docs(readme): update readme for 6.3.1

Signed-off-by: k4yt3x <i@k4yt3x.com>
fix(encoder): decreasing PTS precision with increasing video duration (#1269 )
2026-02-08 13:34:45 +08:00 · 2024-12-21 00:00:00 +00:00 · 2024-12-21 00:00:00 +00:00 · 2024-12-21 00:00:00 +00:00 · 2024-12-21 00:00:00 +00:00 · 2024-12-21 00:00:00 +00:00
91 changed files with 2348 additions and 647 deletions
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -54,7 +54,7 @@ jobs:
            -DCMAKE_BUILD_TYPE=Debug -DCMAKE_INSTALL_PREFIX=/tmp/install \
            -DINSTALL_BIN_DESTINATION=. -DINSTALL_INCLUDE_DESTINATION=include \
            -DINSTALL_LIB_DESTINATION=. -DINSTALL_MODEL_DESTINATION=.
-          cmake --build /tmp/build --config Debug --target install --parallel
+          cmake --build /tmp/build --config Debug --target install

      - name: Upload artifacts
        uses: actions/upload-artifact@v4
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -68,6 +68,8 @@ jobs:
          path: build/video2x-linux-ubuntu-2404-amd64.deb

  windows:
+    # Disabled because it takes too long to build
+    if: false
    name: Build Windows release
    needs:
      - setup
@@ -143,14 +145,14 @@ jobs:
    needs:
      - setup
      # - ubuntu-2404
-      - windows
+      # - windows
      - container
    runs-on: ubuntu-latest
-    outputs:
-      upload_url: ${{ steps.create_release.outputs.upload_url }}
+    # outputs:
+    #   upload_url: ${{ steps.create_release.outputs.upload_url }}
    steps:
-      - name: Download artifacts
-        uses: actions/download-artifact@v4
+      # - name: Download artifacts
+      #   uses: actions/download-artifact@v4

      - name: Create release
        id: create_release
@@ -161,5 +163,3 @@ jobs:
          name: ${{ needs.setup.outputs.version }}
          draft: true
          prerelease: false
-          files: |
-            video2x-windows-amd64.zip
--- a/.gitmodules
+++ b/.gitmodules
@@ -1,6 +1,3 @@
-[submodule "third_party/librealesrgan_ncnn_vulkan"]
-	path = third_party/librealesrgan_ncnn_vulkan
-	url = https://github.com/k4yt3x/librealesrgan-ncnn-vulkan.git
 [submodule "third_party/ncnn"]
 	path = third_party/ncnn
 	url = https://github.com/Tencent/ncnn.git
@@ -10,6 +7,12 @@
 [submodule "third_party/boost"]
 	path = third_party/boost
 	url = https://github.com/boostorg/boost.git
+[submodule "third_party/librealesrgan_ncnn_vulkan"]
+	path = third_party/librealesrgan_ncnn_vulkan
+	url = https://github.com/k4yt3x/librealesrgan-ncnn-vulkan.git
+[submodule "third_party/librealcugan_ncnn_vulkan"]
+	path = third_party/librealcugan_ncnn_vulkan
+	url = https://github.com/k4yt3x/librealcugan-ncnn-vulkan.git
 [submodule "third_party/librife_ncnn_vulkan"]
 	path = third_party/librife_ncnn_vulkan
 	url = https://github.com/k4yt3x/librife-ncnn-vulkan.git
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,39 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

+## [6.3.1] - 2024-12-21
+
+### Fixed
+
+- The issue of decreasing PTS precision with increasing video duration (#1269).
+- (Video2X Qt6) Errors restoring Real-CUGAN configs in the Edit Task dialog.
+- (Video2X Qt6) The incorrect comparison of version numbers.
+
+## [6.3.0] - 2024-12-20
+
+### Added
+
+- A logger manager to provide unified logging across the application.
+- Support for Real-CUGAN ncnn Vulkan (#1198).
+- (Video2X Qt6) A check to ensure the required VC++ Redistributable version is installed.
+- (Video2X Qt6) A configuration manager to save user preferences like the last selected language.
+- (Video2X Qt6) A new logging widget to display logs in the UI.
+- (Video2X Qt6) Custom options `crf=20` and `preset=slow` to the default encoder options.
+- (Video2X Qt6) French translation.
+- (Video2X Qt6) The ability to check for available updates and prompt the user to update.
+- (Video2X Qt6) Tooltips for processor and encoder options.
+
+### Changed
+
+- Improve optimization flags and add namespaces for better code organization.
+- (Video2X Qt6) Add processor names to processed videos instead of `.processed`.
+- (Video2X Qt6) The output video suffix from auto-generated to `.mkv`.
+
+### Fixed
+
+- Make the encoder always use the calculated PTS with corrected math.
+- (Video2X Qt6) The issue where task configs are being restored incorrectly in the UI.
+
 ## [6.2.0] - 2024-12-11

 ### Added
@@ -14,14 +47,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - More `AVCodecContext` options.
 - Support for RIFE ncnn Vulkan.
 - Support for specifying arbitrary `AVOptions` for the encoder (#1232).
- Visual C++ Redistributable version check to the installer. (Video2X Qt6)
+- (Video2X Qt6) Visual C++ Redistributable version check to the installer.

 ### Changed

 - Improve CLI argument validation.
 - Improve error handling and error messages.
 - Improve the CLI help message structure and clarity.
- Improve the UI with a complete redesign. (Video2X Qt6)
+- (Video2X Qt6) Improve the UI with a complete redesign.

 ### Removed

@@ -48,9 +81,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

 - A better timer that gets paused when the processing is paused.
 - Detection for the validity of the provided GPU ID.
- Status bar and processing statistics. (Video2X Qt6)
 - The `--listgpus` option to list available Vulkan GPU devices.
 - Vulkan device selection for libplacebo.
+- (Video2X Qt6) Status bar and processing statistics.

 ### Changed

--- a/CITATION.cff
+++ b/CITATION.cff
@@ -0,0 +1,16 @@
+cff-version: 1.2.0
+title: Video2X
+abstract: >-
+  Video2X is a A machine learning-based video super resolution and frame interpolation framework.
+message: Please cite this software using these metadata.
+type: software
+authors:
+  - name: K4YT3X
+date-released: 2018-02-24
+url: https://video2x.org
+repository-code: https://github.com/k4yt3x/video2x
+license: AGPL-3.0
+keywords:
+  - machine learning
+  - super resolution
+  - frame interpolation
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -1,6 +1,7 @@
 cmake_minimum_required(VERSION 3.10)
-project(video2x VERSION 6.2.0 LANGUAGES CXX)
+project(video2x VERSION 6.3.1 LANGUAGES CXX)

+# The FindBoost module is removed in CMake 3.30
 if(POLICY CMP0167)
  cmake_policy(SET CMP0167 NEW)
 endif()
@@ -15,24 +16,24 @@ if(NOT CMAKE_BUILD_TYPE)
    set(CMAKE_BUILD_TYPE Release)
 endif()

-# Set the default optimization flags for Release builds
-if(CMAKE_BUILD_TYPE STREQUAL "Release")
-    if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
-        set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Ox /GL /LTCG /MD /DNDEBUG")
-    elseif(CMAKE_CXX_COMPILER_ID STREQUAL "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
-        set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -O3 -march=native -flto")
-        set(CMAKE_EXE_LINKER_FLAGS_RELEASE "${CMAKE_EXE_LINKER_FLAGS_RELEASE} -s")
-        set(CMAKE_SHARED_LINKER_FLAGS_RELEASE "${CMAKE_SHARED_LINKER_FLAGS_RELEASE} -s")
-    endif()
-endif()
-
 # Set global compile options for all targets
 if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
    add_compile_options(/W4 /permissive-)
-elseif(CMAKE_CXX_COMPILER_ID STREQUAL "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
+elseif(CMAKE_CXX_COMPILER_ID MATCHES "GNU|Clang")
    add_compile_options(-Wall -Wextra -Wpedantic -Wconversion -Wshadow)
 endif()

+# Set the default optimization flags for Release builds
+if(CMAKE_BUILD_TYPE STREQUAL "Release")
+    if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
+        add_compile_options(/Ox /Ot /GL /DNDEBUG)
+        add_link_options(/LTCG /OPT:REF /OPT:ICF)
+    elseif(CMAKE_CXX_COMPILER_ID MATCHES "GNU|Clang")
+        add_compile_options(-O3 -march=native -ffunction-sections -fdata-sections)
+        add_link_options(-Wl,-s -flto -Wl,--gc-sections)
+    endif()
+endif()
+
 # Build options
 option(BUILD_SHARED_LIBS "Build libvideo2x as a shared library" ON)
 option(BUILD_VIDEO2X_CLI "Build the video2x executable" ON)
@@ -48,8 +49,10 @@ configure_file(
 )

 # Find the required packages
-set(ALL_INCLUDE_DIRS)
-set(ALL_LIBRARIES)
+set(LIBVIDEO2X_INCLUDE_DIRS)
+set(LIBVIDEO2X_LIBS)
+set(VIDEO2X_INCLUDE_DIRS)
+set(VIDEO2X_LIBS)

 # Platform-specific dependencies
 if(WIN32)
@@ -66,11 +69,12 @@ if(WIN32)
        ${FFMPEG_BASE_PATH}/lib/avutil.lib
        ${FFMPEG_BASE_PATH}/lib/swscale.lib
    )
-    list(APPEND ALL_LIBRARIES ${FFMPEG_LIB})
-    list(APPEND ALL_INCLUDE_DIRS ${FFMPEG_BASE_PATH}/include)
+    list(APPEND LIBVIDEO2X_LIBS ${FFMPEG_LIB})
+    list(APPEND LIBVIDEO2X_INCLUDE_DIRS ${FFMPEG_BASE_PATH}/include)
+    list(APPEND VIDEO2X_LIBS ${FFMPEG_LIB})
+    list(APPEND VIDEO2X_INCLUDE_DIRS ${FFMPEG_BASE_PATH}/include)

    # ncnn
-    # TODO: Figure out why this file is not being copied to the install directory
    set(SPIRV_BUILD_PATH
        ${CMAKE_BINARY_DIR}/realesrgan-prefix/src/realesrgan-build/ncnn/glslang/SPIRV
    )
@@ -80,11 +84,8 @@ if(WIN32)
        set(SPIRV_LIB ${SPIRV_BUILD_PATH}/Debug/SPIRVd.lib)
    endif()

-    list(APPEND ALL_LIBRARIES
-        ${NCNN_BASE_PATH}/lib/ncnn.lib
-        ${SPIRV_LIB}
-    )
-    list(APPEND ALL_INCLUDE_DIRS ${NCNN_BASE_PATH}/include/ncnn)
+    list(APPEND LIBVIDEO2X_LIBS ${NCNN_BASE_PATH}/lib/ncnn.lib ${SPIRV_LIB})
+    list(APPEND LIBVIDEO2X_INCLUDE_DIRS ${NCNN_BASE_PATH}/include/ncnn)
 else()
    # FFmpeg
    find_package(PkgConfig REQUIRED)
@@ -98,13 +99,13 @@ else()
    )

    # Loop through each package to find and collect include dirs and libraries
-    set(FFMPEG_LIB)
    foreach(PKG ${FFMPEG_REQUIRED_PKGS})
        pkg_check_modules(${PKG} REQUIRED ${PKG})
-        list(APPEND ALL_INCLUDE_DIRS ${${PKG}_INCLUDE_DIRS})
-        list(APPEND FFMPEG_LIB ${${PKG}_LIBRARIES})
+        list(APPEND LIBVIDEO2X_INCLUDE_DIRS ${${PKG}_INCLUDE_DIRS})
+        list(APPEND LIBVIDEO2X_LIBS ${${PKG}_LIBRARIES})
+        list(APPEND VIDEO2X_INCLUDE_DIRS ${${PKG}_INCLUDE_DIRS})
+        list(APPEND VIDEO2X_LIBS ${${PKG}_LIBRARIES})
    endforeach()
-    list(APPEND ALL_LIBRARIES ${FFMPEG_LIB})
 endif() # WIN32

 # Find ncnn package
@@ -209,32 +210,38 @@ endif()
 # spdlog
 if(USE_SYSTEM_SPDLOG)
    find_package(spdlog REQUIRED)
-    list(APPEND ALL_INCLUDE_DIRS ${spdlog_INCLUDE_DIRS})
+    list(APPEND LIBVIDEO2X_INCLUDE_DIRS ${spdlog_INCLUDE_DIRS})
+    list(APPEND VIDEO2X_INCLUDE_DIRS ${spdlog_INCLUDE_DIRS})
    set(SPDLOG_LIB spdlog::spdlog)
 else()
    add_subdirectory(third_party/spdlog)
    set(SPDLOG_LIB spdlog::spdlog_header_only)
 endif()
-list(APPEND ALL_LIBRARIES ${SPDLOG_LIB})
-
-# Boost
-if(USE_SYSTEM_BOOST)
-    find_package(Boost REQUIRED COMPONENTS program_options)
-    list(APPEND ALL_INCLUDE_DIRS ${Boost_INCLUDE_DIRS})
-else()
-    option(Boost_USE_STATIC_LIBS "" ON)
-    option(Boost_USE_STATIC_RUNTIME "" ON)
-    option(Boost_COMPONENTS "program_options")
-
-    add_subdirectory(third_party/boost)
-    include_directories(${PROJECT_SOURCE_DIR}/third_party/boost/libs/program_options/include)
-    set(BOOST_BASE_PATH ${CMAKE_BINARY_DIR}/third_party/boost/libs/program_options/${CMAKE_BUILD_TYPE})
-endif()
-set(BOOST_LIB Boost::program_options)
+list(APPEND LIBVIDEO2X_LIBS ${SPDLOG_LIB})
+list(APPEND VIDEO2X_LIBS ${SPDLOG_LIB})

+# Find dependencies required for the CLI
 if(BUILD_VIDEO2X_CLI)
+    # Vulkan
    find_package(Vulkan REQUIRED)
-    set(VULKAN_LIB Vulkan::Vulkan)
+    list(APPEND VIDEO2X_LIBS Vulkan::Vulkan)
+
+    # Boost
+    if(USE_SYSTEM_BOOST)
+        find_package(Boost REQUIRED COMPONENTS program_options)
+        list(APPEND LIBVIDEO2X_INCLUDE_DIRS ${Boost_INCLUDE_DIRS})
+    else()
+        option(Boost_USE_STATIC_LIBS "" ON)
+        option(Boost_USE_STATIC_RUNTIME "" ON)
+        option(Boost_COMPONENTS "program_options")
+
+        add_subdirectory(third_party/boost)
+        include_directories(${PROJECT_SOURCE_DIR}/third_party/boost/libs/program_options/include)
+        set(BOOST_BASE_PATH
+            ${CMAKE_BINARY_DIR}/third_party/boost/libs/program_options/${CMAKE_BUILD_TYPE}
+        )
+    endif()
+    list(APPEND VIDEO2X_LIBS Boost::program_options)
 endif()

 # Include ExternalProject module
@@ -253,7 +260,20 @@ ExternalProject_Add(
    INSTALL_COMMAND ${CMAKE_COMMAND} --build . --target install --config ${CMAKE_BUILD_TYPE}
 )

-# Add librealesrgan-ncnn-vulkan as an external project
+# Add librealcugan-ncnn-vulkan as an external project
+ExternalProject_Add(
+    realcugan
+    SOURCE_DIR ${PROJECT_SOURCE_DIR}/third_party/librealcugan_ncnn_vulkan/src
+    CMAKE_ARGS
+        -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
+        -DCMAKE_INSTALL_PREFIX=${CMAKE_BINARY_DIR}/realcugan_install
+        -DCMAKE_POSITION_INDEPENDENT_CODE=ON
+        -DUSE_SYSTEM_NCNN=${USE_SYSTEM_NCNN}
+    BUILD_ALWAYS ON
+    INSTALL_COMMAND ${CMAKE_COMMAND} --build . --target install --config ${CMAKE_BUILD_TYPE}
+)
+
+# Add librife-ncnn-vulkan as an external project
 ExternalProject_Add(
    rife
    SOURCE_DIR ${PROJECT_SOURCE_DIR}/third_party/librife_ncnn_vulkan/src
@@ -267,8 +287,10 @@ ExternalProject_Add(
 )

 # Remove duplicate entries
-list(REMOVE_DUPLICATES ALL_INCLUDE_DIRS)
-list(REMOVE_DUPLICATES ALL_LIBRARIES)
+list(REMOVE_DUPLICATES LIBVIDEO2X_INCLUDE_DIRS)
+list(REMOVE_DUPLICATES LIBVIDEO2X_LIBS)
+list(REMOVE_DUPLICATES VIDEO2X_INCLUDE_DIRS)
+list(REMOVE_DUPLICATES VIDEO2X_LIBS)

 # Create the shared library 'libvideo2x'
 file(GLOB LIBVIDEO2X_SOURCES src/*.cpp)
@@ -281,37 +303,36 @@ else()
 endif()

 # Ensure that the shared library is built after the external projects
-add_dependencies(libvideo2x realesrgan rife)
+add_dependencies(libvideo2x realesrgan realcugan rife)

 # Include directories for the shared library
 target_include_directories(libvideo2x PRIVATE
-    ${ALL_INCLUDE_DIRS}
+    ${LIBVIDEO2X_INCLUDE_DIRS}
    ${CMAKE_CURRENT_BINARY_DIR}
    ${PROJECT_SOURCE_DIR}/include
    ${PROJECT_SOURCE_DIR}/include/libvideo2x
    ${PROJECT_SOURCE_DIR}/third_party/librealesrgan_ncnn_vulkan/src
+    ${PROJECT_SOURCE_DIR}/third_party/librealcugan_ncnn_vulkan/src
    ${PROJECT_SOURCE_DIR}/third_party/librife_ncnn_vulkan/src
 )

 # Compile options for the shared library
-target_compile_options(libvideo2x PRIVATE
-    -fPIC
-    $<$<CONFIG:Release>:-Ofast>
-    $<$<CONFIG:Debug>:-g -DDEBUG>
-)
+target_compile_options(libvideo2x PRIVATE -fPIC $<$<CONFIG:Debug>:-g -DDEBUG>)

 # Define the paths to the shared libraries
 if(WIN32)
    set(REALESRGAN_LIB ${CMAKE_BINARY_DIR}/realesrgan_install/lib/librealesrgan-ncnn-vulkan.lib)
+    set(REALCUGAN_LIB ${CMAKE_BINARY_DIR}/realcugan_install/lib/librealcugan-ncnn-vulkan.lib)
    set(RIFE_LIB ${CMAKE_BINARY_DIR}/rife_install/lib/librife-ncnn-vulkan.lib)
 else()
    set(REALESRGAN_LIB ${CMAKE_BINARY_DIR}/realesrgan_install/lib/librealesrgan-ncnn-vulkan.so)
+    set(REALCUGAN_LIB ${CMAKE_BINARY_DIR}/realcugan_install/lib/librealcugan-ncnn-vulkan.so)
    set(RIFE_LIB ${CMAKE_BINARY_DIR}/rife_install/lib/librife-ncnn-vulkan.so)
 endif()
-list(APPEND ALL_LIBRARIES ${REALESRGAN_LIB} ${RIFE_LIB})
+list(APPEND LIBVIDEO2X_LIBS ${REALESRGAN_LIB} ${REALCUGAN_LIB} ${RIFE_LIB})

 # Link the shared library with the dependencies
-target_link_libraries(libvideo2x PRIVATE ${ALL_LIBRARIES})
+target_link_libraries(libvideo2x PRIVATE ${LIBVIDEO2X_LIBS})

 if(NOT WIN32)
    if(USE_SYSTEM_NCNN)
@@ -329,7 +350,7 @@ if(BUILD_VIDEO2X_CLI)

    # Include directories for the executable
    target_include_directories(video2x PRIVATE
-        ${ALL_INCLUDE_DIRS}
+        ${VIDEO2X_INCLUDE_DIRS}
        ${CMAKE_CURRENT_BINARY_DIR}
        ${PROJECT_SOURCE_DIR}/include
        ${PROJECT_SOURCE_DIR}/tools/video2x/include
@@ -339,13 +360,7 @@ if(BUILD_VIDEO2X_CLI)
    target_compile_options(video2x PRIVATE $<$<CONFIG:Debug>:-g -DDEBUG>)

    # Link the executable with the shared library
-    target_link_libraries(video2x PRIVATE
-        libvideo2x
-        ${FFMPEG_LIB}
-        ${SPDLOG_LIB}
-        ${BOOST_LIB}
-        ${VULKAN_LIB}
-    )
+    target_link_libraries(video2x PRIVATE libvideo2x ${VIDEO2X_LIBS})
 endif()

 # Define the default installation directories
@@ -404,6 +419,7 @@ if(WIN32)
    file(GLOB FFMPEG_DLLS "${FFMPEG_BASE_PATH}/bin/*.dll")
    install(FILES
        ${CMAKE_BINARY_DIR}/realesrgan_install/bin/librealesrgan-ncnn-vulkan.dll
+        ${CMAKE_BINARY_DIR}/realcugan_install/bin/librealcugan-ncnn-vulkan.dll
        ${CMAKE_BINARY_DIR}/rife_install/bin/librife-ncnn-vulkan.dll
        ${FFMPEG_DLLS}
        ${NCNN_BASE_PATH}/bin/ncnn.dll
@@ -414,7 +430,7 @@ if(WIN32)
                    WORLD_READ WORLD_EXECUTE
    )
 else()
-    install(FILES ${REALESRGAN_LIB} ${RIFE_LIB}
+    install(FILES ${REALESRGAN_LIB} ${REALCUGAN_LIB} ${RIFE_LIB}
        DESTINATION ${INSTALL_LIB_DESTINATION}
        PERMISSIONS OWNER_READ OWNER_WRITE OWNER_EXECUTE
                    GROUP_READ GROUP_EXECUTE
--- a/50
+++ b/50
@@ -1,10 +1,9 @@
 .PHONY: build static debug windows windows-debug debian ubuntu clean \
-	test-realesrgan test-libplacebo test-rife \
-	memcheck-realesrgan memcheck-libplacebo memcheck-rife \
-	heaptrack-realesrgan heaptrack-libplacebo heaptrack-rife
+	test-realesrgan test-realcugan test-libplacebo test-rife \
+	memcheck-realesrgan memcheck-realcugan memcheck-libplacebo memcheck-rife \
+	heaptrack-realesrgan heaptrack-realcugan heaptrack-libplacebo heaptrack-rife

 BINDIR=build
-CC=clang
 CXX=clang++

 TEST_VIDEO=data/standard-test.mp4
@@ -13,7 +12,6 @@ TEST_OUTPUT=data/output.mp4
 build:
 	cmake -S . -B $(BINDIR) \
 		-DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
-		-DCMAKE_C_COMPILER=$(CC) \
 		-DCMAKE_CXX_COMPILER=$(CXX) \
 		-DCMAKE_BUILD_TYPE=Release
 	cmake --build $(BINDIR) --config Release --parallel
@@ -22,17 +20,18 @@ build:
 static:
 	cmake -S . -B $(BINDIR) \
 		-DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
-		-DCMAKE_C_COMPILER=$(CC) \
 		-DCMAKE_CXX_COMPILER=$(CXX) \
 		-DCMAKE_BUILD_TYPE=Release \
-		-DBUILD_SHARED_LIBS=OFF
+		-DBUILD_SHARED_LIBS=OFF \
+		-DUSE_SYSTEM_NCNN=OFF \
+		-DUSE_SYSTEM_SPDLOG=OFF \
+		-DUSE_SYSTEM_BOOST=OFF
 	cmake --build $(BINDIR) --config Release --parallel
 	cp $(BINDIR)/compile_commands.json .

 debug:
 	cmake -S . -B $(BINDIR) \
 		-DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
-		-DCMAKE_C_COMPILER=$(CC) \
 		-DCMAKE_CXX_COMPILER=$(CXX) \
 		-DCMAKE_BUILD_TYPE=Debug
 	cmake --build $(BINDIR) --config Debug --parallel
@@ -71,8 +70,7 @@ debian:
 		libomp-dev \
 		libspdlog-dev \
 		libboost-program-options-dev
-	cmake -B /tmp/build -S . -DUSE_SYSTEM_NCNN=OFF \
-		-DCMAKE_C_COMPILER=$(CC) -DCMAKE_CXX_COMPILER=$(CXX) \
+	cmake -B /tmp/build -S . -DUSE_SYSTEM_NCNN=OFF -DCMAKE_CXX_COMPILER=$(CXX) \
 		-DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/tmp/install \
 		-DINSTALL_BIN_DESTINATION=. -DINSTALL_INCLUDE_DESTINATION=include \
 		-DINSTALL_LIB_DESTINATION=. -DINSTALL_MODEL_DESTINATION=.
@@ -93,8 +91,8 @@ ubuntu2404:
 		libomp-dev \
 		libboost-program-options-dev
 	cmake -B build -S . -DUSE_SYSTEM_NCNN=OFF -DUSE_SYSTEM_SPDLOG=OFF -DSPDLOG_NO_EXCEPTIONS=ON \
-		-DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++ \
-		-DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=video2x-linux-ubuntu-amd64/usr
+		-DCMAKE_CXX_COMPILER=g++ -DCMAKE_BUILD_TYPE=Release \
+		-DCMAKE_INSTALL_PREFIX=video2x-linux-ubuntu-amd64/usr
 	cmake --build build --config Release --target install --parallel
 	mkdir -p video2x-linux-ubuntu-amd64/DEBIAN
 	cp packaging/debian/control.ubuntu2404 video2x-linux-ubuntu-amd64/DEBIAN/control
@@ -118,8 +116,8 @@ ubuntu2204:
 		libomp-dev \
 		libboost-program-options-dev
 	cmake -B build -S . -DUSE_SYSTEM_NCNN=OFF -DUSE_SYSTEM_SPDLOG=OFF -DSPDLOG_NO_EXCEPTIONS=ON \
-		-DCMAKE_C_COMPILER=gcc -DCMAKE_CXX_COMPILER=g++ \
-		-DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=video2x-linux-ubuntu-amd64/usr
+		-DCMAKE_CXX_COMPILER=g++ -DCMAKE_BUILD_TYPE=Release \
+		-DCMAKE_INSTALL_PREFIX=video2x-linux-ubuntu-amd64/usr
 	cmake --build build --config Release --target install --parallel
 	mkdir -p video2x-linux-ubuntu-amd64/DEBIAN
 	cp packaging/debian/control.ubuntu2204 video2x-linux-ubuntu-amd64/DEBIAN/control
@@ -132,6 +130,10 @@ test-realesrgan:
 	LD_LIBRARY_PATH=$(BINDIR) $(BINDIR)/video2x -i $(TEST_VIDEO) -o $(TEST_OUTPUT) \
 		-p realesrgan -s 4 --realesrgan-model realesr-animevideov3

+test-realcugan:
+	LD_LIBRARY_PATH=$(BINDIR) $(BINDIR)/video2x -i $(TEST_VIDEO) -o $(TEST_OUTPUT) \
+		-p realcugan -s 4 -n 0 --realcugan-model models-se
+
 test-libplacebo:
 	LD_LIBRARY_PATH=$(BINDIR) $(BINDIR)/video2x -i $(TEST_VIDEO) -o $(TEST_OUTPUT) \
 		-p libplacebo -w 1920 -h 1080 --libplacebo-shader anime4k-v4-a
@@ -153,6 +155,19 @@ memcheck-realesrgan:
 		-p realesrgan -s 2 --realesrgan-model realesr-animevideov3 \
 		-e preset=veryfast -e crf=30

+memcheck-realcugan:
+	LD_LIBRARY_PATH=$(BINDIR) valgrind \
+		--tool=memcheck \
+		--leak-check=full \
+		--show-leak-kinds=all \
+		--track-origins=yes \
+		--show-reachable=yes \
+		--verbose --log-file="valgrind.log" \
+		$(BINDIR)/video2x \
+		-i $(TEST_VIDEO) -o $(TEST_OUTPUT) \
+		-p realcugan -s 2 -n 0 --realcugan-model models-se \
+		-e preset=veryfast -e crf=30
+
 memcheck-libplacebo:
 	LD_LIBRARY_PATH=$(BINDIR) valgrind \
 		--tool=memcheck \
@@ -186,6 +201,13 @@ heaptrack-realesrgan:
 		-p realesrgan -s 4 --realesrgan-model realesr-animevideov3 \
 		-e preset=veryfast -e crf=30

+heaptrack-realcugan:
+	LD_LIBRARY_PATH=$(BINDIR) HEAPTRACK_ENABLE_DEBUGINFOD=1 heaptrack \
+		$(BINDIR)/video2x \
+		-i $(TEST_VIDEO) -o $(TEST_OUTPUT) \
+		-p realcugan -s 4 -n 0 --realcugan-model models-se \
+		-e preset=veryfast -e crf=30
+
 heaptrack-libplacebo:
 	LD_LIBRARY_PATH=$(BINDIR) HEAPTRACK_ENABLE_DEBUGINFOD=1 heaptrack \
 		$(BINDIR)/video2x \
--- a/9
+++ b/9
@@ -4,8 +4,17 @@ Copyright (C) 2018-2024 K4YT3X and contributors.
 This product depends on FFmpeg, which is available under the GNU Lesser General Public License 2.1.
 The source code can be found at https://github.com/FFmpeg/FFmpeg.

+This product depends on Anime4K, which is available under the MIT License.
+The source code can be found at https://github.com/bloc97/Anime4K.
+
 This product depends on Real-ESRGAN ncnn Vulkan, which is available under the MIT License.
 The source code can be found at https://github.com/xinntao/Real-ESRGAN-ncnn-vulkan.

+This product depends on Real-CUGAN ncnn Vulkan, which is available under the MIT License.
+The source code can be found at https://github.com/nihui/realcugan-ncnn-vulkan.
+
+This product depends on RIFE ncnn Vulkan, which is available under the MIT License.
+The source code can be found at https://github.com/nihui/rife-ncnn-vulkan.
+
 This product depends on ncnn, which is available under the BSD 3-Clause License.
 The source code can be found at https://github.com/Tencent/ncnn.
--- a/README.md
+++ b/README.md
@@ -27,23 +27,19 @@ Version 6.0.0 is a complete rewrite of this project in C/C++. It:
 - genuinely works this time, with much less hassle compared to the 5.0.0 beta;
 - is blazing fast, thanks to the new optimized pipeline and the efficiency of C/C++;
 - is cross-platform, available now for both Windows and Linux;
- offers significantly better output quality with Anime4K v4, RealESRGAN, and RIFE;
+- offers significantly better output quality with Anime4K v4, RealESRGAN, RealCUGAN, and RIFE;
 - supports two modes: filtering (upscaling) and frame interpolation;
 - supports Anime4K v4 and all custom MPV-compatible GLSL shaders;
- supports RealESRGAN (all three models) via ncnn and Vulkan;
- supports RIFE via ncnn and Vulkan;
- requires zero additional disk space during processing, just space for the final output; and
- exports a standard C function for easy integration into other projects! (documentation is on the way)
-
-Support for RealCUGAN is coming soon.
+- supports RealESRGAN, RealCUGAN, and RIFE (all models) via ncnn and Vulkan;
+- requires zero additional disk space during processing, just space for the final output.

 </details>

-![6.1.0-screenshot](https://github.com/user-attachments/assets/57aa11d0-dd01-49e9-b6b0-2d2f21a363ac)
+![6.3.0-screenshot](https://github.com/user-attachments/assets/c5442f84-5ffc-4476-915f-a0fc188a2cb3)

 ## [🪟 Install on Windows](https://docs.video2x.org/installing/windows-qt6.html)

-**[Download the Latest Windows Installer Executable (6.1.1)](https://github.com/k4yt3x/video2x/releases/download/6.1.1/video2x-qt6-windows-amd64-installer.exe)**
+**[Download the Latest Windows Installer Executable (6.3.1)](https://github.com/k4yt3x/video2x/releases/download/6.3.1/video2x-qt6-windows-amd64-installer.exe)**

 You can download the latest Windows release on the [releases page](https://github.com/k4yt3x/video2x/releases/latest). For basic GUI usage, refer to the [documentation](https://docs.video2x.org/running/desktop.html). If you're unable to download directly from GitHub, try the [mirror site](https://files.k4yt3x.com/Projects/Video2X). The GUI currently supports the following languages:

@@ -51,6 +47,7 @@ You can download the latest Windows release on the [releases page](https://githu
 - 简体中文（中国）
 - 日本語（日本）
 - Português (Portugal)
+- Français (France)

 ## [🐧 Install on Linux](https://docs.video2x.org/installing/linux.html)

@@ -92,6 +89,8 @@ _Upscale demo: Spirited Away's movie trailer_
 The following clip can be used to test if your setup works properly. This is also the standard clip used for running performance benchmarks.

 - [Standard Test Clip (240P)](https://files.k4yt3x.com/Resources/Videos/standard-test.mp4) 4.54 MiB
+- [Real-CUGAN Upscaled Sample (1704P)](https://files.k4yt3x.com/Resources/Videos/standard-realcugan.mp4) 3.5 MiB
+- [Real-ESRGAN Upscaled Sample (1704P)](https://files.k4yt3x.com/Resources/Videos/standard-realesrgan.mp4) 3.1 MiB
 - [waifu2x Upscaled Sample (1080P)](https://files.k4yt3x.com/Resources/Videos/standard-waifu2x.mp4) 4.54 MiB
 - [Ground Truth (1080P)](https://files.k4yt3x.com/Resources/Videos/standard-original.mp4) 22.2 MiB

@@ -109,10 +108,12 @@ This project includes or depends on these following projects:

 | Project                                                                               | License         |
 | ------------------------------------------------------------------------------------- | --------------- |
-| [bloc97/Anime4K](https://github.com/bloc97/Anime4K)                                   | MIT License     |
 | [FFmpeg/FFmpeg](https://www.ffmpeg.org/)                                              | LGPLv2.1, GPLv2 |
-| [xinntao/Real-ESRGAN-ncnn-vulkan](https://github.com/xinntao/Real-ESRGAN-ncnn-vulkan) | MIT License     |
 | [Tencent/ncnn](https://github.com/Tencent/ncnn)                                       | BSD 3-Clause    |
+| [bloc97/Anime4K](https://github.com/bloc97/Anime4K)                                   | MIT License     |
+| [nihui/realcugan-ncnn-vulkan](https://github.com/nihui/realcugan-ncnn-vulkan)         | MIT License     |
+| [nihui/rife-ncnn-vulkan](https://github.com/nihui/rife-ncnn-vulkan)                   | MIT License     |
+| [xinntao/Real-ESRGAN-ncnn-vulkan](https://github.com/xinntao/Real-ESRGAN-ncnn-vulkan) | MIT License     |

 More licensing information can be found in the [NOTICE](NOTICE) file.

--- a/include/libvideo2x/avutils.h
+++ b/include/libvideo2x/avutils.h
@@ -4,6 +4,9 @@ extern "C" {
 #include <libavformat/avformat.h>
 }

+namespace video2x {
+namespace avutils {
+
 AVRational get_video_frame_rate(AVFormatContext *ifmt_ctx, int in_vstream_idx);

 int64_t get_video_frame_count(AVFormatContext *ifmt_ctx, int in_vstream_idx);
@@ -17,3 +20,6 @@ void av_bufferref_deleter(AVBufferRef *bufferref);
 void av_frame_deleter(AVFrame *frame);

 void av_packet_deleter(AVPacket *packet);
+
+}  // namespace avutils
+}  // namespace video2x
--- a/include/libvideo2x/conversions.h
+++ b/include/libvideo2x/conversions.h
@@ -7,6 +7,9 @@ extern "C" {

 #include <mat.h>

+namespace video2x {
+namespace conversions {
+
 // Convert AVFrame to another pixel format
 AVFrame *convert_avframe_pix_fmt(AVFrame *src_frame, AVPixelFormat pix_fmt);

@@ -15,3 +18,6 @@ ncnn::Mat avframe_to_ncnn_mat(AVFrame *frame);

 // Convert ncnn::Mat to AVFrame
 AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt);
+
+}  // namespace conversions
+}  // namespace video2x
--- a/include/libvideo2x/decoder.h
+++ b/include/libvideo2x/decoder.h
@@ -7,6 +7,9 @@ extern "C" {
 #include <libavformat/avformat.h>
 }

+namespace video2x {
+namespace decoder {
+
 class Decoder {
   public:
    Decoder();
@@ -26,3 +29,6 @@ class Decoder {
    AVCodecContext *dec_ctx_;
    int in_vstream_idx_;
 };
+
+}  // namespace decoder
+}  // namespace video2x
--- a/include/libvideo2x/encoder.h
+++ b/include/libvideo2x/encoder.h
@@ -12,6 +12,9 @@ extern "C" {

 #include "fsutils.h"

+namespace video2x {
+namespace encoder {
+
 // Encoder configurations
 struct EncoderConfig {
    // Non-AVCodecContext options
@@ -42,7 +45,7 @@ struct EncoderConfig {
    int delay = -1;

    // Extra AVOptions
-    std::vector<std::pair<StringType, StringType>> extra_opts;
+    std::vector<std::pair<fsutils::StringType, fsutils::StringType>> extra_opts;
 };

 class Encoder {
@@ -76,3 +79,6 @@ class Encoder {
    int out_vstream_idx_;
    int *stream_map_;
 };
+
+}  // namespace encoder
+}  // namespace video2x
--- a/include/libvideo2x/filter_libplacebo.h
+++ b/include/libvideo2x/filter_libplacebo.h
@@ -10,6 +10,9 @@ extern "C" {

 #include "processor.h"

+namespace video2x {
+namespace processors {
+
 // FilterLibplacebo class definition
 class FilterLibplacebo : public Filter {
   public:
@@ -56,3 +59,6 @@ class FilterLibplacebo : public Filter {
    AVRational in_time_base_;
    AVRational out_time_base_;
 };
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/filter_realcugan.h
+++ b/include/libvideo2x/filter_realcugan.h
@@ -0,0 +1,63 @@
+#pragma once
+
+extern "C" {
+#include <libavcodec/avcodec.h>
+}
+
+#include "processor.h"
+#include "realcugan.h"
+
+namespace video2x {
+namespace processors {
+
+// FilterRealcugan class definition
+class FilterRealcugan : public Filter {
+   public:
+    // Constructor
+    FilterRealcugan(
+        int gpuid = 0,
+        bool tta_mode = false,
+        int scaling_factor = 4,
+        int noise_level = -1,
+        int num_threads = 1,
+        int syncgap = 3,
+        const fsutils::StringType model_name = STR("models-pro")
+    );
+
+    // Destructor
+    virtual ~FilterRealcugan() override;
+
+    // Initializes the filter with decoder and encoder contexts
+    int init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *hw_ctx) override;
+
+    // Processes an input frame and returns the processed frame
+    int filter(AVFrame *in_frame, AVFrame **out_frame) override;
+
+    // Returns the filter's type
+    ProcessorType get_processor_type() const override { return ProcessorType::RealCUGAN; }
+
+    // Returns the filter's output dimensions
+    void get_output_dimensions(
+        const ProcessorConfig &proc_cfg,
+        int in_width,
+        int in_height,
+        int &out_width,
+        int &out_height
+    ) const override;
+
+   private:
+    RealCUGAN *realcugan_;
+    int gpuid_;
+    bool tta_mode_;
+    int scaling_factor_;
+    int noise_level_;
+    int num_threads_;
+    int syncgap_;
+    const fsutils::StringType model_name_;
+    AVRational in_time_base_;
+    AVRational out_time_base_;
+    AVPixelFormat out_pix_fmt_;
+};
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/filter_realesrgan.h
+++ b/include/libvideo2x/filter_realesrgan.h
@@ -7,6 +7,9 @@ extern "C" {
 #include "processor.h"
 #include "realesrgan.h"

+namespace video2x {
+namespace processors {
+
 // FilterRealesrgan class definition
 class FilterRealesrgan : public Filter {
   public:
@@ -15,7 +18,7 @@ class FilterRealesrgan : public Filter {
        int gpuid = 0,
        bool tta_mode = false,
        int scaling_factor = 4,
-        const StringType model_name = STR("realesr-animevideov3")
+        const fsutils::StringType model_name = STR("realesr-animevideov3")
    );

    // Destructor
@@ -44,8 +47,11 @@ class FilterRealesrgan : public Filter {
    int gpuid_;
    bool tta_mode_;
    int scaling_factor_;
-    const StringType model_name_;
+    const fsutils::StringType model_name_;
    AVRational in_time_base_;
    AVRational out_time_base_;
    AVPixelFormat out_pix_fmt_;
 };
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/fsutils.h
+++ b/include/libvideo2x/fsutils.h
@@ -3,6 +3,9 @@
 #include <filesystem>
 #include <string>

+namespace video2x {
+namespace fsutils {
+
 #ifdef _WIN32
 typedef wchar_t CharType;
 #define STR(x) L##x
@@ -23,8 +26,11 @@ std::filesystem::path find_resource_file(const std::filesystem::path &path);

 std::string path_to_u8string(const std::filesystem::path &path);

-std::string wstring_to_u8string(const StringType &wstr);
+std::string wstring_to_u8string(const fsutils::StringType &wstr);

-StringType path_to_string_type(const std::filesystem::path &path);
+fsutils::StringType path_to_string_type(const std::filesystem::path &path);

-StringType to_string_type(int value);
+fsutils::StringType to_string_type(int value);
+
+}  // namespace fsutils
+}  // namespace video2x
--- a/include/libvideo2x/interpolator_rife.h
+++ b/include/libvideo2x/interpolator_rife.h
@@ -7,6 +7,9 @@ extern "C" {
 #include "processor.h"
 #include "rife.h"

+namespace video2x {
+namespace processors {
+
 // InterpolatorRIFE class definition
 class InterpolatorRIFE : public Interpolator {
   public:
@@ -17,7 +20,7 @@ class InterpolatorRIFE : public Interpolator {
        bool tta_temporal_mode = false,
        bool uhd_mode = false,
        int num_threads = 1,
-        const StringType model_name = STR("rife-v4.6")
+        const fsutils::StringType model_name = STR("rife-v4.6")
    );

    // Destructor
@@ -49,8 +52,11 @@ class InterpolatorRIFE : public Interpolator {
    bool tta_temporal_mode_;
    bool uhd_mode_;
    int num_threads_;
-    const StringType model_name_;
+    const fsutils::StringType model_name_;
    AVRational in_time_base_;
    AVRational out_time_base_;
    AVPixelFormat out_pix_fmt_;
 };
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/libplacebo.h
+++ b/include/libvideo2x/libplacebo.h
@@ -7,6 +7,9 @@ extern "C" {
 #include <libavfilter/avfilter.h>
 }

+namespace video2x {
+namespace processors {
+
 int init_libplacebo(
    AVFilterGraph **filter_graph,
    AVFilterContext **buffersrc_ctx,
@@ -17,3 +20,6 @@ int init_libplacebo(
    uint32_t vk_device_index,
    const std::filesystem::path &shader_path
 );
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/libvideo2x.h
+++ b/include/libvideo2x/libvideo2x.h
@@ -12,18 +12,10 @@ extern "C" {
 #include "avutils.h"
 #include "decoder.h"
 #include "encoder.h"
-#include "logging.h"
+#include "libvideo2x_export.h"
 #include "processor.h"

-#ifdef _WIN32
-#ifdef LIBVIDEO2X_EXPORTS
-#define LIBVIDEO2X_API __declspec(dllexport)
-#else
-#define LIBVIDEO2X_API __declspec(dllimport)
-#endif
-#else
-#define LIBVIDEO2X_API
-#endif
+namespace video2x {

 enum class VideoProcessorState {
    Idle,
@@ -37,11 +29,10 @@ enum class VideoProcessorState {
 class LIBVIDEO2X_API VideoProcessor {
   public:
    VideoProcessor(
-        const ProcessorConfig proc_cfg,
-        const EncoderConfig enc_cfg,
-        const uint32_t vk_device_index = 0,
+        const processors::ProcessorConfig proc_cfg,
+        const encoder::EncoderConfig enc_cfg,
+        const uint32_t vk_device_idx = 0,
        const AVHWDeviceType hw_device_type = AV_HWDEVICE_TYPE_NONE,
-        const Video2xLogLevel = Video2xLogLevel::Info,
        const bool benchmark = false
    );

@@ -55,14 +46,17 @@ class LIBVIDEO2X_API VideoProcessor {
    void abort() { state_.store(VideoProcessorState::Aborted); }

    VideoProcessorState get_state() const { return state_.load(); }
-    int64_t get_processed_frames() const { return frame_index_.load(); }
+    int64_t get_processed_frames() const { return frame_idx_.load(); }
    int64_t get_total_frames() const { return total_frames_.load(); }

   private:
-    [[nodiscard]] int
-    process_frames(Decoder &decoder, Encoder &encoder, std::unique_ptr<Processor> &processor);
+    [[nodiscard]] int process_frames(
+        decoder::Decoder &decoder,
+        encoder::Encoder &encoder,
+        std::unique_ptr<processors::Processor> &processor
+    );

-    [[nodiscard]] int write_frame(AVFrame *frame, Encoder &encoder);
+    [[nodiscard]] int write_frame(AVFrame *frame, encoder::Encoder &encoder);

    [[nodiscard]] inline int write_raw_packet(
        AVPacket *packet,
@@ -72,27 +66,29 @@ class LIBVIDEO2X_API VideoProcessor {
    );

    [[nodiscard]] inline int process_filtering(
-        std::unique_ptr<Processor> &processor,
-        Encoder &encoder,
+        std::unique_ptr<processors::Processor> &processor,
+        encoder::Encoder &encoder,
        AVFrame *frame,
        AVFrame *proc_frame
    );

    [[nodiscard]] inline int process_interpolation(
-        std::unique_ptr<Processor> &processor,
-        Encoder &encoder,
-        std::unique_ptr<AVFrame, decltype(&av_frame_deleter)> &prev_frame,
+        std::unique_ptr<processors::Processor> &processor,
+        encoder::Encoder &encoder,
+        std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)> &prev_frame,
        AVFrame *frame,
        AVFrame *proc_frame
    );

-    ProcessorConfig proc_cfg_;
-    EncoderConfig enc_cfg_;
-    uint32_t vk_device_index_ = 0;
+    processors::ProcessorConfig proc_cfg_;
+    encoder::EncoderConfig enc_cfg_;
+    uint32_t vk_device_idx_ = 0;
    AVHWDeviceType hw_device_type_ = AV_HWDEVICE_TYPE_NONE;
    bool benchmark_ = false;

    std::atomic<VideoProcessorState> state_ = VideoProcessorState::Idle;
-    std::atomic<int64_t> frame_index_ = 0;
+    std::atomic<int64_t> frame_idx_ = 0;
    std::atomic<int64_t> total_frames_ = 0;
 };
+
+}  // namespace video2x
--- a/include/libvideo2x/libvideo2x_export.h
+++ b/include/libvideo2x/libvideo2x_export.h
@@ -0,0 +1,11 @@
+#pragma once
+
+#ifdef _WIN32
+#ifdef LIBVIDEO2X_EXPORTS
+#define LIBVIDEO2X_API __declspec(dllexport)
+#else
+#define LIBVIDEO2X_API __declspec(dllimport)
+#endif
+#else
+#define LIBVIDEO2X_API
+#endif
--- a/include/libvideo2x/logger_manager.h
+++ b/include/libvideo2x/logger_manager.h
@@ -0,0 +1,48 @@
+#pragma once
+
+#include <memory>
+#include <string>
+#include <vector>
+
+#include <spdlog/logger.h>
+#include <spdlog/sinks/sink.h>
+
+#include "libvideo2x_export.h"
+
+namespace video2x {
+namespace logger_manager {
+
+class LIBVIDEO2X_API LoggerManager {
+   public:
+    LoggerManager(const LoggerManager &) = delete;
+    LoggerManager &operator=(const LoggerManager &) = delete;
+
+    static LoggerManager &instance();
+
+    std::shared_ptr<spdlog::logger> logger();
+
+    bool reconfigure_logger(
+        const std::string &logger_name,
+        const std::vector<spdlog::sink_ptr> &sinks,
+        const std::string &pattern = "%+"
+    );
+
+    bool set_log_level(const std::string &level_str);
+
+    void hook_ffmpeg_logging();
+    void unhook_ffmpeg_logging();
+
+   private:
+    LoggerManager();
+
+    std::shared_ptr<spdlog::logger> logger_ = nullptr;
+};
+
+}  // namespace logger_manager
+
+// Convenience function to get the logger instance
+inline std::shared_ptr<spdlog::logger> logger() {
+    return logger_manager::LoggerManager::instance().logger();
+}
+
+}  // namespace video2x
--- a/include/libvideo2x/logging.h
+++ b/include/libvideo2x/logging.h
@@ -1,20 +0,0 @@
-#pragma once
-
-#include <optional>
-
-#include "fsutils.h"
-
-enum class Video2xLogLevel {
-    Unknown,
-    Trace,
-    Debug,
-    Info,
-    Warning,
-    Error,
-    Critical,
-    Off
-};
-
-void set_log_level(Video2xLogLevel log_level);
-
-std::optional<Video2xLogLevel> find_log_level_by_name(const StringType &log_level_name);
--- a/include/libvideo2x/processor.h
+++ b/include/libvideo2x/processor.h
@@ -11,6 +11,9 @@ extern "C" {

 #include "fsutils.h"

+namespace video2x {
+namespace processors {
+
 enum class ProcessingMode {
    Filter,
    Interpolate,
@@ -20,16 +23,24 @@ enum class ProcessorType {
    None,
    Libplacebo,
    RealESRGAN,
+    RealCUGAN,
    RIFE,
 };

 struct LibplaceboConfig {
-    StringType shader_path;
+    fsutils::StringType shader_path;
 };

 struct RealESRGANConfig {
    bool tta_mode = false;
-    StringType model_name;
+    fsutils::StringType model_name;
+};
+
+struct RealCUGANConfig {
+    bool tta_mode = false;
+    int num_threads = 1;
+    int syncgap = 3;
+    fsutils::StringType model_name;
 };

 struct RIFEConfig {
@@ -37,7 +48,7 @@ struct RIFEConfig {
    bool tta_temporal_mode = false;
    bool uhd_mode = false;
    int num_threads = 0;
-    StringType model_name;
+    fsutils::StringType model_name;
 };

 // Unified filter configuration
@@ -46,16 +57,17 @@ struct ProcessorConfig {
    int width = 0;
    int height = 0;
    int scaling_factor = 0;
+    int noise_level = -1;
    int frm_rate_mul = 0;
    float scn_det_thresh = 0.0f;
-    std::variant<LibplaceboConfig, RealESRGANConfig, RIFEConfig> config;
+    std::variant<LibplaceboConfig, RealESRGANConfig, RealCUGANConfig, RIFEConfig> config;
 };

 class Processor {
   public:
    virtual ~Processor() = default;
    virtual int init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *hw_ctx) = 0;
-    virtual int flush(std::vector<AVFrame *> &_) { return 0; }
+    virtual int flush(std::vector<AVFrame *> &) { return 0; }
    virtual ProcessingMode get_processing_mode() const = 0;
    virtual ProcessorType get_processor_type() const = 0;
    virtual void get_output_dimensions(
@@ -81,3 +93,6 @@ class Interpolator : public Processor {
    virtual int
    interpolate(AVFrame *prev_frame, AVFrame *in_frame, AVFrame **out_frame, float time_step) = 0;
 };
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/processor_factory.h
+++ b/include/libvideo2x/processor_factory.h
@@ -6,6 +6,9 @@

 #include "processor.h"

+namespace video2x {
+namespace processors {
+
 // Processor Factory Class
 class ProcessorFactory {
   public:
@@ -31,3 +34,6 @@ class ProcessorFactory {
    // Static initializer for default processors
    static void init_default_processors(ProcessorFactory &factory);
 };
+
+}  // namespace processors
+}  // namespace video2x
--- a/include/libvideo2x/version.h.in
+++ b/include/libvideo2x/version.h.in
@@ -1,3 +1,7 @@
 #pragma once

+namespace video2x {
+
 #define LIBVIDEO2X_VERSION_STRING "@PROJECT_VERSION@"
+
+}  // namespace video2x
--- a/models/realcugan/models-nose/up2x-no-denoise.bin
+++ b/models/realcugan/models-nose/up2x-no-denoise.bin
--- a/models/realcugan/models-nose/up2x-no-denoise.param
+++ b/models/realcugan/models-nose/up2x-no-denoise.param
@@ -0,0 +1,37 @@
+7767517
+35 39
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 4 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Crop                     slice_43                 1 1 5 9 -23309=2,4,4 -23310=2,-4,-4 -23311=2,1,2
+Deconvolution            deconvrelu_0             1 1 8 10 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_0                    2 1 10 9 11
+Convolution              convrelu_5               1 1 11 12 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_20                1 1 12 13 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_1              1 2 13 14 15
+Convolution              convrelu_6               1 1 14 16 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_7               1 1 16 17 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_2              1 2 17 18 19
+Convolution              convrelu_8               1 1 18 20 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_9               1 1 20 21 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 21 22 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 22 23 24
+Convolution              convrelu_11              1 1 23 25 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_12              1 1 25 26 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_13              1 1 26 27 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Crop                     slice_44                 1 1 24 28 -23309=2,4,4 -23310=2,-4,-4 -23311=2,1,2
+Deconvolution            deconvrelu_1             1 1 27 29 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 29 28 30
+Convolution              convrelu_14              1 1 30 31 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 31 32 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Crop                     slice_45                 1 1 19 33 -23309=2,16,16 -23310=2,-16,-16 -23311=2,1,2
+Deconvolution            deconvrelu_2             1 1 32 34 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_2                    2 1 34 33 35
+Convolution              convrelu_16              1 1 35 36 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_18                  1 1 36 37 0=3 1=3 5=1 6=1728
+Crop                     slice_46                 1 1 15 38 -23309=2,20,20 -23310=2,-20,-20 -23311=2,1,2
+BinaryOp                 add_3                    2 1 37 38 out0
--- a/models/realcugan/models-pro/up2x-conservative.bin
+++ b/models/realcugan/models-pro/up2x-conservative.bin
--- a/models/realcugan/models-pro/up2x-conservative.param
+++ b/models/realcugan/models-pro/up2x-conservative.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-pro/up2x-denoise3x.bin
+++ b/models/realcugan/models-pro/up2x-denoise3x.bin
--- a/models/realcugan/models-pro/up2x-denoise3x.param
+++ b/models/realcugan/models-pro/up2x-denoise3x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-pro/up2x-no-denoise.bin
+++ b/models/realcugan/models-pro/up2x-no-denoise.bin
--- a/models/realcugan/models-pro/up2x-no-denoise.param
+++ b/models/realcugan/models-pro/up2x-no-denoise.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-pro/up3x-conservative.bin
+++ b/models/realcugan/models-pro/up3x-conservative.bin
--- a/models/realcugan/models-pro/up3x-conservative.param
+++ b/models/realcugan/models-pro/up3x-conservative.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-pro/up3x-denoise3x.bin
+++ b/models/realcugan/models-pro/up3x-denoise3x.bin
--- a/models/realcugan/models-pro/up3x-denoise3x.param
+++ b/models/realcugan/models-pro/up3x-denoise3x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-pro/up3x-no-denoise.bin
+++ b/models/realcugan/models-pro/up3x-no-denoise.bin
--- a/models/realcugan/models-pro/up3x-no-denoise.param
+++ b/models/realcugan/models-pro/up3x-no-denoise.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 10 11 12
+Pooling                  mean_64                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_2                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_3                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 20 21 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 30 31 32
+Pooling                  mean_65                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_4                    2 1 31 35 36 0=2
+Split                    splitncnn_5              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 41 42 43
+Pooling                  mean_66                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_5                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 52 53 54
+Pooling                  mean_67                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_7                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_8                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 63 64 0=3 1=3 5=1 6=1728
+BinaryOp                 add_9                    2 1 64 62 out0
--- a/models/realcugan/models-se/up2x-conservative.bin
+++ b/models/realcugan/models-se/up2x-conservative.bin
--- a/models/realcugan/models-se/up2x-conservative.param
+++ b/models/realcugan/models-se/up2x-conservative.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up2x-denoise1x.bin
+++ b/models/realcugan/models-se/up2x-denoise1x.bin
--- a/models/realcugan/models-se/up2x-denoise1x.param
+++ b/models/realcugan/models-se/up2x-denoise1x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up2x-denoise2x.bin
+++ b/models/realcugan/models-se/up2x-denoise2x.bin
--- a/models/realcugan/models-se/up2x-denoise2x.param
+++ b/models/realcugan/models-se/up2x-denoise2x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up2x-denoise3x.bin
+++ b/models/realcugan/models-se/up2x-denoise3x.bin
--- a/models/realcugan/models-se/up2x-denoise3x.param
+++ b/models/realcugan/models-se/up2x-denoise3x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up2x-no-denoise.bin
+++ b/models/realcugan/models-se/up2x-no-denoise.bin
--- a/models/realcugan/models-se/up2x-no-denoise.param
+++ b/models/realcugan/models-se/up2x-no-denoise.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=4 3=2 4=3 5=1 6=3072
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up3x-conservative.bin
+++ b/models/realcugan/models-se/up3x-conservative.bin
--- a/models/realcugan/models-se/up3x-conservative.param
+++ b/models/realcugan/models-se/up3x-conservative.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up3x-denoise3x.bin
+++ b/models/realcugan/models-se/up3x-denoise3x.bin
--- a/models/realcugan/models-se/up3x-denoise3x.param
+++ b/models/realcugan/models-se/up3x-denoise3x.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up3x-no-denoise.bin
+++ b/models/realcugan/models-se/up3x-no-denoise.bin
--- a/models/realcugan/models-se/up3x-no-denoise.param
+++ b/models/realcugan/models-se/up3x-no-denoise.param
@@ -0,0 +1,57 @@
+7767517
+55 63
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 2 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 2 3 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_0              1 2 3 4 5
+Convolution              convrelu_2               1 1 5 6 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 6 7 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 7 8 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 8 9 10
+Pooling                  mean_64                  1 1 10 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 12 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 12 13 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 9 13 14 0=2
+Crop                     pad_9                    1 1 4 15 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 14 16 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 15 16 17
+Convolution              convrelu_6               1 1 17 18 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_48                1 1 18 19 0=3 1=5 3=3 4=2 5=1 6=4800
+Split                    splitncnn_2              1 2 19 20 21
+Convolution              convrelu_7               1 1 21 22 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 22 23 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_3              1 2 23 24 25
+Convolution              convrelu_9               1 1 25 26 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 26 27 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 27 28 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 28 29 30
+Pooling                  mean_65                  1 1 30 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 32 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 32 33 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 29 33 34 0=2
+Split                    splitncnn_5              1 2 34 35 36
+Convolution              convrelu_13              1 1 36 37 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 37 38 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 38 39 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_6              1 2 39 40 41
+Pooling                  mean_66                  1 1 41 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 43 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 43 44 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 40 44 45 0=2
+Crop                     pad_10                   1 1 35 46 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 45 47 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 46 47 48
+Convolution              convrelu_17              1 1 48 49 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 49 50 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 50 51 52
+Pooling                  mean_67                  1 1 52 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 54 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 54 55 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 51 55 56 0=2
+Crop                     pad_11                   1 1 24 57 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 56 58 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 57 58 59
+Crop                     pad_12                   1 1 20 60 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 59 61 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_46                  1 1 61 62 0=3 1=3 5=1 6=1728
+BinaryOp                 add_7                    2 1 62 60 out0
--- a/models/realcugan/models-se/up4x-conservative.bin
+++ b/models/realcugan/models-se/up4x-conservative.bin
--- a/models/realcugan/models-se/up4x-conservative.param
+++ b/models/realcugan/models-se/up4x-conservative.param
@@ -0,0 +1,60 @@
+7767517
+58 66
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_2              1 2 10 11 12
+Pooling                  mean_68                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_51                1 1 20 21 0=64 1=4 3=2 4=3 5=1 6=65536
+Split                    splitncnn_3              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_5              1 2 30 31 32
+Pooling                  mean_69                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 31 35 36 0=2
+Split                    splitncnn_6              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 41 42 43
+Pooling                  mean_70                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_8              1 2 52 53 54
+Pooling                  mean_71                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_48                  1 1 63 64 0=64 1=3 5=1 6=36864
+BinaryOp                 add_7                    2 1 64 62 65
+Convolution              conv_49                  1 1 65 66 0=12 1=3 5=1 6=6912
+Crop                     pad_13                   1 1 66 67 -23309=2,1,1 -23310=2,-1,-1 -23311=2,-2,-1
+PixelShuffle             pixelshuffle_66          1 1 67 out0 0=2
--- a/models/realcugan/models-se/up4x-denoise3x.bin
+++ b/models/realcugan/models-se/up4x-denoise3x.bin
--- a/models/realcugan/models-se/up4x-denoise3x.param
+++ b/models/realcugan/models-se/up4x-denoise3x.param
@@ -0,0 +1,60 @@
+7767517
+58 66
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_2              1 2 10 11 12
+Pooling                  mean_68                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_51                1 1 20 21 0=64 1=4 3=2 4=3 5=1 6=65536
+Split                    splitncnn_3              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_5              1 2 30 31 32
+Pooling                  mean_69                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 31 35 36 0=2
+Split                    splitncnn_6              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 41 42 43
+Pooling                  mean_70                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_8              1 2 52 53 54
+Pooling                  mean_71                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_48                  1 1 63 64 0=64 1=3 5=1 6=36864
+BinaryOp                 add_7                    2 1 64 62 65
+Convolution              conv_49                  1 1 65 66 0=12 1=3 5=1 6=6912
+Crop                     pad_13                   1 1 66 67 -23309=2,1,1 -23310=2,-1,-1 -23311=2,-2,-1
+PixelShuffle             pixelshuffle_66          1 1 67 out0 0=2
--- a/models/realcugan/models-se/up4x-no-denoise.bin
+++ b/models/realcugan/models-se/up4x-no-denoise.bin
--- a/models/realcugan/models-se/up4x-no-denoise.param
+++ b/models/realcugan/models-se/up4x-no-denoise.param
@@ -0,0 +1,60 @@
+7767517
+58 66
+Input                    in0                      0 1 in0
+Convolution              convrelu_0               1 1 in0 4 0=32 1=3 5=1 6=864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_1               1 1 4 5 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_1              1 2 5 6 7
+Convolution              convrelu_2               1 1 7 8 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_3               1 1 8 9 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_4               1 1 9 10 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_2              1 2 10 11 12
+Pooling                  mean_68                  1 1 12 gap0 0=1 4=1
+InnerProduct             convrelu_5               1 1 gap0 14 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_21           1 1 14 15 0=64 1=1 2=512 9=4
+BinaryOp                 mul_0                    2 1 11 15 16 0=2
+Crop                     pad_9                    1 1 6 17 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_0             1 1 16 18 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_1                    2 1 17 18 19
+Convolution              convrelu_6               1 1 19 20 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Deconvolution            deconv_51                1 1 20 21 0=64 1=4 3=2 4=3 5=1 6=65536
+Split                    splitncnn_3              1 2 21 22 23
+Convolution              convrelu_7               1 1 23 24 0=32 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_8               1 1 24 25 0=64 1=3 5=1 6=18432 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_4              1 2 25 26 27
+Convolution              convrelu_9               1 1 27 28 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_10              1 1 28 29 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_11              1 1 29 30 0=128 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_5              1 2 30 31 32
+Pooling                  mean_69                  1 1 32 gap1 0=1 4=1
+InnerProduct             convrelu_12              1 1 gap1 34 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_22           1 1 34 35 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_2                    2 1 31 35 36 0=2
+Split                    splitncnn_6              1 2 36 37 38
+Convolution              convrelu_13              1 1 38 39 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_14              1 1 39 40 0=256 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_15              1 1 40 41 0=128 1=3 5=1 6=294912 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_7              1 2 41 42 43
+Pooling                  mean_70                  1 1 43 gap2 0=1 4=1
+InnerProduct             convrelu_16              1 1 gap2 45 0=16 1=1 2=2048 9=1
+InnerProduct             convsigmoid_23           1 1 45 46 0=128 1=1 2=2048 9=4
+BinaryOp                 mul_3                    2 1 42 46 47 0=2
+Crop                     pad_10                   1 1 37 48 -23309=2,4,4 -23310=2,-4,-4 -23311=2,-2,-1
+Deconvolution            deconvrelu_1             1 1 47 49 0=128 1=2 3=2 5=1 6=65536 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_4                    2 1 48 49 50
+Convolution              convrelu_17              1 1 50 51 0=64 1=3 5=1 6=73728 9=2 -23310=1,1.000000e-01
+Convolution              convrelu_18              1 1 51 52 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Split                    splitncnn_8              1 2 52 53 54
+Pooling                  mean_71                  1 1 54 gap3 0=1 4=1
+InnerProduct             convrelu_19              1 1 gap3 56 0=8 1=1 2=512 9=1
+InnerProduct             convsigmoid_24           1 1 56 57 0=64 1=1 2=512 9=4
+BinaryOp                 mul_5                    2 1 53 57 58 0=2
+Crop                     pad_11                   1 1 26 59 -23309=2,16,16 -23310=2,-16,-16 -23311=2,-2,-1
+Deconvolution            deconvrelu_2             1 1 58 60 0=64 1=2 3=2 5=1 6=16384 9=2 -23310=1,1.000000e-01
+BinaryOp                 add_6                    2 1 59 60 61
+Crop                     pad_12                   1 1 22 62 -23309=2,20,20 -23310=2,-20,-20 -23311=2,-2,-1
+Convolution              convrelu_20              1 1 61 63 0=64 1=3 5=1 6=36864 9=2 -23310=1,1.000000e-01
+Convolution              conv_48                  1 1 63 64 0=64 1=3 5=1 6=36864
+BinaryOp                 add_7                    2 1 64 62 65
+Convolution              conv_49                  1 1 65 66 0=12 1=3 5=1 6=6912
+Crop                     pad_13                   1 1 66 67 -23309=2,1,1 -23310=2,-1,-1 -23311=2,-2,-1
+PixelShuffle             pixelshuffle_66          1 1 67 out0 0=2
--- a/src/avutils.cpp
+++ b/src/avutils.cpp
@@ -10,6 +10,10 @@ extern "C" {
 #include <spdlog/spdlog.h>

 #include "conversions.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace avutils {

 AVRational get_video_frame_rate(AVFormatContext *ifmt_ctx, int in_vstream_idx) {
    AVRational frame_rate = ifmt_ctx->streams[in_vstream_idx]->avg_frame_rate;
@@ -75,7 +79,7 @@ AVPixelFormat get_encoder_default_pix_fmt(const AVCodec *encoder, AVPixelFormat
    );
    if (ret < 0) {
        av_strerror(ret, errbuf, sizeof(errbuf));
-        spdlog::error("Failed to get supported pixel formats: {}", errbuf);
+        logger()->error("Failed to get supported pixel formats: {}", errbuf);
        return AV_PIX_FMT_NONE;
    }

@@ -116,7 +120,7 @@ AVPixelFormat get_encoder_default_pix_fmt(const AVCodec *encoder, AVPixelFormat
        }
    }
    if (best_pix_fmt == AV_PIX_FMT_NONE) {
-        spdlog::error("No suitable pixel format found for encoder");
+        logger()->error("No suitable pixel format found for encoder");
    }

    if (target_pix_fmt != AV_PIX_FMT_NONE && best_pix_fmt != target_pix_fmt) {
@@ -133,12 +137,12 @@ AVPixelFormat get_encoder_default_pix_fmt(const AVCodec *encoder, AVPixelFormat

 float get_frame_diff(AVFrame *frame1, AVFrame *frame2) {
    if (!frame1 || !frame2) {
-        spdlog::error("Invalid frame(s) provided for comparison");
+        logger()->error("Invalid frame(s) provided for comparison");
        return -1.0f;
    }

    if (frame1->width != frame2->width || frame1->height != frame2->height) {
-        spdlog::error("Frame dimensions do not match");
+        logger()->error("Frame dimensions do not match");
        return -1.0f;
    }

@@ -147,11 +151,11 @@ float get_frame_diff(AVFrame *frame1, AVFrame *frame2) {

    // Convert both frames to the target pixel format using the provided function
    AVPixelFormat target_pix_fmt = AV_PIX_FMT_RGB24;
-    AVFrame *rgb_frame1 = convert_avframe_pix_fmt(frame1, target_pix_fmt);
-    AVFrame *rgb_frame2 = convert_avframe_pix_fmt(frame2, target_pix_fmt);
+    AVFrame *rgb_frame1 = conversions::convert_avframe_pix_fmt(frame1, target_pix_fmt);
+    AVFrame *rgb_frame2 = conversions::convert_avframe_pix_fmt(frame2, target_pix_fmt);

    if (!rgb_frame1 || !rgb_frame2) {
-        spdlog::error("Failed to convert frames to target pixel format");
+        logger()->error("Failed to convert frames to target pixel format");
        if (rgb_frame1) {
            av_frame_free(&rgb_frame1);
        }
@@ -208,3 +212,6 @@ void av_packet_deleter(AVPacket *packet) {
        packet = nullptr;
    }
 }
+
+}  // namespace avutils
+}  // namespace video2x
--- a/src/conversions.cpp
+++ b/src/conversions.cpp
@@ -5,11 +5,16 @@

 #include <spdlog/spdlog.h>

+#include "logger_manager.h"
+
+namespace video2x {
+namespace conversions {
+
 // Convert AVFrame format
 AVFrame *convert_avframe_pix_fmt(AVFrame *src_frame, AVPixelFormat pix_fmt) {
    AVFrame *dst_frame = av_frame_alloc();
    if (dst_frame == nullptr) {
-        spdlog::error("Failed to allocate destination AVFrame.");
+        logger()->error("Failed to allocate destination AVFrame.");
        return nullptr;
    }

@@ -19,7 +24,7 @@ AVFrame *convert_avframe_pix_fmt(AVFrame *src_frame, AVPixelFormat pix_fmt) {

    // Allocate memory for the converted frame
    if (av_frame_get_buffer(dst_frame, 32) < 0) {
-        spdlog::error("Failed to allocate memory for AVFrame.");
+        logger()->error("Failed to allocate memory for AVFrame.");
        av_frame_free(&dst_frame);
        return nullptr;
    }
@@ -39,7 +44,7 @@ AVFrame *convert_avframe_pix_fmt(AVFrame *src_frame, AVPixelFormat pix_fmt) {
    );

    if (sws_ctx == nullptr) {
-        spdlog::error("Failed to initialize swscale context.");
+        logger()->error("Failed to initialize swscale context.");
        av_frame_free(&dst_frame);
        return nullptr;
    }
@@ -69,7 +74,7 @@ ncnn::Mat avframe_to_ncnn_mat(AVFrame *frame) {
    if (frame->format != AV_PIX_FMT_BGR24) {
        converted_frame = convert_avframe_pix_fmt(frame, AV_PIX_FMT_BGR24);
        if (!converted_frame) {
-            spdlog::error("Failed to convert AVFrame to BGR24.");
+            logger()->error("Failed to convert AVFrame to BGR24.");
            return ncnn::Mat();
        }
    } else {
@@ -107,7 +112,7 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {
    // Step 1: Allocate a destination AVFrame for the specified pixel format
    AVFrame *dst_frame = av_frame_alloc();
    if (!dst_frame) {
-        spdlog::error("Failed to allocate destination AVFrame.");
+        logger()->error("Failed to allocate destination AVFrame.");
        return nullptr;
    }

@@ -117,7 +122,7 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {

    // Allocate memory for the frame buffer
    if (av_frame_get_buffer(dst_frame, 32) < 0) {
-        spdlog::error("Failed to allocate memory for destination AVFrame.");
+        logger()->error("Failed to allocate memory for destination AVFrame.");
        av_frame_free(&dst_frame);
        return nullptr;
    }
@@ -125,7 +130,7 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {
    // Step 2: Convert ncnn::Mat to BGR AVFrame
    AVFrame *bgr_frame = av_frame_alloc();
    if (!bgr_frame) {
-        spdlog::error("Failed to allocate intermediate BGR AVFrame.");
+        logger()->error("Failed to allocate intermediate BGR AVFrame.");
        av_frame_free(&dst_frame);
        return nullptr;
    }
@@ -136,7 +141,7 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {

    // Allocate memory for the intermediate BGR frame
    if (av_frame_get_buffer(bgr_frame, 32) < 0) {
-        spdlog::error("Failed to allocate memory for BGR AVFrame.");
+        logger()->error("Failed to allocate memory for BGR AVFrame.");
        av_frame_free(&dst_frame);
        av_frame_free(&bgr_frame);
        return nullptr;
@@ -166,7 +171,7 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {
    );

    if (sws_ctx == nullptr) {
-        spdlog::error("Failed to initialize swscale context.");
+        logger()->error("Failed to initialize swscale context.");
        av_frame_free(&bgr_frame);
        av_frame_free(&dst_frame);
        return nullptr;
@@ -188,10 +193,13 @@ AVFrame *ncnn_mat_to_avframe(const ncnn::Mat &mat, AVPixelFormat pix_fmt) {
    av_frame_free(&bgr_frame);

    if (ret != dst_frame->height) {
-        spdlog::error("Failed to convert BGR AVFrame to destination pixel format.");
+        logger()->error("Failed to convert BGR AVFrame to destination pixel format.");
        av_frame_free(&dst_frame);
        return nullptr;
    }

    return dst_frame;
 }
+
+}  // namespace conversions
+}  // namespace video2x
--- a/src/decoder.cpp
+++ b/src/decoder.cpp
@@ -2,6 +2,11 @@

 #include <spdlog/spdlog.h>

+#include "logger_manager.h"
+
+namespace video2x {
+namespace decoder {
+
 AVPixelFormat Decoder::hw_pix_fmt_ = AV_PIX_FMT_NONE;

 Decoder::Decoder() : fmt_ctx_(nullptr), dec_ctx_(nullptr), in_vstream_idx_(-1) {}
@@ -23,7 +28,7 @@ AVPixelFormat Decoder::get_hw_format(AVCodecContext *_, const AVPixelFormat *pix
            return *p;
        }
    }
-    spdlog::error("Failed to get HW surface format.");
+    logger()->error("Failed to get HW surface format.");
    return AV_PIX_FMT_NONE;
 }

@@ -36,20 +41,20 @@ int Decoder::init(

    // Open the input file
    if ((ret = avformat_open_input(&fmt_ctx_, in_fpath.u8string().c_str(), nullptr, nullptr)) < 0) {
-        spdlog::error("Could not open input file '{}'", in_fpath.u8string());
+        logger()->error("Could not open input file '{}'", in_fpath.u8string());
        return ret;
    }

    // Retrieve stream information
    if ((ret = avformat_find_stream_info(fmt_ctx_, nullptr)) < 0) {
-        spdlog::error("Failed to retrieve input stream information");
+        logger()->error("Failed to retrieve input stream information");
        return ret;
    }

    // Find the first video stream
    ret = av_find_best_stream(fmt_ctx_, AVMEDIA_TYPE_VIDEO, -1, -1, nullptr, 0);
    if (ret < 0) {
-        spdlog::error("Could not find video stream in the input file");
+        logger()->error("Could not find video stream in the input file");
        return ret;
    }

@@ -59,7 +64,7 @@ int Decoder::init(
    // Find the decoder for the video stream
    const AVCodec *decoder = avcodec_find_decoder(video_stream->codecpar->codec_id);
    if (!decoder) {
-        spdlog::error(
+        logger()->error(
            "Failed to find decoder for codec ID {}",
            static_cast<int>(video_stream->codecpar->codec_id)
        );
@@ -69,13 +74,13 @@ int Decoder::init(
    // Allocate the decoder context
    dec_ctx_ = avcodec_alloc_context3(decoder);
    if (!dec_ctx_) {
-        spdlog::error("Failed to allocate the decoder context");
+        logger()->error("Failed to allocate the decoder context");
        return AVERROR(ENOMEM);
    }

    // Copy codec parameters from input stream to decoder context
    if ((ret = avcodec_parameters_to_context(dec_ctx_, video_stream->codecpar)) < 0) {
-        spdlog::error("Failed to copy decoder parameters to input decoder context");
+        logger()->error("Failed to copy decoder parameters to input decoder context");
        return ret;
    }

@@ -93,7 +98,7 @@ int Decoder::init(
        for (int i = 0;; i++) {
            const AVCodecHWConfig *config = avcodec_get_hw_config(decoder, i);
            if (config == nullptr) {
-                spdlog::error(
+                logger()->error(
                    "Decoder {} does not support device type {}.",
                    decoder->name,
                    av_hwdevice_get_type_name(hw_type)
@@ -110,7 +115,7 @@ int Decoder::init(

    // Open the decoder
    if ((ret = avcodec_open2(dec_ctx_, decoder, nullptr)) < 0) {
-        spdlog::error("Failed to open decoder for stream #{}", stream_index);
+        logger()->error("Failed to open decoder for stream #{}", stream_index);
        return ret;
    }

@@ -130,3 +135,6 @@ AVCodecContext *Decoder::get_codec_context() const {
 int Decoder::get_video_stream_index() const {
    return in_vstream_idx_;
 }
+
+}  // namespace decoder
+}  // namespace video2x
--- a/src/encoder.cpp
+++ b/src/encoder.cpp
@@ -6,9 +6,14 @@ extern "C" {
 #include <libavutil/opt.h>
 }

+#include "logger_manager.h"
+
 #include "avutils.h"
 #include "conversions.h"

+namespace video2x {
+namespace encoder {
+
 Encoder::Encoder()
    : ofmt_ctx_(nullptr), enc_ctx_(nullptr), out_vstream_idx_(-1), stream_map_(nullptr) {}

@@ -43,14 +48,14 @@ int Encoder::init(
    // Allocate the output format context
    avformat_alloc_output_context2(&ofmt_ctx_, nullptr, nullptr, out_fpath.u8string().c_str());
    if (!ofmt_ctx_) {
-        spdlog::error("Could not create output context");
+        logger()->error("Could not create output context");
        return AVERROR_UNKNOWN;
    }

    // Find the encoder
    const AVCodec *encoder = avcodec_find_encoder(enc_cfg.codec);
    if (!encoder) {
-        spdlog::error(
+        logger()->error(
            "Required video encoder not found for codec {}", avcodec_get_name(enc_cfg.codec)
        );
        return AVERROR_ENCODER_NOT_FOUND;
@@ -59,7 +64,7 @@ int Encoder::init(
    // Create a new video stream in the output file
    AVStream *out_vstream = avformat_new_stream(ofmt_ctx_, nullptr);
    if (!out_vstream) {
-        spdlog::error("Failed to allocate the output video stream");
+        logger()->error("Failed to allocate the output video stream");
        return AVERROR_UNKNOWN;
    }
    out_vstream_idx_ = out_vstream->index;
@@ -67,7 +72,7 @@ int Encoder::init(
    // Allocate the encoder context
    enc_ctx_ = avcodec_alloc_context3(encoder);
    if (!enc_ctx_) {
-        spdlog::error("Failed to allocate the encoder context");
+        logger()->error("Failed to allocate the encoder context");
        return AVERROR(ENOMEM);
    }

@@ -116,16 +121,16 @@ int Encoder::init(
        enc_ctx_->pix_fmt = enc_cfg.pix_fmt;
    } else {
        // Automatically select the pixel format
-        enc_ctx_->pix_fmt = get_encoder_default_pix_fmt(encoder, dec_ctx->pix_fmt);
+        enc_ctx_->pix_fmt = avutils::get_encoder_default_pix_fmt(encoder, dec_ctx->pix_fmt);
        if (enc_ctx_->pix_fmt == AV_PIX_FMT_NONE) {
-            spdlog::error("Could not get the default pixel format for the encoder");
+            logger()->error("Could not get the default pixel format for the encoder");
            return AVERROR(EINVAL);
        }
        spdlog::debug("Auto-selected pixel format: {}", av_get_pix_fmt_name(enc_ctx_->pix_fmt));
    }

    if (frm_rate_mul > 0) {
-        AVRational in_frame_rate = get_video_frame_rate(ifmt_ctx, in_vstream_idx);
+        AVRational in_frame_rate = avutils::get_video_frame_rate(ifmt_ctx, in_vstream_idx);
        enc_ctx_->framerate = {in_frame_rate.num * frm_rate_mul, in_frame_rate.den};
        enc_ctx_->time_base = av_inv_q(enc_ctx_->framerate);
    } else {
@@ -146,8 +151,8 @@ int Encoder::init(

    // Set extra AVOptions
    for (const auto &[opt_name, opt_value] : enc_cfg.extra_opts) {
-        std::string opt_name_str = wstring_to_u8string(opt_name);
-        std::string opt_value_str = wstring_to_u8string(opt_value);
+        std::string opt_name_str = fsutils::wstring_to_u8string(opt_name);
+        std::string opt_value_str = fsutils::wstring_to_u8string(opt_value);
        spdlog::debug("Setting encoder option '{}' to '{}'", opt_name_str, opt_value_str);

        if (av_opt_set(enc_ctx_->priv_data, opt_name_str.c_str(), opt_value_str.c_str(), 0) < 0) {
@@ -162,14 +167,14 @@ int Encoder::init(

    // Open the encoder
    if ((ret = avcodec_open2(enc_ctx_, encoder, nullptr)) < 0) {
-        spdlog::error("Cannot open video encoder");
+        logger()->error("Cannot open video encoder");
        return ret;
    }

    // Copy encoder parameters to output video stream
    ret = avcodec_parameters_from_context(out_vstream->codecpar, enc_ctx_);
    if (ret < 0) {
-        spdlog::error("Failed to copy encoder parameters to output video stream");
+        logger()->error("Failed to copy encoder parameters to output video stream");
        return ret;
    }

@@ -183,7 +188,7 @@ int Encoder::init(
        stream_map_ =
            reinterpret_cast<int *>(av_malloc_array(ifmt_ctx->nb_streams, sizeof(*stream_map_)));
        if (!stream_map_) {
-            spdlog::error("Could not allocate stream mapping");
+            logger()->error("Could not allocate stream mapping");
            return AVERROR(ENOMEM);
        }

@@ -209,14 +214,14 @@ int Encoder::init(
            // Create corresponding output stream for audio and subtitle streams
            AVStream *out_stream = avformat_new_stream(ofmt_ctx_, nullptr);
            if (!out_stream) {
-                spdlog::error("Failed allocating output stream");
+                logger()->error("Failed allocating output stream");
                return AVERROR_UNKNOWN;
            }

            // Copy codec parameters from input to output
            ret = avcodec_parameters_copy(out_stream->codecpar, in_codecpar);
            if (ret < 0) {
-                spdlog::error("Failed to copy codec parameters");
+                logger()->error("Failed to copy codec parameters");
                return ret;
            }
            out_stream->codecpar->codec_tag = 0;
@@ -234,7 +239,7 @@ int Encoder::init(
    if (!(ofmt_ctx_->oformat->flags & AVFMT_NOFILE)) {
        ret = avio_open(&ofmt_ctx_->pb, out_fpath.u8string().c_str(), AVIO_FLAG_WRITE);
        if (ret < 0) {
-            spdlog::error("Could not open output file '{}'", out_fpath.u8string());
+            logger()->error("Could not open output file '{}'", out_fpath.u8string());
            return ret;
        }
    }
@@ -242,7 +247,7 @@ int Encoder::init(
    // Write the output file header
    ret = avformat_write_header(ofmt_ctx_, nullptr);
    if (ret < 0) {
-        spdlog::error("Error writing output file header");
+        logger()->error("Error writing output file header");
        return ret;
    }

@@ -253,16 +258,17 @@ int Encoder::write_frame(AVFrame *frame, int64_t frame_idx) {
    AVFrame *converted_frame = nullptr;
    int ret;

-    // Set the frame's presentation timestamp if not set
-    if (frame->pts <= 0) {
-        frame->pts = frame_idx;
-    }
+    // Let the encoder decide the frame type
+    frame->pict_type = AV_PICTURE_TYPE_NONE;
+
+    // Calculate this frame's presentation timestamp (PTS)
+    frame->pts = av_rescale_q(frame_idx, av_inv_q(enc_ctx_->framerate), enc_ctx_->time_base);

    // Convert the frame to the encoder's pixel format if needed
    if (frame->format != enc_ctx_->pix_fmt) {
-        converted_frame = convert_avframe_pix_fmt(frame, enc_ctx_->pix_fmt);
+        converted_frame = conversions::convert_avframe_pix_fmt(frame, enc_ctx_->pix_fmt);
        if (!converted_frame) {
-            spdlog::error("Error converting frame to encoder's pixel format");
+            logger()->error("Error converting frame to encoder's pixel format");
            return AVERROR_EXTERNAL;
        }
        converted_frame->pts = frame->pts;
@@ -270,7 +276,7 @@ int Encoder::write_frame(AVFrame *frame, int64_t frame_idx) {

    AVPacket *enc_pkt = av_packet_alloc();
    if (!enc_pkt) {
-        spdlog::error("Could not allocate AVPacket");
+        logger()->error("Could not allocate AVPacket");
        return AVERROR(ENOMEM);
    }

@@ -282,7 +288,7 @@ int Encoder::write_frame(AVFrame *frame, int64_t frame_idx) {
        ret = avcodec_send_frame(enc_ctx_, frame);
    }
    if (ret < 0) {
-        spdlog::error("Error sending frame to encoder");
+        logger()->error("Error sending frame to encoder");
        av_packet_free(&enc_pkt);
        return ret;
    }
@@ -294,7 +300,7 @@ int Encoder::write_frame(AVFrame *frame, int64_t frame_idx) {
            av_packet_unref(enc_pkt);
            break;
        } else if (ret < 0) {
-            spdlog::error("Error encoding frame");
+            logger()->error("Error encoding frame");
            av_packet_free(&enc_pkt);
            return ret;
        }
@@ -309,7 +315,7 @@ int Encoder::write_frame(AVFrame *frame, int64_t frame_idx) {
        ret = av_interleaved_write_frame(ofmt_ctx_, enc_pkt);
        av_packet_unref(enc_pkt);
        if (ret < 0) {
-            spdlog::error("Error muxing packet");
+            logger()->error("Error muxing packet");
            av_packet_free(&enc_pkt);
            return ret;
        }
@@ -323,14 +329,14 @@ int Encoder::flush() {
    int ret;
    AVPacket *enc_pkt = av_packet_alloc();
    if (!enc_pkt) {
-        spdlog::error("Could not allocate AVPacket");
+        logger()->error("Could not allocate AVPacket");
        return AVERROR(ENOMEM);
    }

    // Send a NULL frame to signal the encoder to flush
    ret = avcodec_send_frame(enc_ctx_, nullptr);
    if (ret < 0) {
-        spdlog::error("Error sending NULL frame to encoder during flush");
+        logger()->error("Error sending NULL frame to encoder during flush");
        av_packet_free(&enc_pkt);
        return ret;
    }
@@ -342,7 +348,7 @@ int Encoder::flush() {
            av_packet_unref(enc_pkt);
            break;
        } else if (ret < 0) {
-            spdlog::error("Error encoding packet during flush");
+            logger()->error("Error encoding packet during flush");
            av_packet_free(&enc_pkt);
            return ret;
        }
@@ -357,7 +363,7 @@ int Encoder::flush() {
        ret = av_interleaved_write_frame(ofmt_ctx_, enc_pkt);
        av_packet_unref(enc_pkt);
        if (ret < 0) {
-            spdlog::error("Error muxing packet during flush");
+            logger()->error("Error muxing packet during flush");
            av_packet_free(&enc_pkt);
            return ret;
        }
@@ -382,3 +388,6 @@ int Encoder::get_output_video_stream_index() const {
 int *Encoder::get_stream_map() const {
    return stream_map_;
 }
+
+}  // namespace encoder
+}  // namespace video2x
--- a/src/filter_libplacebo.cpp
+++ b/src/filter_libplacebo.cpp
@@ -6,6 +6,10 @@

 #include "fsutils.h"
 #include "libplacebo.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {

 FilterLibplacebo::FilterLibplacebo(
    uint32_t vk_device_index,
@@ -39,20 +43,20 @@ FilterLibplacebo::~FilterLibplacebo() {
 int FilterLibplacebo::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *) {
    // Construct the shader path
    std::filesystem::path shader_full_path;
-    if (filepath_is_readable(shader_path_)) {
+    if (fsutils::filepath_is_readable(shader_path_)) {
        // If the shader path is directly readable, use it
        shader_full_path = shader_path_;
    } else {
        // Construct the fallback path using std::filesystem
-        shader_full_path = find_resource_file(
+        shader_full_path = fsutils::find_resource_file(
            std::filesystem::path(STR("models")) / STR("libplacebo") /
-            (path_to_string_type(shader_path_) + STR(".glsl"))
+            (fsutils::path_to_string_type(shader_path_) + STR(".glsl"))
        );
    }

    // Check if the shader file exists
    if (!std::filesystem::exists(shader_full_path)) {
-        spdlog::error("libplacebo shader file not found: '{}'", shader_path_.u8string());
+        logger()->error("libplacebo shader file not found: '{}'", shader_path_.u8string());
        return -1;
    }

@@ -87,14 +91,14 @@ int FilterLibplacebo::filter(AVFrame *in_frame, AVFrame **out_frame) {
    // Get the filtered frame
    *out_frame = av_frame_alloc();
    if (*out_frame == nullptr) {
-        spdlog::error("Failed to allocate output frame");
+        logger()->error("Failed to allocate output frame");
        return -1;
    }

    // Feed the frame to the filter graph
    ret = av_buffersrc_add_frame(buffersrc_ctx_, in_frame);
    if (ret < 0) {
-        spdlog::error("Error while feeding the filter graph");
+        logger()->error("Error while feeding the filter graph");
        av_frame_free(out_frame);
        return ret;
    }
@@ -115,7 +119,7 @@ int FilterLibplacebo::filter(AVFrame *in_frame, AVFrame **out_frame) {
 int FilterLibplacebo::flush(std::vector<AVFrame *> &flushed_frames) {
    int ret = av_buffersrc_add_frame(buffersrc_ctx_, nullptr);
    if (ret < 0) {
-        spdlog::error("Error while flushing filter graph");
+        logger()->error("Error while flushing filter graph");
        return ret;
    }

@@ -156,3 +160,6 @@ void FilterLibplacebo::get_output_dimensions(
    out_width = proc_cfg.width;
    out_height = proc_cfg.height;
 }
+
+}  // namespace processors
+}  // namespace video2x
--- a/src/filter_realcugan.cpp
+++ b/src/filter_realcugan.cpp
@@ -0,0 +1,207 @@
+#include "filter_realcugan.h"
+
+#include <cstdint>
+#include <cstdio>
+#include <filesystem>
+
+#include <spdlog/spdlog.h>
+
+#include "conversions.h"
+#include "fsutils.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {
+
+FilterRealcugan::FilterRealcugan(
+    int gpuid,
+    bool tta_mode,
+    int scaling_factor,
+    int noise_level,
+    int num_threads,
+    int syncgap,
+    const fsutils::StringType model_name
+)
+    : realcugan_(nullptr),
+      gpuid_(gpuid),
+      tta_mode_(tta_mode),
+      scaling_factor_(scaling_factor),
+      noise_level_(noise_level),
+      num_threads_(num_threads),
+      syncgap_(syncgap),
+      model_name_(std::move(model_name)) {}
+
+FilterRealcugan::~FilterRealcugan() {
+    if (realcugan_) {
+        delete realcugan_;
+        realcugan_ = nullptr;
+    }
+}
+
+int FilterRealcugan::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *) {
+    // Construct the model paths using std::filesystem
+    std::filesystem::path model_param_path;
+    std::filesystem::path model_bin_path;
+
+    fsutils::StringType model_base_name =
+        STR("up") + fsutils::to_string_type(scaling_factor_) + STR("x-");
+
+    switch (noise_level_) {
+        case -1:
+            model_base_name += STR("conservative");
+            break;
+        case 0:
+            model_base_name += STR("no-denoise");
+            break;
+        default:
+            model_base_name += STR("denoise") + fsutils::to_string_type(noise_level_) + STR("x");
+            break;
+    }
+
+    fsutils::StringType param_file_name = model_base_name + STR(".param");
+    fsutils::StringType bin_file_name = model_base_name + STR(".bin");
+
+    // Find the model paths by model name if provided
+    model_param_path =
+        std::filesystem::path(STR("models")) / STR("realcugan") / model_name_ / param_file_name;
+    model_bin_path =
+        std::filesystem::path(STR("models")) / STR("realcugan") / model_name_ / bin_file_name;
+
+    // Get the full paths using a function that possibly modifies or validates the path
+    std::filesystem::path model_param_full_path = fsutils::find_resource_file(model_param_path);
+    std::filesystem::path model_bin_full_path = fsutils::find_resource_file(model_bin_path);
+
+    // Check if the model files exist
+    if (!std::filesystem::exists(model_param_full_path)) {
+        logger()->error("RealCUGAN model param file not found: {}", model_param_path.u8string());
+        return -1;
+    }
+    if (!std::filesystem::exists(model_bin_full_path)) {
+        logger()->error("RealCUGAN model bin file not found: {}", model_bin_path.u8string());
+        return -1;
+    }
+
+    // Create a new RealCUGAN instance
+    realcugan_ = new RealCUGAN(gpuid_, tta_mode_, num_threads_);
+
+    // Store the time bases
+    in_time_base_ = dec_ctx->time_base;
+    out_time_base_ = enc_ctx->time_base;
+    out_pix_fmt_ = enc_ctx->pix_fmt;
+
+    // Load the model
+    if (realcugan_->load(model_param_full_path, model_bin_full_path) != 0) {
+        logger()->error("Failed to load RealCUGAN model");
+        return -1;
+    }
+
+    // Set syncgap to 0 for models-nose
+    if (model_name_.find(STR("models-nose")) != fsutils::StringType::npos) {
+        syncgap_ = 0;
+    }
+
+    // Set realcugan parameters
+    realcugan_->scale = scaling_factor_;
+    realcugan_->noise = noise_level_;
+    realcugan_->prepadding = 10;
+
+    // Set prepadding based on scaling factor
+    if (scaling_factor_ == 2) {
+        realcugan_->prepadding = 18;
+    }
+    if (scaling_factor_ == 3) {
+        realcugan_->prepadding = 14;
+    }
+    if (scaling_factor_ == 4) {
+        realcugan_->prepadding = 19;
+    }
+
+    // Calculate tilesize based on GPU heap budget
+    uint32_t heap_budget = ncnn::get_gpu_device(gpuid_)->get_heap_budget();
+    if (scaling_factor_ == 2) {
+        if (heap_budget > 1300) {
+            realcugan_->tilesize = 400;
+        } else if (heap_budget > 800) {
+            realcugan_->tilesize = 300;
+        } else if (heap_budget > 400) {
+            realcugan_->tilesize = 200;
+        } else if (heap_budget > 200) {
+            realcugan_->tilesize = 100;
+        } else {
+            realcugan_->tilesize = 32;
+        }
+    }
+    if (scaling_factor_ == 3) {
+        if (heap_budget > 3300) {
+            realcugan_->tilesize = 400;
+        } else if (heap_budget > 1900) {
+            realcugan_->tilesize = 300;
+        } else if (heap_budget > 950) {
+            realcugan_->tilesize = 200;
+        } else if (heap_budget > 320) {
+            realcugan_->tilesize = 100;
+        } else {
+            realcugan_->tilesize = 32;
+        }
+    }
+    if (scaling_factor_ == 4) {
+        if (heap_budget > 1690) {
+            realcugan_->tilesize = 400;
+        } else if (heap_budget > 980) {
+            realcugan_->tilesize = 300;
+        } else if (heap_budget > 530) {
+            realcugan_->tilesize = 200;
+        } else if (heap_budget > 240) {
+            realcugan_->tilesize = 100;
+        } else {
+            realcugan_->tilesize = 32;
+        }
+    }
+
+    return 0;
+}
+
+int FilterRealcugan::filter(AVFrame *in_frame, AVFrame **out_frame) {
+    int ret;
+
+    // Convert the input frame to RGB24
+    ncnn::Mat in_mat = conversions::avframe_to_ncnn_mat(in_frame);
+    if (in_mat.empty()) {
+        logger()->error("Failed to convert AVFrame to ncnn::Mat");
+        return -1;
+    }
+
+    // Allocate space for output ncnn::Mat
+    int output_width = in_mat.w * realcugan_->scale;
+    int output_height = in_mat.h * realcugan_->scale;
+    ncnn::Mat out_mat = ncnn::Mat(output_width, output_height, static_cast<size_t>(3), 3);
+
+    ret = realcugan_->process(in_mat, out_mat);
+    if (ret != 0) {
+        logger()->error("RealCUGAN processing failed");
+        return ret;
+    }
+
+    // Convert ncnn::Mat to AVFrame
+    *out_frame = conversions::ncnn_mat_to_avframe(out_mat, out_pix_fmt_);
+
+    // Rescale PTS to encoder's time base
+    (*out_frame)->pts = av_rescale_q(in_frame->pts, in_time_base_, out_time_base_);
+
+    // Return the processed frame to the caller
+    return ret;
+}
+
+void FilterRealcugan::get_output_dimensions(
+    const ProcessorConfig &,
+    int in_width,
+    int in_height,
+    int &out_width,
+    int &out_height
+) const {
+    out_width = in_width * scaling_factor_;
+    out_height = in_height * scaling_factor_;
+}
+
+}  // namespace processors
+}  // namespace video2x
--- a/src/filter_realesrgan.cpp
+++ b/src/filter_realesrgan.cpp
@@ -8,12 +8,16 @@

 #include "conversions.h"
 #include "fsutils.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {

 FilterRealesrgan::FilterRealesrgan(
    int gpuid,
    bool tta_mode,
    int scaling_factor,
-    const StringType model_name
+    const fsutils::StringType model_name
 )
    : realesrgan_(nullptr),
      gpuid_(gpuid),
@@ -28,31 +32,31 @@ FilterRealesrgan::~FilterRealesrgan() {
    }
 }

-int FilterRealesrgan::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *_) {
+int FilterRealesrgan::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVBufferRef *) {
    // Construct the model paths using std::filesystem
    std::filesystem::path model_param_path;
    std::filesystem::path model_bin_path;

-    StringType param_file_name =
-        model_name_ + STR("-x") + to_string_type(scaling_factor_) + STR(".param");
-    StringType bin_file_name =
-        model_name_ + STR("-x") + to_string_type(scaling_factor_) + STR(".bin");
+    fsutils::StringType param_file_name =
+        model_name_ + STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".param");
+    fsutils::StringType bin_file_name =
+        model_name_ + STR("-x") + fsutils::to_string_type(scaling_factor_) + STR(".bin");

    // Find the model paths by model name if provided
    model_param_path = std::filesystem::path(STR("models")) / STR("realesrgan") / param_file_name;
    model_bin_path = std::filesystem::path(STR("models")) / STR("realesrgan") / bin_file_name;

    // Get the full paths using a function that possibly modifies or validates the path
-    std::filesystem::path model_param_full_path = find_resource_file(model_param_path);
-    std::filesystem::path model_bin_full_path = find_resource_file(model_bin_path);
+    std::filesystem::path model_param_full_path = fsutils::find_resource_file(model_param_path);
+    std::filesystem::path model_bin_full_path = fsutils::find_resource_file(model_bin_path);

    // Check if the model files exist
    if (!std::filesystem::exists(model_param_full_path)) {
-        spdlog::error("RealESRGAN model param file not found: {}", model_param_path.u8string());
+        logger()->error("RealESRGAN model param file not found: {}", model_param_path.u8string());
        return -1;
    }
    if (!std::filesystem::exists(model_bin_full_path)) {
-        spdlog::error("RealESRGAN model bin file not found: {}", model_bin_path.u8string());
+        logger()->error("RealESRGAN model bin file not found: {}", model_bin_path.u8string());
        return -1;
    }

@@ -66,7 +70,7 @@ int FilterRealesrgan::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVB

    // Load the model
    if (realesrgan_->load(model_param_full_path, model_bin_full_path) != 0) {
-        spdlog::error("Failed to load RealESRGAN model");
+        logger()->error("Failed to load RealESRGAN model");
        return -1;
    }

@@ -93,9 +97,9 @@ int FilterRealesrgan::filter(AVFrame *in_frame, AVFrame **out_frame) {
    int ret;

    // Convert the input frame to RGB24
-    ncnn::Mat in_mat = avframe_to_ncnn_mat(in_frame);
+    ncnn::Mat in_mat = conversions::avframe_to_ncnn_mat(in_frame);
    if (in_mat.empty()) {
-        spdlog::error("Failed to convert AVFrame to ncnn::Mat");
+        logger()->error("Failed to convert AVFrame to ncnn::Mat");
        return -1;
    }

@@ -106,12 +110,12 @@ int FilterRealesrgan::filter(AVFrame *in_frame, AVFrame **out_frame) {

    ret = realesrgan_->process(in_mat, out_mat);
    if (ret != 0) {
-        spdlog::error("RealESRGAN processing failed");
+        logger()->error("RealESRGAN processing failed");
        return ret;
    }

    // Convert ncnn::Mat to AVFrame
-    *out_frame = ncnn_mat_to_avframe(out_mat, out_pix_fmt_);
+    *out_frame = conversions::ncnn_mat_to_avframe(out_mat, out_pix_fmt_);

    // Rescale PTS to encoder's time base
    (*out_frame)->pts = av_rescale_q(in_frame->pts, in_time_base_, out_time_base_);
@@ -130,3 +134,6 @@ void FilterRealesrgan::get_output_dimensions(
    out_width = in_width * scaling_factor_;
    out_height = in_height * scaling_factor_;
 }
+
+}  // namespace processors
+}  // namespace video2x
--- a/src/fsutils.cpp
+++ b/src/fsutils.cpp
@@ -10,6 +10,11 @@

 #include <spdlog/spdlog.h>

+#include "logger_manager.h"
+
+namespace video2x {
+namespace fsutils {
+
 #if _WIN32
 static std::filesystem::path get_executable_directory() {
    std::vector<wchar_t> filepath(MAX_PATH);
@@ -17,7 +22,7 @@ static std::filesystem::path get_executable_directory() {
    // Get the executable path, expanding the buffer if necessary
    DWORD size = GetModuleFileNameW(NULL, filepath.data(), static_cast<DWORD>(filepath.size()));
    if (size == 0) {
-        spdlog::error("Error getting executable path: {}", GetLastError());
+        logger()->error("Error getting executable path: {}", GetLastError());
        return std::filesystem::path();
    }

@@ -26,7 +31,7 @@ static std::filesystem::path get_executable_directory() {
        filepath.resize(filepath.size() * 2);
        size = GetModuleFileNameW(NULL, filepath.data(), static_cast<DWORD>(filepath.size()));
        if (size == 0) {
-            spdlog::error("Error getting executable path: {}", GetLastError());
+            logger()->error("Error getting executable path: {}", GetLastError());
            return std::filesystem::path();
        }
    }
@@ -41,7 +46,7 @@ static std::filesystem::path get_executable_directory() {
    std::filesystem::path filepath = std::filesystem::read_symlink("/proc/self/exe", ec);

    if (ec) {
-        spdlog::error("Error reading /proc/self/exe: {}", ec.message());
+        logger()->error("Error reading /proc/self/exe: {}", ec.message());
        return std::filesystem::path();
    }

@@ -120,7 +125,7 @@ std::string wstring_to_u8string(const std::string &str) {
 }
 #endif

-StringType path_to_string_type(const std::filesystem::path &path) {
+fsutils::StringType path_to_string_type(const std::filesystem::path &path) {
 #if _WIN32
    return path.wstring();
 #else
@@ -128,10 +133,13 @@ StringType path_to_string_type(const std::filesystem::path &path) {
 #endif
 }

-StringType to_string_type(int value) {
+fsutils::StringType to_string_type(int value) {
 #if _WIN32
    return std::to_wstring(value);
 #else
    return std::to_string(value);
 #endif
 }
+
+}  // namespace fsutils
+}  // namespace video2x
--- a/src/interpolator_rife.cpp
+++ b/src/interpolator_rife.cpp
@@ -7,6 +7,10 @@

 #include "conversions.h"
 #include "fsutils.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {

 InterpolatorRIFE::InterpolatorRIFE(
    int gpuid,
@@ -14,7 +18,7 @@ InterpolatorRIFE::InterpolatorRIFE(
    bool tta_temporal_mode,
    bool uhd_mode,
    int num_threads,
-    const StringType model_name
+    const fsutils::StringType model_name
 )
    : rife_(nullptr),
      gpuid_(gpuid),
@@ -39,25 +43,25 @@ int InterpolatorRIFE::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVB
    model_param_dir = std::filesystem::path(STR("models")) / STR("rife") / model_name_;

    // Get the full paths using a function that possibly modifies or validates the path
-    std::filesystem::path model_param_full_path = find_resource_file(model_param_dir);
+    std::filesystem::path model_param_full_path = fsutils::find_resource_file(model_param_dir);

    // Check if the model files exist
    if (!std::filesystem::exists(model_param_full_path)) {
-        spdlog::error("RIFE model param directory not found: {}", model_param_dir.u8string());
+        logger()->error("RIFE model param directory not found: {}", model_param_dir.u8string());
        return -1;
    }

    // Automatically infer the RIFE model generation based on the model name
    bool rife_v2 = false;
    bool rife_v4 = false;
-    if (model_name_.find(STR("rife-v2")) != StringType::npos) {
+    if (model_name_.find(STR("rife-v2")) != fsutils::StringType::npos) {
        rife_v2 = true;
-    } else if (model_name_.find(STR("rife-v3")) != StringType::npos) {
+    } else if (model_name_.find(STR("rife-v3")) != fsutils::StringType::npos) {
        rife_v2 = true;
-    } else if (model_name_.find(STR("rife-v4")) != StringType::npos) {
+    } else if (model_name_.find(STR("rife-v4")) != fsutils::StringType::npos) {
        rife_v4 = true;
-    } else if (model_name_.find(STR("rife")) == StringType::npos) {
-        spdlog::critical("Failed to infer RIFE model generation from model name");
+    } else if (model_name_.find(STR("rife")) == fsutils::StringType::npos) {
+        logger()->critical("Failed to infer RIFE model generation from model name");
        return -1;
    }

@@ -72,7 +76,7 @@ int InterpolatorRIFE::init(AVCodecContext *dec_ctx, AVCodecContext *enc_ctx, AVB

    // Load the model
    if (rife_->load(model_param_full_path) != 0) {
-        spdlog::error("Failed to load RIFE model");
+        logger()->error("Failed to load RIFE model");
        return -1;
    }

@@ -87,15 +91,15 @@ int InterpolatorRIFE::interpolate(
 ) {
    int ret;

-    ncnn::Mat in_mat1 = avframe_to_ncnn_mat(prev_frame);
+    ncnn::Mat in_mat1 = conversions::avframe_to_ncnn_mat(prev_frame);
    if (in_mat1.empty()) {
-        spdlog::error("Failed to convert AVFrame to ncnn::Mat");
+        logger()->error("Failed to convert AVFrame to ncnn::Mat");
        return -1;
    }

-    ncnn::Mat in_mat2 = avframe_to_ncnn_mat(in_frame);
+    ncnn::Mat in_mat2 = conversions::avframe_to_ncnn_mat(in_frame);
    if (in_mat2.empty()) {
-        spdlog::error("Failed to convert AVFrame to ncnn::Mat");
+        logger()->error("Failed to convert AVFrame to ncnn::Mat");
        return -1;
    }

@@ -104,12 +108,12 @@ int InterpolatorRIFE::interpolate(

    ret = rife_->process(in_mat1, in_mat2, time_step, out_mat);
    if (ret != 0) {
-        spdlog::error("RIFE processing failed");
+        logger()->error("RIFE processing failed");
        return ret;
    }

    // Convert ncnn::Mat to AVFrame
-    *out_frame = ncnn_mat_to_avframe(out_mat, out_pix_fmt_);
+    *out_frame = conversions::ncnn_mat_to_avframe(out_mat, out_pix_fmt_);

    // Rescale PTS to encoder's time base
    (*out_frame)->pts = av_rescale_q(in_frame->pts, in_time_base_, out_time_base_);
@@ -128,3 +132,6 @@ void InterpolatorRIFE::get_output_dimensions(
    out_width = in_width;
    out_height = in_height;
 }
+
+}  // namespace processors
+}  // namespace video2x
--- a/src/libplacebo.cpp
+++ b/src/libplacebo.cpp
@@ -11,6 +11,11 @@ extern "C" {

 #include <spdlog/spdlog.h>

+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {
+
 int init_libplacebo(
    AVFilterGraph **filter_graph,
    AVFilterContext **buffersrc_ctx,
@@ -29,20 +34,20 @@ int init_libplacebo(
        &vk_hw_device_ctx, AV_HWDEVICE_TYPE_VULKAN, std::to_string(vk_device_index).c_str(), NULL, 0
    );
    if (ret < 0) {
-        spdlog::error("Failed to create Vulkan hardware device context for libplacebo.");
+        logger()->error("Failed to create Vulkan hardware device context for libplacebo.");
        vk_hw_device_ctx = nullptr;
    }

    AVFilterGraph *graph = avfilter_graph_alloc();
    if (!graph) {
-        spdlog::error("Unable to create filter graph.");
+        logger()->error("Unable to create filter graph.");
        return AVERROR(ENOMEM);
    }

    // Create buffer source
    const AVFilter *buffersrc = avfilter_get_by_name("buffer");
    if (!buffersrc) {
-        spdlog::error("Filter 'buffer' not found.");
+        logger()->error("Filter 'buffer' not found.");
        avfilter_graph_free(&graph);
        return AVERROR_FILTER_NOT_FOUND;
    }
@@ -79,7 +84,7 @@ int init_libplacebo(
    spdlog::debug("Buffer source args: {}", args);
    ret = avfilter_graph_create_filter(buffersrc_ctx, buffersrc, "in", args.c_str(), NULL, graph);
    if (ret < 0) {
-        spdlog::error("Cannot create buffer source.");
+        logger()->error("Cannot create buffer source.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -89,7 +94,7 @@ int init_libplacebo(
    // Create the libplacebo filter
    const AVFilter *libplacebo_filter = avfilter_get_by_name("libplacebo");
    if (!libplacebo_filter) {
-        spdlog::error("Filter 'libplacebo' not found.");
+        logger()->error("Filter 'libplacebo' not found.");
        avfilter_graph_free(&graph);
        return AVERROR_FILTER_NOT_FOUND;
    }
@@ -112,7 +117,7 @@ int init_libplacebo(
        &libplacebo_ctx, libplacebo_filter, "libplacebo", filter_args.c_str(), NULL, graph
    );
    if (ret < 0) {
-        spdlog::error("Cannot create libplacebo filter.");
+        logger()->error("Cannot create libplacebo filter.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -126,7 +131,7 @@ int init_libplacebo(
    // Link buffersrc to libplacebo
    ret = avfilter_link(last_filter, 0, libplacebo_ctx, 0);
    if (ret < 0) {
-        spdlog::error("Error connecting buffersrc to libplacebo filter.");
+        logger()->error("Error connecting buffersrc to libplacebo filter.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -137,7 +142,7 @@ int init_libplacebo(
    const AVFilter *buffersink = avfilter_get_by_name("buffersink");
    ret = avfilter_graph_create_filter(buffersink_ctx, buffersink, "out", NULL, NULL, graph);
    if (ret < 0) {
-        spdlog::error("Cannot create buffer sink.");
+        logger()->error("Cannot create buffer sink.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -145,7 +150,7 @@ int init_libplacebo(
    // Link libplacebo to buffersink
    ret = avfilter_link(last_filter, 0, *buffersink_ctx, 0);
    if (ret < 0) {
-        spdlog::error("Error connecting libplacebo filter to buffersink.");
+        logger()->error("Error connecting libplacebo filter to buffersink.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -153,7 +158,7 @@ int init_libplacebo(
    // Configure the filter graph
    ret = avfilter_graph_config(graph, NULL);
    if (ret < 0) {
-        spdlog::error("Error configuring the filter graph.");
+        logger()->error("Error configuring the filter graph.");
        avfilter_graph_free(&graph);
        return ret;
    }
@@ -161,3 +166,6 @@ int init_libplacebo(
    *filter_graph = graph;
    return 0;
 }
+
+}  // namespace processors
+}  // namespace video2x
--- a/src/libvideo2x.cpp
+++ b/src/libvideo2x.cpp
@@ -9,25 +9,24 @@ extern "C" {
 #include "avutils.h"
 #include "decoder.h"
 #include "encoder.h"
-#include "logging.h"
+#include "logger_manager.h"
 #include "processor.h"
 #include "processor_factory.h"

+namespace video2x {
+
 VideoProcessor::VideoProcessor(
-    const ProcessorConfig proc_cfg,
-    const EncoderConfig enc_cfg,
-    const uint32_t vk_device_index,
+    const processors::ProcessorConfig proc_cfg,
+    const encoder::EncoderConfig enc_cfg,
+    const uint32_t vk_device_idx,
    const AVHWDeviceType hw_device_type,
-    const Video2xLogLevel log_level,
    const bool benchmark
 )
    : proc_cfg_(proc_cfg),
      enc_cfg_(enc_cfg),
-      vk_device_index_(vk_device_index),
+      vk_device_idx_(vk_device_idx),
      hw_device_type_(hw_device_type),
-      benchmark_(benchmark) {
-    set_log_level(log_level);
-}
+      benchmark_(benchmark) {}

 int VideoProcessor::process(
    const std::filesystem::path in_fname,
@@ -40,7 +39,7 @@ int VideoProcessor::process(
        // Format and log the error message
        char errbuf[AV_ERROR_MAX_STRING_SIZE];
        av_strerror(error_code, errbuf, sizeof(errbuf));
-        spdlog::critical("{}: {}", msg, errbuf);
+        logger()->critical("{}: {}", msg, errbuf);

        // Set the video processor state to failed and return the error code
        state_.store(VideoProcessorState::Failed);
@@ -51,8 +50,8 @@ int VideoProcessor::process(
    state_.store(VideoProcessorState::Running);

    // Create a smart pointer to manage the hardware device context
-    std::unique_ptr<AVBufferRef, decltype(&av_bufferref_deleter)> hw_ctx(
-        nullptr, &av_bufferref_deleter
+    std::unique_ptr<AVBufferRef, decltype(&avutils::av_bufferref_deleter)> hw_ctx(
+        nullptr, &avutils::av_bufferref_deleter
    );

    // Initialize hardware device context
@@ -66,7 +65,7 @@ int VideoProcessor::process(
    }

    // Initialize input decoder
-    Decoder decoder;
+    decoder::Decoder decoder;
    ret = decoder.init(hw_device_type_, hw_ctx.get(), in_fname);
    if (ret < 0) {
        return handle_error(ret, "Failed to initialize decoder");
@@ -77,8 +76,8 @@ int VideoProcessor::process(
    int in_vstream_idx = decoder.get_video_stream_index();

    // Create and initialize the appropriate filter
-    std::unique_ptr<Processor> processor(
-        ProcessorFactory::instance().create_processor(proc_cfg_, vk_device_index_)
+    std::unique_ptr<processors::Processor> processor(
+        processors::ProcessorFactory::instance().create_processor(proc_cfg_, vk_device_idx_)
    );
    if (processor == nullptr) {
        return handle_error(-1, "Failed to create filter instance");
@@ -94,7 +93,7 @@ int VideoProcessor::process(
    }

    // Initialize the encoder
-    Encoder encoder;
+    encoder::Encoder encoder;
    ret = encoder.init(
        hw_ctx.get(),
        out_fname,
@@ -140,9 +139,9 @@ int VideoProcessor::process(

 // Process frames using the selected filter.
 int VideoProcessor::process_frames(
-    Decoder &decoder,
-    Encoder &encoder,
-    std::unique_ptr<Processor> &processor
+    decoder::Decoder &decoder,
+    encoder::Encoder &encoder,
+    std::unique_ptr<processors::Processor> &processor
 ) {
    char errbuf[AV_ERROR_MAX_STRING_SIZE];
    int ret = 0;
@@ -156,29 +155,31 @@ int VideoProcessor::process_frames(

    // Reference to the previous frame does not require allocation
    // It will be cloned from the current frame
-    std::unique_ptr<AVFrame, decltype(&av_frame_deleter)> prev_frame(nullptr, &av_frame_deleter);
+    std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)> prev_frame(
+        nullptr, &avutils::av_frame_deleter
+    );

    // Allocate space for the decoded frames
-    std::unique_ptr<AVFrame, decltype(&av_frame_deleter)> frame(
-        av_frame_alloc(), &av_frame_deleter
+    std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)> frame(
+        av_frame_alloc(), &avutils::av_frame_deleter
    );
    if (frame == nullptr) {
-        spdlog::critical("Error allocating frame");
+        logger()->critical("Error allocating frame");
        return AVERROR(ENOMEM);
    }

    // Allocate space for the decoded packets
-    std::unique_ptr<AVPacket, decltype(&av_packet_deleter)> packet(
-        av_packet_alloc(), &av_packet_deleter
+    std::unique_ptr<AVPacket, decltype(&avutils::av_packet_deleter)> packet(
+        av_packet_alloc(), &avutils::av_packet_deleter
    );
    if (packet == nullptr) {
-        spdlog::critical("Error allocating packet");
+        logger()->critical("Error allocating packet");
        return AVERROR(ENOMEM);
    }

    // Set the total number of frames in the VideoProcessingContext
    spdlog::debug("Estimating the total number of frames to process");
-    total_frames_ = get_video_frame_count(ifmt_ctx, in_vstream_idx);
+    total_frames_ = avutils::get_video_frame_count(ifmt_ctx, in_vstream_idx);

    if (total_frames_ <= 0) {
        spdlog::warn("Unable to determine the total number of frames");
@@ -188,7 +189,7 @@ int VideoProcessor::process_frames(
    }

    // Set total frames for interpolation
-    if (processor->get_processing_mode() == ProcessingMode::Interpolate) {
+    if (processor->get_processing_mode() == processors::ProcessingMode::Interpolate) {
        total_frames_.store(total_frames_.load() * proc_cfg_.frm_rate_mul);
    }

@@ -201,7 +202,7 @@ int VideoProcessor::process_frames(
                break;
            }
            av_strerror(ret, errbuf, sizeof(errbuf));
-            spdlog::critical("Error reading packet: {}", errbuf);
+            logger()->critical("Error reading packet: {}", errbuf);
            return ret;
        }

@@ -210,7 +211,7 @@ int VideoProcessor::process_frames(
            ret = avcodec_send_packet(dec_ctx, packet.get());
            if (ret < 0) {
                av_strerror(ret, errbuf, sizeof(errbuf));
-                spdlog::critical("Error sending packet to decoder: {}", errbuf);
+                logger()->critical("Error sending packet to decoder: {}", errbuf);
                return ret;
            }

@@ -229,33 +230,33 @@ int VideoProcessor::process_frames(
                    break;
                } else if (ret < 0) {
                    av_strerror(ret, errbuf, sizeof(errbuf));
-                    spdlog::critical("Error decoding video frame: {}", errbuf);
+                    logger()->critical("Error decoding video frame: {}", errbuf);
                    return ret;
                }

                // Process the frame based on the selected processing mode
                AVFrame *proc_frame = nullptr;
                switch (processor->get_processing_mode()) {
-                    case ProcessingMode::Filter: {
+                    case processors::ProcessingMode::Filter: {
                        ret = process_filtering(processor, encoder, frame.get(), proc_frame);
                        break;
                    }
-                    case ProcessingMode::Interpolate: {
+                    case processors::ProcessingMode::Interpolate: {
                        ret = process_interpolation(
                            processor, encoder, prev_frame, frame.get(), proc_frame
                        );
                        break;
                    }
                    default:
-                        spdlog::critical("Unknown processing mode");
+                        logger()->critical("Unknown processing mode");
                        return -1;
                }
                if (ret < 0 && ret != AVERROR(EAGAIN)) {
                    return ret;
                }
                av_frame_unref(frame.get());
-                frame_index_++;
-                spdlog::debug("Processed frame {}/{}", frame_index_.load(), total_frames_.load());
+                frame_idx_++;
+                spdlog::debug("Processed frame {}/{}", frame_idx_.load(), total_frames_.load());
            }
        } else if (enc_cfg_.copy_streams && stream_map[packet->stream_index] >= 0) {
            ret = write_raw_packet(packet.get(), ifmt_ctx, ofmt_ctx, stream_map);
@@ -271,14 +272,14 @@ int VideoProcessor::process_frames(
    ret = processor->flush(raw_flushed_frames);
    if (ret < 0) {
        av_strerror(ret, errbuf, sizeof(errbuf));
-        spdlog::critical("Error flushing filter: {}", errbuf);
+        logger()->critical("Error flushing filter: {}", errbuf);
        return ret;
    }

    // Wrap flushed frames in unique_ptrs
-    std::vector<std::unique_ptr<AVFrame, decltype(&av_frame_deleter)>> flushed_frames;
+    std::vector<std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)>> flushed_frames;
    for (AVFrame *raw_frame : raw_flushed_frames) {
-        flushed_frames.emplace_back(raw_frame, &av_frame_deleter);
+        flushed_frames.emplace_back(raw_frame, &avutils::av_frame_deleter);
    }

    // Encode and write all flushed frames
@@ -287,31 +288,29 @@ int VideoProcessor::process_frames(
        if (ret < 0) {
            return ret;
        }
-        frame_index_++;
+        frame_idx_++;
    }

    // Flush the encoder
    ret = encoder.flush();
    if (ret < 0) {
        av_strerror(ret, errbuf, sizeof(errbuf));
-        spdlog::critical("Error flushing encoder: {}", errbuf);
+        logger()->critical("Error flushing encoder: {}", errbuf);
        return ret;
    }

    return ret;
 }

-int VideoProcessor::write_frame(AVFrame *frame, Encoder &encoder) {
+int VideoProcessor::write_frame(AVFrame *frame, encoder::Encoder &encoder) {
    char errbuf[AV_ERROR_MAX_STRING_SIZE];
    int ret = 0;

    if (!benchmark_) {
-        // Set the frame type to none to let the encoder decide
-        frame->pict_type = AV_PICTURE_TYPE_NONE;
-        ret = encoder.write_frame(frame, frame_index_);
+        ret = encoder.write_frame(frame, frame_idx_);
        if (ret < 0) {
            av_strerror(ret, errbuf, sizeof(errbuf));
-            spdlog::critical("Error encoding/writing frame: {}", errbuf);
+            logger()->critical("Error encoding/writing frame: {}", errbuf);
        }
    }
    return ret;
@@ -327,23 +326,23 @@ int VideoProcessor::write_raw_packet(
    int ret = 0;

    AVStream *in_stream = ifmt_ctx->streams[packet->stream_index];
-    int out_stream_index = stream_map[packet->stream_index];
-    AVStream *out_stream = ofmt_ctx->streams[out_stream_index];
+    int out_stream_idx = stream_map[packet->stream_index];
+    AVStream *out_stream = ofmt_ctx->streams[out_stream_idx];

    av_packet_rescale_ts(packet, in_stream->time_base, out_stream->time_base);
-    packet->stream_index = out_stream_index;
+    packet->stream_index = out_stream_idx;

    ret = av_interleaved_write_frame(ofmt_ctx, packet);
    if (ret < 0) {
        av_strerror(ret, errbuf, sizeof(errbuf));
-        spdlog::critical("Error muxing audio/subtitle packet: {}", errbuf);
+        logger()->critical("Error muxing audio/subtitle packet: {}", errbuf);
    }
    return ret;
 }

 int VideoProcessor::process_filtering(
-    std::unique_ptr<Processor> &processor,
-    Encoder &encoder,
+    std::unique_ptr<processors::Processor> &processor,
+    encoder::Encoder &encoder,
    AVFrame *frame,
    AVFrame *proc_frame
 ) {
@@ -351,7 +350,7 @@ int VideoProcessor::process_filtering(
    int ret = 0;

    // Cast the processor to a Filter
-    Filter *filter = static_cast<Filter *>(processor.get());
+    processors::Filter *filter = static_cast<processors::Filter *>(processor.get());

    // Process the frame using the filter
    ret = filter->filter(frame, &proc_frame);
@@ -359,19 +358,20 @@ int VideoProcessor::process_filtering(
    // Write the processed frame
    if (ret < 0 && ret != AVERROR(EAGAIN)) {
        av_strerror(ret, errbuf, sizeof(errbuf));
-        spdlog::critical("Error filtering frame: {}", errbuf);
+        logger()->critical("Error filtering frame: {}", errbuf);
    } else if (ret == 0 && proc_frame != nullptr) {
-        auto processed_frame =
-            std::unique_ptr<AVFrame, decltype(&av_frame_deleter)>(proc_frame, &av_frame_deleter);
+        auto processed_frame = std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)>(
+            proc_frame, &avutils::av_frame_deleter
+        );
        ret = write_frame(processed_frame.get(), encoder);
    }
    return ret;
 }

 int VideoProcessor::process_interpolation(
-    std::unique_ptr<Processor> &processor,
-    Encoder &encoder,
-    std::unique_ptr<AVFrame, decltype(&av_frame_deleter)> &prev_frame,
+    std::unique_ptr<processors::Processor> &processor,
+    encoder::Encoder &encoder,
+    std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)> &prev_frame,
    AVFrame *frame,
    AVFrame *proc_frame
 ) {
@@ -379,7 +379,8 @@ int VideoProcessor::process_interpolation(
    int ret = 0;

    // Cast the processor to an Interpolator
-    Interpolator *interpolator = static_cast<Interpolator *>(processor.get());
+    processors::Interpolator *interpolator =
+        static_cast<processors::Interpolator *>(processor.get());

    // Calculate the time step for each frame
    float time_step = 1.0f / static_cast<float>(proc_cfg_.frm_rate_mul);
@@ -388,12 +389,10 @@ int VideoProcessor::process_interpolation(
    // Check if a scene change is detected
    bool skip_frame = false;
    if (proc_cfg_.scn_det_thresh < 100.0 && prev_frame.get() != nullptr) {
-        float frame_diff = get_frame_diff(prev_frame.get(), frame);
+        float frame_diff = avutils::get_frame_diff(prev_frame.get(), frame);
        if (frame_diff > proc_cfg_.scn_det_thresh) {
            spdlog::debug(
-                "Scene change detected ({:.2f}%), skipping frame {}",
-                frame_diff,
-                frame_index_.load()
+                "Scene change detected ({:.2f}%), skipping frame {}", frame_diff, frame_idx_.load()
            );
            skip_frame = true;
        }
@@ -418,29 +417,29 @@ int VideoProcessor::process_interpolation(
        // Write the interpolated frame
        if (ret < 0 && ret != AVERROR(EAGAIN)) {
            av_strerror(ret, errbuf, sizeof(errbuf));
-            spdlog::critical("Error interpolating frame: {}", errbuf);
+            logger()->critical("Error interpolating frame: {}", errbuf);
            return ret;
        } else if (ret == 0 && proc_frame != nullptr) {
-            auto processed_frame = std::unique_ptr<AVFrame, decltype(&av_frame_deleter)>(
-                proc_frame, &av_frame_deleter
+            auto processed_frame = std::unique_ptr<AVFrame, decltype(&avutils::av_frame_deleter)>(
+                proc_frame, &avutils::av_frame_deleter
            );

-            processed_frame->pts = frame_index_;
            ret = write_frame(processed_frame.get(), encoder);
            if (ret < 0) {
                return ret;
            }
        }

-        frame_index_++;
+        frame_idx_++;
        current_time_step += time_step;
    }

    // Write the original frame
-    frame->pts = frame_index_;
    ret = write_frame(frame, encoder);

    // Update the previous frame with the current frame
    prev_frame.reset(av_frame_clone(frame));
    return ret;
 }
+
+}  // namespace video2x
--- a/src/logger_manager.cpp
+++ b/src/logger_manager.cpp
@@ -0,0 +1,116 @@
+#include "logger_manager.h"
+
+extern "C" {
+#include <libavutil/log.h>
+}
+
+#include <spdlog/sinks/stdout_color_sinks.h>
+#include <spdlog/spdlog.h>
+
+static spdlog::level::level_enum ffmpeg_level_to_spdlog(int av_level) {
+    if (av_level <= AV_LOG_PANIC) {
+        return spdlog::level::critical;
+    } else if (av_level <= AV_LOG_ERROR) {
+        return spdlog::level::err;
+    } else if (av_level <= AV_LOG_WARNING) {
+        return spdlog::level::warn;
+    } else if (av_level <= AV_LOG_INFO) {
+        return spdlog::level::info;
+    } else if (av_level <= AV_LOG_VERBOSE) {
+        return spdlog::level::debug;
+    } else if (av_level == AV_LOG_DEBUG) {
+        return spdlog::level::debug;
+    } else {
+        // AV_LOG_TRACE or beyond (if supported by FFmpeg)
+        return spdlog::level::trace;
+    }
+}
+
+static void ffmpeg_log_callback(void *avcl, int level, const char *fmt, va_list vargs) {
+    // Format the message the same way as the default callback
+    char line[1024];
+    int print_prefix = 1;
+    av_log_format_line(avcl, level, fmt, vargs, line, sizeof(line), &print_prefix);
+
+    // Trim trailing newlines
+    std::string message = line;
+    while (!message.empty() && (message.back() == '\n' || message.back() == '\r')) {
+        message.pop_back();
+    }
+
+    // Forward the formatted FFmpeg log message to the logger
+    video2x::logger()->log(ffmpeg_level_to_spdlog(level), "[FFmpeg] {}", message);
+}
+
+namespace video2x {
+namespace logger_manager {
+
+LoggerManager::LoggerManager() {
+    auto console_sink = std::make_shared<spdlog::sinks::stdout_color_sink_mt>();
+    console_sink->set_pattern("%+");
+    console_sink->set_level(spdlog::level::info);
+    logger_ = std::make_shared<spdlog::logger>("video2x", console_sink);
+    spdlog::register_logger(logger_);
+}
+
+LoggerManager &LoggerManager::instance() {
+    static LoggerManager instance;
+    return instance;
+}
+
+std::shared_ptr<spdlog::logger> LoggerManager::logger() {
+    return logger_;
+}
+
+bool LoggerManager::reconfigure_logger(
+    const std::string &logger_name,
+    const std::vector<spdlog::sink_ptr> &sinks,
+    const std::string &pattern
+) {
+    if (logger_name.empty() || sinks.empty()) {
+        return false;
+    }
+
+    // Create a new logger with the given name, sinks, and pattern
+    std::shared_ptr<spdlog::logger> new_logger =
+        std::make_shared<spdlog::logger>(logger_name, sinks.begin(), sinks.end());
+    new_logger->set_pattern(pattern);
+
+    // Maintain the log level from the previous logger
+    if (logger_ != nullptr) {
+        new_logger->set_level(logger_->level());
+    }
+
+    // If a logger with the same name exists, remove it first
+    std::shared_ptr<spdlog::logger> old_logger = spdlog::get(logger_name);
+    if (old_logger != nullptr) {
+        spdlog::drop(logger_name);
+        logger_ = nullptr;
+    }
+
+    // Replace the internal logger_ member and register the new one
+    logger_ = new_logger;
+    spdlog::register_logger(logger_);
+    return true;
+}
+
+bool LoggerManager::set_log_level(const std::string &level_str) {
+    spdlog::level::level_enum log_level = spdlog::level::from_str(level_str);
+    if (log_level == spdlog::level::off && level_str != "off") {
+        // Invalid level_str
+        return false;
+    }
+    logger_->set_level(log_level);
+    return true;
+}
+
+void LoggerManager::hook_ffmpeg_logging() {
+    av_log_set_callback(ffmpeg_log_callback);
+}
+
+void LoggerManager::unhook_ffmpeg_logging() {
+    av_log_set_callback(nullptr);
+}
+
+}  // namespace logger_manager
+}  // namespace video2x
--- a/src/logging.cpp
+++ b/src/logging.cpp
@@ -1,44 +0,0 @@
-#include "logging.h"
-
-extern "C" {
-#include <libavutil/avutil.h>
-}
-
-#include <spdlog/spdlog.h>
-
-void set_log_level(Video2xLogLevel log_level) {
-    switch (log_level) {
-        case Video2xLogLevel::Trace:
-            av_log_set_level(AV_LOG_TRACE);
-            spdlog::set_level(spdlog::level::trace);
-            break;
-        case Video2xLogLevel::Debug:
-            av_log_set_level(AV_LOG_DEBUG);
-            spdlog::set_level(spdlog::level::debug);
-            break;
-        case Video2xLogLevel::Info:
-            av_log_set_level(AV_LOG_INFO);
-            spdlog::set_level(spdlog::level::info);
-            break;
-        case Video2xLogLevel::Warning:
-            av_log_set_level(AV_LOG_WARNING);
-            spdlog::set_level(spdlog::level::warn);
-            break;
-        case Video2xLogLevel::Error:
-            av_log_set_level(AV_LOG_ERROR);
-            spdlog::set_level(spdlog::level::err);
-            break;
-        case Video2xLogLevel::Critical:
-            av_log_set_level(AV_LOG_FATAL);
-            spdlog::set_level(spdlog::level::critical);
-            break;
-        case Video2xLogLevel::Off:
-            av_log_set_level(AV_LOG_QUIET);
-            spdlog::set_level(spdlog::level::off);
-            break;
-        default:
-            av_log_set_level(AV_LOG_INFO);
-            spdlog::set_level(spdlog::level::info);
-            break;
-    }
-}
--- a/src/processor_factory.cpp
+++ b/src/processor_factory.cpp
@@ -4,8 +4,13 @@
 #include <utility>

 #include "filter_libplacebo.h"
+#include "filter_realcugan.h"
 #include "filter_realesrgan.h"
 #include "interpolator_rife.h"
+#include "logger_manager.h"
+
+namespace video2x {
+namespace processors {

 // Access the singleton instance
 ProcessorFactory &ProcessorFactory::instance() {
@@ -33,7 +38,7 @@ std::unique_ptr<Processor> ProcessorFactory::create_processor(
 ) const {
    auto it = creators.find(proc_cfg.processor_type);
    if (it == creators.end()) {
-        spdlog::critical(
+        logger()->critical(
            "Processor type not registered: {}", static_cast<int>(proc_cfg.processor_type)
        );
        return nullptr;
@@ -51,11 +56,11 @@ void ProcessorFactory::init_default_processors(ProcessorFactory &factory) {
           uint32_t vk_device_index) -> std::unique_ptr<Processor> {
            const auto &config = std::get<LibplaceboConfig>(proc_cfg.config);
            if (config.shader_path.empty()) {
-                spdlog::critical("Shader path must be provided for the libplacebo filter");
+                logger()->critical("Shader path must be provided for the libplacebo filter");
                return nullptr;
            }
            if (proc_cfg.width <= 0 || proc_cfg.height <= 0) {
-                spdlog::critical(
+                logger()->critical(
                    "Output width and height must be provided for the libplacebo filter"
                );
                return nullptr;
@@ -75,11 +80,11 @@ void ProcessorFactory::init_default_processors(ProcessorFactory &factory) {
           uint32_t vk_device_index) -> std::unique_ptr<Processor> {
            const auto &config = std::get<RealESRGANConfig>(proc_cfg.config);
            if (proc_cfg.scaling_factor <= 0) {
-                spdlog::critical("Scaling factor must be provided for the RealESRGAN filter");
+                logger()->critical("Scaling factor must be provided for the RealESRGAN filter");
                return nullptr;
            }
            if (config.model_name.empty()) {
-                spdlog::critical("Model name must be provided for the RealESRGAN filter");
+                logger()->critical("Model name must be provided for the RealESRGAN filter");
                return nullptr;
            }
            return std::make_unique<FilterRealesrgan>(
@@ -91,13 +96,38 @@ void ProcessorFactory::init_default_processors(ProcessorFactory &factory) {
        }
    );

+    factory.register_processor(
+        ProcessorType::RealCUGAN,
+        [](const ProcessorConfig &proc_cfg,
+           uint32_t vk_device_index) -> std::unique_ptr<Processor> {
+            const auto &config = std::get<RealCUGANConfig>(proc_cfg.config);
+            if (proc_cfg.scaling_factor <= 0) {
+                logger()->critical("Scaling factor must be provided for the RealCUGAN filter");
+                return nullptr;
+            }
+            if (config.model_name.empty()) {
+                logger()->critical("Model name must be provided for the RealCUGAN filter");
+                return nullptr;
+            }
+            return std::make_unique<FilterRealcugan>(
+                static_cast<int>(vk_device_index),
+                config.tta_mode,
+                proc_cfg.scaling_factor,
+                proc_cfg.noise_level,
+                config.num_threads,
+                config.syncgap,
+                config.model_name
+            );
+        }
+    );
+
    factory.register_processor(
        ProcessorType::RIFE,
        [](const ProcessorConfig &proc_cfg,
           uint32_t vk_device_index) -> std::unique_ptr<Processor> {
            const auto &cfg = std::get<RIFEConfig>(proc_cfg.config);
            if (cfg.model_name.empty()) {
-                spdlog::critical("Model name must be provided for the RIFE filter");
+                logger()->critical("Model name must be provided for the RIFE filter");
                return nullptr;
            }
            return std::make_unique<InterpolatorRIFE>(
@@ -111,3 +141,6 @@ void ProcessorFactory::init_default_processors(ProcessorFactory &factory) {
        }
    );
 }
+
+}  // namespace processors
+}  // namespace video2x
--- a/third_party/librealcugan_ncnn_vulkan
+++ b/third_party/librealcugan_ncnn_vulkan
--- a/third_party/librealesrgan_ncnn_vulkan
+++ b/third_party/librealesrgan_ncnn_vulkan
--- a/third_party/librife_ncnn_vulkan
+++ b/third_party/librife_ncnn_vulkan
--- a/third_party/ncnn
+++ b/third_party/ncnn
--- a/third_party/spdlog
+++ b/third_party/spdlog
--- a/tools/video2x/include/argparse.h
+++ b/tools/video2x/include/argparse.h
@@ -5,10 +5,7 @@

 // Structure to hold parsed arguments
 struct Arguments {
-    Video2xLogLevel log_level = Video2xLogLevel::Info;
    bool no_progress = false;
-
-    // General options
    std::filesystem::path in_fname;
    std::filesystem::path out_fname;
    uint32_t vk_device_index = 0;
@@ -24,6 +21,6 @@ struct Arguments {
    char *argv[],
 #endif
    Arguments &arguments,
-    ProcessorConfig &proc_cfg,
-    EncoderConfig &enc_cfg
+    video2x::processors::ProcessorConfig &proc_cfg,
+    video2x::encoder::EncoderConfig &enc_cfg
 );
--- a/tools/video2x/include/logging.h
+++ b/tools/video2x/include/logging.h
@@ -1,15 +0,0 @@
-#pragma once
-
-#include <atomic>
-#include <optional>
-
-#include <libvideo2x/libvideo2x.h>
-#include <spdlog/spdlog.h>
-
-extern std::atomic<bool> newline_required;
-
-void set_spdlog_level(Video2xLogLevel log_level);
-
-std::optional<Video2xLogLevel> find_log_level_by_name(const StringType &log_level_name);
-
-void newline_safe_ffmpeg_log_callback(void *ptr, int level, const char *fmt, va_list vl);
--- a/tools/video2x/include/newline_safe_sink.h
+++ b/tools/video2x/include/newline_safe_sink.h
@@ -0,0 +1,22 @@
+#pragma once
+
+#include <atomic>
+
+#include <spdlog/sinks/ansicolor_sink.h>
+
+class newline_safe_sink : public spdlog::sinks::ansicolor_stdout_sink_mt {
+   public:
+    newline_safe_sink() = default;
+    ~newline_safe_sink() = default;
+
+    newline_safe_sink(const newline_safe_sink &) = delete;
+    newline_safe_sink &operator=(const newline_safe_sink &) = delete;
+
+    void log(const spdlog::details::log_msg &msg);
+
+    void set_needs_newline(bool needs_newline) { needs_newline_.store(needs_newline); };
+    bool get_needs_newline() { return needs_newline_.load(); };
+
+   private:
+    std::atomic<bool> needs_newline_ = false;
+};
--- a/tools/video2x/include/validators.h
+++ b/tools/video2x/include/validators.h
@@ -61,8 +61,10 @@ void validate_greater_equal_one(const T &value, const std::string &option_name)
    }
 }

-void validate_anime4k_shader_name(const StringType &shader_name);
+void validate_anime4k_shader_name(const video2x::fsutils::StringType &shader_name);

-void validate_realesrgan_model_name(const StringType &model_name);
+void validate_realesrgan_model_name(const video2x::fsutils::StringType &model_name);

-void validate_rife_model_name(const StringType &model_name);
+void validate_realcugan_model_name(const video2x::fsutils::StringType &model_name);
+
+void validate_rife_model_name(const video2x::fsutils::StringType &model_name);
--- a/tools/video2x/src/argparse.cpp
+++ b/tools/video2x/src/argparse.cpp
@@ -7,13 +7,13 @@
 #include <cwchar>
 #endif

+#include <libvideo2x/logger_manager.h>
 #include <libvideo2x/version.h>
 #include <spdlog/spdlog.h>
-#include <vulkan_utils.h>
 #include <boost/program_options.hpp>

-#include "logging.h"
 #include "validators.h"
+#include "vulkan_utils.h"

 #ifdef _WIN32
 #define BOOST_PROGRAM_OPTIONS_WCHAR_T
@@ -59,8 +59,8 @@ int parse_args(
    char *argv[],
 #endif
    Arguments &arguments,
-    ProcessorConfig &proc_cfg,
-    EncoderConfig &enc_cfg
+    video2x::processors::ProcessorConfig &proc_cfg,
+    video2x::encoder::EncoderConfig &enc_cfg
 ) {
    try {
        // clang-format off
@@ -68,19 +68,20 @@ int parse_args(
        all_opts.add_options()
            ("help", "Display this help page")
            ("version,V", "Print program version and exit")
-            ("log-level", PO_STR_VALUE<StringType>()->default_value(STR("info"), "info"),
+            ("log-level", PO_STR_VALUE<video2x::fsutils::StringType>()
+                ->default_value(STR("info"), "info"),
                "Set verbosity level (trace, debug, info, warn, error, critical, none)")
            ("no-progress", po::bool_switch(&arguments.no_progress),
                "Do not display the progress bar")
            ("list-devices,l", "List the available Vulkan devices (GPUs)")

            // General Processing Options
-            ("input,i", PO_STR_VALUE<StringType>(), "Input video file path")
-            ("output,o", PO_STR_VALUE<StringType>(), "Output video file path")
-            ("processor,p", PO_STR_VALUE<StringType>(),
-                "Processor to use (libplacebo, realesrgan, rife)")
-            ("hwaccel,a", PO_STR_VALUE<StringType>()->default_value(STR("none"), "none"),
-                "Hardware acceleration method (decoding)")
+            ("input,i", PO_STR_VALUE<video2x::fsutils::StringType>(), "Input video file path")
+            ("output,o", PO_STR_VALUE<video2x::fsutils::StringType>(), "Output video file path")
+            ("processor,p", PO_STR_VALUE<video2x::fsutils::StringType>(),
+                "Processor to use (libplacebo, realesrgan, realcugan, rife)")
+            ("hwaccel,a", PO_STR_VALUE<video2x::fsutils::StringType>()
+                ->default_value(STR("none"), "none"), "Hardware acceleration method (decoding)")
            ("device,d", po::value<uint32_t>(&arguments.vk_device_index)->default_value(0),
                "Vulkan device index (GPU ID)")
            ("benchmark,b", po::bool_switch(&arguments.benchmark),
@@ -90,10 +91,10 @@ int parse_args(

        po::options_description encoder_opts("Encoder options");
        encoder_opts.add_options()
-            ("codec,c", PO_STR_VALUE<StringType>()->default_value(STR("libx264"), "libx264"),
-                "Output codec")
+            ("codec,c", PO_STR_VALUE<video2x::fsutils::StringType>()
+                ->default_value(STR("libx264"), "libx264"), "Output codec")
            ("no-copy-streams", "Do not copy audio and subtitle streams")
-            ("pix-fmt", PO_STR_VALUE<StringType>(), "Output pixel format")
+            ("pix-fmt", PO_STR_VALUE<video2x::fsutils::StringType>(), "Output pixel format")
            ("bit-rate", po::value<int64_t>(&enc_cfg.bit_rate)->default_value(0),
                "Bitrate in bits per second")
            ("rc-buffer-size", po::value<int>(&enc_cfg.rc_buffer_size)->default_value(0),
@@ -118,8 +119,8 @@ int parse_args(
                "Delay in milliseconds for encoder")

            // Extra encoder options (key-value pairs)
-            ("extra-encoder-option,e", PO_STR_VALUE<std::vector<StringType>>()->multitoken(),
-                "Additional AVOption(s) for the encoder (format: -e key=value)")
+            ("extra-encoder-option,e", PO_STR_VALUE<std::vector<video2x::fsutils::StringType>>()
+                ->multitoken(), "Additional AVOption(s) for the encoder (format: -e key=value)")
        ;

        po::options_description upscale_opts("Upscaling options");
@@ -130,6 +131,8 @@ int parse_args(
                ->notifier([](int v) { validate_greater_equal_one(v, "height"); }), "Output height")
            ("scaling-factor,s", po::value<int>(&proc_cfg.scaling_factor)
                ->notifier([](int v) { validate_min(v, "scaling-factor", 2); }), "Scaling factor")
+            ("noise-level,n", po::value<int>(&proc_cfg.noise_level)
+                ->notifier([](int v) { validate_min(v, "noise-level", 0); }), "Noise level")
        ;

        po::options_description interp_opts("Frame interpolation options");
@@ -144,7 +147,7 @@ int parse_args(

        po::options_description libplacebo_opts("libplacebo options");
        libplacebo_opts.add_options()
-            ("libplacebo-shader", PO_STR_VALUE<StringType>()
+            ("libplacebo-shader", PO_STR_VALUE<video2x::fsutils::StringType>()
                ->default_value(STR("anime4k-v4-a"), "anime4k-v4-a")
                ->notifier(validate_anime4k_shader_name),
                "Name/path of the GLSL shader file to use (built-in: anime4k-v4-a, anime4k-v4-a+a, "
@@ -153,17 +156,29 @@ int parse_args(

        po::options_description realesrgan_opts("RealESRGAN options");
        realesrgan_opts.add_options()
-            ("realesrgan-model", PO_STR_VALUE<StringType>()
+            ("realesrgan-model", PO_STR_VALUE<video2x::fsutils::StringType>()
                ->default_value(STR("realesr-animevideov3"), "realesr-animevideov3")
                ->notifier(validate_realesrgan_model_name),
                "Name of the RealESRGAN model to use (realesr-animevideov3, realesrgan-plus-anime, "
                "realesrgan-plus)")
        ;

+        po::options_description realcugan_opts("RealCUGAN options");
+        realcugan_opts.add_options()
+            ("realcugan-model", PO_STR_VALUE<video2x::fsutils::StringType>()
+                ->default_value(STR("models-se"), "models-se")
+                ->notifier(validate_realcugan_model_name),
+                "Name of the RealCUGAN model to use (models-nose, models-pro, models-se)")
+            ("realcugan-threads", po::value<int>()->default_value(1),
+                "Number of threads to use for RealCUGAN")
+            ("realcugan-syncgap", po::value<int>()->default_value(3),
+                "Sync gap mode; 0:no sync, 1: accurate sync: 2 = rough sync, 3: very rough sync")
+        ;
+
        po::options_description rife_opts("RIFE options");
        rife_opts.add_options()
-            ("rife-model", PO_STR_VALUE<StringType>()->default_value(STR("rife-v4.6"), "rife-v4.6")
-                ->notifier(validate_rife_model_name),
+            ("rife-model", PO_STR_VALUE<video2x::fsutils::StringType>()
+                ->default_value(STR("rife-v4.6"), "rife-v4.6")->notifier(validate_rife_model_name),
                "Name of the RIFE model to use (rife, rife-HD, rife-UHD, rife-anime, rife-v2, "
                "rife-v2.3, rife-v2.4, rife-v3.0, rife-v3.1, rife-v4, rife-v4.6)")
            ("rife-uhd", "Enable Ultra HD mode")
@@ -176,6 +191,7 @@ int parse_args(
            .add(interp_opts)
            .add(libplacebo_opts)
            .add(realesrgan_opts)
+            .add(realcugan_opts)
            .add(rife_opts);

        po::variables_map vm;
@@ -220,66 +236,69 @@ int parse_args(
        }

        if (vm.count("log-level")) {
-            std::optional<Video2xLogLevel> log_level =
-                find_log_level_by_name(vm["log-level"].as<StringType>());
-            if (!log_level.has_value()) {
-                spdlog::critical("Invalid log level specified.");
+            if (!video2x::logger_manager::LoggerManager::instance().set_log_level(
+                    wstring_to_u8string(vm["log-level"].as<video2x::fsutils::StringType>())
+                )) {
+                video2x::logger()->critical("Invalid log level specified.");
                return -1;
            }
-            arguments.log_level = log_level.value();
        }
-        set_spdlog_level(arguments.log_level);
+        video2x::logger_manager::LoggerManager::instance().hook_ffmpeg_logging();

        // Print program banner
-        spdlog::info("Video2X version {}", LIBVIDEO2X_VERSION_STRING);
-        // spdlog::info("Copyright (C) 2018-2024 K4YT3X and contributors.");
-        // spdlog::info("Licensed under GNU AGPL version 3.");
+        video2x::logger()->info("Video2X version {}", LIBVIDEO2X_VERSION_STRING);
+        // video2x::logger()->info("Copyright (C) 2018-2024 K4YT3X and contributors.");
+        // video2x::logger()->info("Licensed under GNU AGPL version 3.");

        // Assign positional arguments
        if (vm.count("input")) {
-            arguments.in_fname = std::filesystem::path(vm["input"].as<StringType>());
-            spdlog::info("Processing file: {}", arguments.in_fname.u8string());
+            arguments.in_fname =
+                std::filesystem::path(vm["input"].as<video2x::fsutils::StringType>());
+            video2x::logger()->info("Processing file: {}", arguments.in_fname.u8string());
        } else {
-            spdlog::critical("Input file path is required.");
+            video2x::logger()->critical("Input file path is required.");
            return -1;
        }

        if (vm.count("output")) {
-            arguments.out_fname = std::filesystem::path(vm["output"].as<StringType>());
+            arguments.out_fname =
+                std::filesystem::path(vm["output"].as<video2x::fsutils::StringType>());
        } else if (!arguments.benchmark) {
-            spdlog::critical("Output file path is required.");
+            video2x::logger()->critical("Output file path is required.");
            return -1;
        }

        // Parse processor type
        if (vm.count("processor")) {
-            StringType processor_type_str = vm["processor"].as<StringType>();
+            video2x::fsutils::StringType processor_type_str =
+                vm["processor"].as<video2x::fsutils::StringType>();
            if (processor_type_str == STR("libplacebo")) {
-                proc_cfg.processor_type = ProcessorType::Libplacebo;
+                proc_cfg.processor_type = video2x::processors::ProcessorType::Libplacebo;
            } else if (processor_type_str == STR("realesrgan")) {
-                proc_cfg.processor_type = ProcessorType::RealESRGAN;
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RealESRGAN;
+            } else if (processor_type_str == STR("realcugan")) {
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RealCUGAN;
            } else if (processor_type_str == STR("rife")) {
-                proc_cfg.processor_type = ProcessorType::RIFE;
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RIFE;
            } else {
-                spdlog::critical(
-                    "Invalid processor specified. Must be 'libplacebo', 'realesrgan', or 'rife'."
-                );
+                video2x::logger()->critical("Invalid processor specified.");
                return -1;
            }
        } else {
-            spdlog::critical("Processor type is required.");
+            video2x::logger()->critical("Processor type is required.");
            return -1;
        }

        // Parse hardware acceleration method
        arguments.hw_device_type = AV_HWDEVICE_TYPE_NONE;
        if (vm.count("hwaccel")) {
-            StringType hwaccel_str = vm["hwaccel"].as<StringType>();
+            video2x::fsutils::StringType hwaccel_str =
+                vm["hwaccel"].as<video2x::fsutils::StringType>();
            if (hwaccel_str != STR("none")) {
                arguments.hw_device_type =
                    av_hwdevice_find_type_by_name(wstring_to_u8string(hwaccel_str).c_str());
                if (arguments.hw_device_type == AV_HWDEVICE_TYPE_NONE) {
-                    spdlog::critical(
+                    video2x::logger()->critical(
                        "Invalid hardware device type '{}'.", wstring_to_u8string(hwaccel_str)
                    );
                    return -1;
@@ -290,11 +309,13 @@ int parse_args(
        // Parse codec to AVCodec
        enc_cfg.codec = AV_CODEC_ID_H264;
        if (vm.count("codec")) {
-            StringType codec_str = vm["codec"].as<StringType>();
+            video2x::fsutils::StringType codec_str = vm["codec"].as<video2x::fsutils::StringType>();
            const AVCodec *codec =
                avcodec_find_encoder_by_name(wstring_to_u8string(codec_str).c_str());
            if (codec == nullptr) {
-                spdlog::critical("Codec '{}' not found.", wstring_to_u8string(codec_str));
+                video2x::logger()->critical(
+                    "Codec '{}' not found.", wstring_to_u8string(codec_str)
+                );
                return -1;
            }
            enc_cfg.codec = codec->id;
@@ -306,11 +327,12 @@ int parse_args(
        // Parse pixel format to AVPixelFormat
        enc_cfg.pix_fmt = AV_PIX_FMT_NONE;
        if (vm.count("pix-fmt")) {
-            StringType pix_fmt_str = vm["pix-fmt"].as<StringType>();
+            video2x::fsutils::StringType pix_fmt_str =
+                vm["pix-fmt"].as<video2x::fsutils::StringType>();
            if (!pix_fmt_str.empty()) {
                enc_cfg.pix_fmt = av_get_pix_fmt(wstring_to_u8string(pix_fmt_str).c_str());
                if (enc_cfg.pix_fmt == AV_PIX_FMT_NONE) {
-                    spdlog::critical(
+                    video2x::logger()->critical(
                        "Invalid pixel format '{}'.", wstring_to_u8string(pix_fmt_str)
                    );
                    return -1;
@@ -320,14 +342,17 @@ int parse_args(

        // Parse extra AVOptions
        if (vm.count("extra-encoder-option")) {
-            for (const auto &opt : vm["extra-encoder-option"].as<std::vector<StringType>>()) {
+            for (const auto &opt :
+                 vm["extra-encoder-option"].as<std::vector<video2x::fsutils::StringType>>()) {
                size_t eq_pos = opt.find('=');
-                if (eq_pos != StringType::npos) {
-                    StringType key = opt.substr(0, eq_pos);
-                    StringType value = opt.substr(eq_pos + 1);
+                if (eq_pos != video2x::fsutils::StringType::npos) {
+                    video2x::fsutils::StringType key = opt.substr(0, eq_pos);
+                    video2x::fsutils::StringType value = opt.substr(eq_pos + 1);
                    enc_cfg.extra_opts.push_back(std::make_pair(key, value));
                } else {
-                    spdlog::critical("Invalid extra AVOption format: {}", wstring_to_u8string(opt));
+                    video2x::logger()->critical(
+                        "Invalid extra AVOption format: {}", wstring_to_u8string(opt)
+                    );
                    return -1;
                }
            }
@@ -335,69 +360,120 @@ int parse_args(

        // Parse processor-specific configurations
        switch (proc_cfg.processor_type) {
-            case ProcessorType::Libplacebo: {
+            case video2x::processors::ProcessorType::Libplacebo: {
                if (!vm.count("libplacebo-shader")) {
-                    spdlog::critical("Shader name/path must be set for libplacebo.");
+                    video2x::logger()->critical("Shader name/path must be set for libplacebo.");
                    return -1;
                }
                if (proc_cfg.width <= 0 || proc_cfg.height <= 0) {
-                    spdlog::critical("Output width and height must be set for libplacebo.");
+                    video2x::logger()->critical(
+                        "Output width and height must be set for libplacebo."
+                    );
                    return -1;
                }

-                proc_cfg.processor_type = ProcessorType::Libplacebo;
-                LibplaceboConfig libplacebo_config;
-                libplacebo_config.shader_path = vm["libplacebo-shader"].as<StringType>();
+                proc_cfg.processor_type = video2x::processors::ProcessorType::Libplacebo;
+                video2x::processors::LibplaceboConfig libplacebo_config;
+                libplacebo_config.shader_path =
+                    vm["libplacebo-shader"].as<video2x::fsutils::StringType>();
                proc_cfg.config = libplacebo_config;
                break;
            }
-            case ProcessorType::RealESRGAN: {
+            case video2x::processors::ProcessorType::RealESRGAN: {
                if (!vm.count("realesrgan-model")) {
-                    spdlog::critical("RealESRGAN model name must be set for RealESRGAN.");
+                    video2x::logger()->critical("RealESRGAN model name must be set for RealESRGAN."
+                    );
                    return -1;
                }
-                if (proc_cfg.scaling_factor != 2 && proc_cfg.scaling_factor != 3 &&
-                    proc_cfg.scaling_factor != 4) {
-                    spdlog::critical("Scaling factor must be set to 2, 3, or 4 for RealESRGAN.");
+                if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) {
+                    video2x::logger()->critical(
+                        "Scaling factor must be set to 2, 3, or 4 for RealESRGAN."
+                    );
                    return -1;
                }

-                proc_cfg.processor_type = ProcessorType::RealESRGAN;
-                RealESRGANConfig realesrgan_config;
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RealESRGAN;
+                video2x::processors::RealESRGANConfig realesrgan_config;
                realesrgan_config.tta_mode = false;
-                realesrgan_config.model_name = vm["realesrgan-model"].as<StringType>();
+                realesrgan_config.model_name =
+                    vm["realesrgan-model"].as<video2x::fsutils::StringType>();
                proc_cfg.config = realesrgan_config;
                break;
            }
-            case ProcessorType::RIFE: {
-                if (!vm.count("rife-model")) {
-                    spdlog::critical("RIFE model name must be set for RIFE.");
+            case video2x::processors::ProcessorType::RealCUGAN: {
+                if (!vm.count("realcugan-model")) {
+                    video2x::logger()->critical("RealCUGAN model name must be set for RealCUGAN.");
                    return -1;
                }
-                if (proc_cfg.frm_rate_mul < 2) {
-                    spdlog::critical("Frame rate multiplier must be set to at least 2 for RIFE.");
+                if (vm.count("realcugan-threads") && vm["realcugan-threads"].as<int>() < 1) {
+                    video2x::logger()->critical(
+                        "Number of threads must be at least 1 for RealCUGAN."
+                    );
+                    return -1;
+                }
+                if (vm.count("realcugan-syncgap") && (vm["realcugan-syncgap"].as<int>() < 0 ||
+                                                      vm["realcugan-syncgap"].as<int>() > 3)) {
+                    video2x::logger()->critical(
+                        "Sync gap mode must be set to 0, 1, 2, or 3 for RealCUGAN."
+                    );
+                    return -1;
+                }
+                if (proc_cfg.scaling_factor < 2 || proc_cfg.scaling_factor > 4) {
+                    video2x::logger()->critical(
+                        "Scaling factor must be set to 2, 3, or 4 for RealCUGAN."
+                    );
+                    return -1;
+                }
+                if (proc_cfg.noise_level < -1 || proc_cfg.noise_level > 3) {
+                    video2x::logger()->critical(
+                        "Noise level must be set to -1, 0, 1, 2, or 3 for RealCUGAN."
+                    );
                    return -1;
                }

-                proc_cfg.processor_type = ProcessorType::RIFE;
-                RIFEConfig rife_config;
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RealCUGAN;
+                video2x::processors::RealCUGANConfig realcugan_config;
+                realcugan_config.tta_mode = false;
+                realcugan_config.model_name =
+                    vm["realcugan-model"].as<video2x::fsutils::StringType>();
+                realcugan_config.num_threads = vm["realcugan-threads"].as<int>();
+                realcugan_config.syncgap = vm["realcugan-syncgap"].as<int>();
+                proc_cfg.config = realcugan_config;
+                break;
+            }
+            case video2x::processors::ProcessorType::RIFE: {
+                if (!vm.count("rife-model")) {
+                    video2x::logger()->critical("RIFE model name must be set for RIFE.");
+                    return -1;
+                }
+                if (proc_cfg.frm_rate_mul < 2) {
+                    video2x::logger()->critical(
+                        "Frame rate multiplier must be set to at least 2 for RIFE."
+                    );
+                    return -1;
+                }
+
+                proc_cfg.processor_type = video2x::processors::ProcessorType::RIFE;
+                video2x::processors::RIFEConfig rife_config;
                rife_config.tta_mode = false;
                rife_config.tta_temporal_mode = false;
                rife_config.uhd_mode = vm.count("rife-uhd") > 0;
                rife_config.num_threads = 0;
-                rife_config.model_name = vm["rife-model"].as<StringType>();
+                rife_config.model_name = vm["rife-model"].as<video2x::fsutils::StringType>();
                proc_cfg.config = rife_config;
                break;
            }
            default:
-                spdlog::critical("Invalid processor type.");
+                video2x::logger()->critical("Invalid processor type.");
                return -1;
        }
    } catch (const po::error &e) {
-        spdlog::critical("Error parsing arguments: {}", e.what());
+        video2x::logger()->critical("Error parsing arguments: {}", e.what());
        return -1;
    } catch (const std::exception &e) {
-        spdlog::critical("Unexpected exception caught while parsing options: {}", e.what());
+        video2x::logger()->critical(
+            "Unexpected exception caught while parsing options: {}", e.what()
+        );
        return -1;
    }

@@ -406,17 +482,19 @@ int parse_args(
    int get_vulkan_dev_ret = get_vulkan_device_prop(arguments.vk_device_index, &dev_props);
    if (get_vulkan_dev_ret != 0) {
        if (get_vulkan_dev_ret == -2) {
-            spdlog::critical("Invalid Vulkan device ID specified.");
+            video2x::logger()->critical("Invalid Vulkan device ID specified.");
            return -1;
        } else {
-            spdlog::warn("Unable to validate Vulkan device ID.");
+            video2x::logger()->warn("Unable to validate Vulkan device ID.");
            return -1;
        }
    } else {
        // Warn if the selected device is a CPU
-        spdlog::info("Using Vulkan device: {} ({:#x})", dev_props.deviceName, dev_props.deviceID);
+        video2x::logger()->info(
+            "Using Vulkan device: {} ({:#x})", dev_props.deviceName, dev_props.deviceID
+        );
        if (dev_props.deviceType == VK_PHYSICAL_DEVICE_TYPE_CPU) {
-            spdlog::warn("The selected Vulkan device is a CPU device.");
+            video2x::logger()->warn("The selected Vulkan device is a CPU device.");
        }
    }
    return 0;
--- a/tools/video2x/src/logging.cpp
+++ b/tools/video2x/src/logging.cpp
@@ -1,77 +0,0 @@
-#include "logging.h"
-
-#include <algorithm>
-#include <unordered_map>
-
-extern "C" {
-#include <libavutil/log.h>
-}
-
-std::atomic<bool> newline_required = false;
-
-void set_spdlog_level(Video2xLogLevel log_level) {
-    switch (log_level) {
-        case Video2xLogLevel::Trace:
-            spdlog::set_level(spdlog::level::trace);
-            break;
-        case Video2xLogLevel::Debug:
-            spdlog::set_level(spdlog::level::debug);
-            break;
-        case Video2xLogLevel::Info:
-            spdlog::set_level(spdlog::level::info);
-            break;
-        case Video2xLogLevel::Warning:
-            spdlog::set_level(spdlog::level::warn);
-            break;
-        case Video2xLogLevel::Error:
-            spdlog::set_level(spdlog::level::err);
-            break;
-        case Video2xLogLevel::Critical:
-            spdlog::set_level(spdlog::level::critical);
-            break;
-        case Video2xLogLevel::Off:
-            spdlog::set_level(spdlog::level::off);
-            break;
-        default:
-            spdlog::set_level(spdlog::level::info);
-            break;
-    }
-}
-
-std::optional<Video2xLogLevel> find_log_level_by_name(const StringType &log_level_name) {
-    // Static map to store the mapping
-    static const std::unordered_map<StringType, Video2xLogLevel> log_level_map = {
-        {STR("trace"), Video2xLogLevel::Trace},
-        {STR("debug"), Video2xLogLevel::Debug},
-        {STR("info"), Video2xLogLevel::Info},
-        {STR("warning"), Video2xLogLevel::Warning},
-        {STR("warn"), Video2xLogLevel::Warning},
-        {STR("error"), Video2xLogLevel::Error},
-        {STR("critical"), Video2xLogLevel::Critical},
-        {STR("off"), Video2xLogLevel::Off},
-        {STR("none"), Video2xLogLevel::Off}
-    };
-
-    // Normalize the input to lowercase
-    StringType normalized_name = log_level_name;
-    std::transform(
-        normalized_name.begin(), normalized_name.end(), normalized_name.begin(), ::tolower
-    );
-
-    // Lookup the log level in the map
-    auto it = log_level_map.find(normalized_name);
-    if (it != log_level_map.end()) {
-        return it->second;
-    }
-
-    return std::nullopt;
-}
-
-// Newline-safe log callback for FFmpeg
-void newline_safe_ffmpeg_log_callback(void *ptr, int level, const char *fmt, va_list vl) {
-    if (level <= av_log_get_level() && newline_required.load()) {
-        putchar('\n');
-        newline_required.store(false);
-    }
-    av_log_default_callback(ptr, level, fmt, vl);
-}
--- a/tools/video2x/src/newline_safe_sink.cpp
+++ b/tools/video2x/src/newline_safe_sink.cpp
@@ -0,0 +1,9 @@
+#include "newline_safe_sink.h"
+
+void newline_safe_sink::log(const spdlog::details::log_msg &msg) {
+    if (needs_newline_.exchange(false)) {
+        std::fputs("\n", stdout);
+    }
+
+    spdlog::sinks::ansicolor_stdout_sink_mt::log(msg);
+}
--- a/tools/video2x/src/validators.cpp
+++ b/tools/video2x/src/validators.cpp
@@ -2,8 +2,8 @@

 #include <unordered_set>

-void validate_anime4k_shader_name(const StringType &shader_name) {
-    static const std::unordered_set<StringType> valid_anime4k_shaders = {
+void validate_anime4k_shader_name(const video2x::fsutils::StringType &shader_name) {
+    static const std::unordered_set<video2x::fsutils::StringType> valid_anime4k_shaders = {
        STR("anime4k-v4-a"),
        STR("anime4k-v4-a+a"),
        STR("anime4k-v4-b"),
@@ -22,8 +22,8 @@ void validate_anime4k_shader_name(const StringType &shader_name) {
    }
 }

-void validate_realesrgan_model_name(const StringType &model_name) {
-    static const std::unordered_set<StringType> valid_realesrgan_models = {
+void validate_realesrgan_model_name(const video2x::fsutils::StringType &model_name) {
+    static const std::unordered_set<video2x::fsutils::StringType> valid_realesrgan_models = {
        STR("realesrgan-plus"), STR("realesrgan-plus-anime"), STR("realesr-animevideov3")
    };
    if (valid_realesrgan_models.count(model_name) == 0) {
@@ -36,8 +36,21 @@ void validate_realesrgan_model_name(const StringType &model_name) {
    }
 }

-void validate_rife_model_name(const StringType &model_name) {
-    static const std::unordered_set<StringType> valid_realesrgan_models = {
+void validate_realcugan_model_name(const video2x::fsutils::StringType &model_name) {
+    static const std::unordered_set<video2x::fsutils::StringType> valid_realcugan_models = {
+        STR("models-nose"), STR("models-pro"), STR("models-se")
+    };
+    if (valid_realcugan_models.count(model_name) == 0) {
+        throw po::validation_error(
+            po::validation_error::invalid_option_value,
+            "realcugan-model",
+            "realcugan-model must be one of: models-nose, models-pro, models-se"
+        );
+    }
+}
+
+void validate_rife_model_name(const video2x::fsutils::StringType &model_name) {
+    static const std::unordered_set<video2x::fsutils::StringType> valid_rife_models = {
        STR("rife"),
        STR("rife-HD"),
        STR("rife-UHD"),
@@ -50,7 +63,7 @@ void validate_rife_model_name(const StringType &model_name) {
        STR("rife-v4"),
        STR("rife-v4.6"),
    };
-    if (valid_realesrgan_models.count(model_name) == 0) {
+    if (valid_rife_models.count(model_name) == 0) {
        throw po::validation_error(
            po::validation_error::invalid_option_value,
            "rife-model",
--- a/tools/video2x/src/video2x.cpp
+++ b/tools/video2x/src/video2x.cpp
@@ -9,10 +9,10 @@
 #include <unistd.h>
 #endif

-#include <spdlog/spdlog.h>
+#include <libvideo2x/logger_manager.h>

 #include "argparse.h"
-#include "logging.h"
+#include "newline_safe_sink.h"
 #include "timer.h"

 // Set UNIX terminal input to non-blocking mode
@@ -53,10 +53,20 @@ int wmain(int argc, wchar_t *argv[]) {
 #else
 int main(int argc, char **argv) {
 #endif
-    // Initialize arguments structures
+    // Initialize newline-safe logger with custom formatting pattern
+    std::shared_ptr<newline_safe_sink> logger_sink = std::make_shared<newline_safe_sink>();
+    std::vector<spdlog::sink_ptr> sinks = {logger_sink};
+    if (!video2x::logger_manager::LoggerManager::instance().reconfigure_logger(
+            "video2x", sinks, "[%Y-%m-%d %H:%M:%S] [%^%l%$] %v"
+        )) {
+        std::cerr << "Error: Failed to configure logger." << std::endl;
+        return 1;
+    }
+
+    // Initialize argument and configuration structs
    Arguments arguments;
-    ProcessorConfig proc_cfg;
-    EncoderConfig enc_cfg;
+    video2x::processors::ProcessorConfig proc_cfg;
+    video2x::encoder::EncoderConfig enc_cfg;

    // Parse command line arguments
    int parse_ret = parse_args(argc, argv, arguments, proc_cfg, enc_cfg);
@@ -72,18 +82,10 @@ int main(int argc, char **argv) {
    }

    // Create video processor object
-    VideoProcessor video_processor = VideoProcessor(
-        proc_cfg,
-        enc_cfg,
-        arguments.vk_device_index,
-        arguments.hw_device_type,
-        arguments.log_level,
-        arguments.benchmark
+    video2x::VideoProcessor video_processor = video2x::VideoProcessor(
+        proc_cfg, enc_cfg, arguments.vk_device_index, arguments.hw_device_type, arguments.benchmark
    );

-    // Register a newline-safe log callback for FFmpeg
-    av_log_set_callback(newline_safe_ffmpeg_log_callback);
-
    // Create a thread for video processing
    int proc_ret = 0;
    std::atomic<bool> completed = false;  // Use atomic for thread-safe updates
@@ -91,7 +93,7 @@ int main(int argc, char **argv) {
        proc_ret = video_processor.process(arguments.in_fname, arguments.out_fname);
        completed.store(true, std::memory_order_relaxed);
    });
-    spdlog::info("Press [space] to pause/resume, [q] to abort.");
+    video2x::logger()->info("Press [space] to pause/resume, [q] to abort.");

    // Setup timer
    Timer timer;
@@ -123,14 +125,14 @@ int main(int argc, char **argv) {
        if (ch == ' ' || ch == '\n') {
            {
                // Toggle pause state
-                if (video_processor.get_state() == VideoProcessorState::Paused) {
+                if (video_processor.get_state() == video2x::VideoProcessorState::Paused) {
                    video_processor.resume();
                } else {
                    video_processor.pause();
                }

                // Print message based on current state and pause/resume the timer
-                if (video_processor.get_state() == VideoProcessorState::Paused) {
+                if (video_processor.get_state() == video2x::VideoProcessorState::Paused) {
                    std::cout
                        << "\r\033[KProcessing paused; press [space] to resume, [q] to abort.";
                    std::cout.flush();
@@ -140,15 +142,11 @@ int main(int argc, char **argv) {
                    std::cout.flush();
                    timer.resume();
                }
-                newline_required.store(true);
+                logger_sink->set_needs_newline(true);
            }
        } else if (ch == 'q' || ch == 'Q') {
            // Abort processing
-            if (newline_required.load()) {
-                putchar('\n');
-            }
-            spdlog::warn("Aborting gracefully; press Ctrl+C to terminate forcefully.");
-            newline_required.store(false);
+            video2x::logger()->warn("Aborting gracefully; press Ctrl+C to terminate forcefully.");
            video_processor.abort();
            break;
        }
@@ -159,7 +157,7 @@ int main(int argc, char **argv) {
            int64_t total_frames = video_processor.get_total_frames();

            // Print the progress bar if processing is not paused
-            if (video_processor.get_state() != VideoProcessorState::Paused &&
+            if (video_processor.get_state() != video2x::VideoProcessorState::Paused &&
                (total_frames > 0 || processed_frames > 0)) {
                double percentage = total_frames > 0 ? static_cast<double>(processed_frames) *
                                                           100.0 / static_cast<double>(total_frames)
@@ -192,7 +190,7 @@ int main(int argc, char **argv) {
                          << ":" << std::setw(2) << std::setfill('0') << minutes_remaining << ":"
                          << std::setw(2) << std::setfill('0') << seconds_remaining;
                std::cout.flush();
-                newline_required.store(true);
+                logger_sink->set_needs_newline(true);
            }
        }

@@ -208,44 +206,43 @@ int main(int argc, char **argv) {
    // Join the processing thread to ensure it completes before exiting
    processing_thread.join();

-    // Print a newline if progress bar was displayed
-    if (newline_required.load()) {
-        std::cout << '\n';
-    }
-
    // Print final message based on processing result
-    if (video_processor.get_state() == VideoProcessorState::Aborted) {
-        spdlog::warn("Video processing aborted");
+    if (video_processor.get_state() == video2x::VideoProcessorState::Aborted) {
+        video2x::logger()->warn("Video processing aborted");
        return 2;
-    } else if (proc_ret != 0 || video_processor.get_state() == VideoProcessorState::Failed) {
-        spdlog::critical("Video processing failed with error code {}", proc_ret);
+    } else if (proc_ret != 0 ||
+               video_processor.get_state() == video2x::VideoProcessorState::Failed) {
+        video2x::logger()->critical("Video processing failed with error code {}", proc_ret);
        return 1;
    } else {
-        spdlog::info("Video processed successfully");
+        video2x::logger()->info("Video processed successfully");
    }

-    // Calculate statistics
-    int64_t processed_frames = video_processor.get_processed_frames();
-    int time_elapsed = static_cast<int>(timer.get_elapsed_time() / 1000);
-    auto [hours_elapsed, minutes_elapsed, seconds_elapsed] =
-        calculate_time_components(time_elapsed);
-    float average_speed_fps = static_cast<float>(processed_frames) /
-                              (time_elapsed > 0 ? static_cast<float>(time_elapsed) : 1);
+    // Print the processing summary if the log level is info or lower
+    if (video2x::logger()->level() <= spdlog::level::info) {
+        // Calculate statistics
+        int64_t processed_frames = video_processor.get_processed_frames();
+        int time_elapsed = static_cast<int>(timer.get_elapsed_time() / 1000);
+        auto [hours_elapsed, minutes_elapsed, seconds_elapsed] =
+            calculate_time_components(time_elapsed);
+        float average_speed_fps = static_cast<float>(processed_frames) /
+                                  (time_elapsed > 0 ? static_cast<float>(time_elapsed) : 1);

-    // Print processing summary
-    std::cout << "====== Video2X " << (arguments.benchmark ? "Benchmark" : "Processing")
-              << " summary ======" << std::endl;
-    std::cout << "Video file processed: " << arguments.in_fname.u8string() << std::endl;
-    std::cout << "Total frames processed: " << processed_frames << std::endl;
-    std::cout << "Total time taken: " << std::setw(2) << std::setfill('0') << hours_elapsed << ":"
-              << std::setw(2) << std::setfill('0') << minutes_elapsed << ":" << std::setw(2)
-              << std::setfill('0') << seconds_elapsed << std::endl;
-    std::cout << "Average processing speed: " << std::fixed << std::setprecision(2)
-              << average_speed_fps << " FPS" << std::endl;
+        // Print processing summary
+        std::cout << "====== Video2X " << (arguments.benchmark ? "Benchmark" : "Processing")
+                  << " summary ======" << std::endl;
+        std::cout << "Video file processed: " << arguments.in_fname.u8string() << std::endl;
+        std::cout << "Total frames processed: " << processed_frames << std::endl;
+        std::cout << "Total time taken: " << std::setw(2) << std::setfill('0') << hours_elapsed
+                  << ":" << std::setw(2) << std::setfill('0') << minutes_elapsed << ":"
+                  << std::setw(2) << std::setfill('0') << seconds_elapsed << std::endl;
+        std::cout << "Average processing speed: " << std::fixed << std::setprecision(2)
+                  << average_speed_fps << " FPS" << std::endl;

-    // Print additional information if not in benchmark mode
-    if (!arguments.benchmark) {
-        std::cout << "Output written to: " << arguments.out_fname.u8string() << std::endl;
+        // Print additional information if not in benchmark mode
+        if (!arguments.benchmark) {
+            std::cout << "Output written to: " << arguments.out_fname.u8string() << std::endl;
+        }
    }

    return 0;
--- a/tools/video2x/src/vulkan_utils.cpp
+++ b/tools/video2x/src/vulkan_utils.cpp
@@ -3,7 +3,7 @@
 #include <iostream>
 #include <vector>

-#include <spdlog/spdlog.h>
+#include <libvideo2x/logger_manager.h>

 static int enumerate_vulkan_devices(VkInstance *instance, std::vector<VkPhysicalDevice> &devices) {
    // Create a Vulkan instance
@@ -12,7 +12,7 @@ static int enumerate_vulkan_devices(VkInstance *instance, std::vector<VkPhysical

    VkResult result = vkCreateInstance(&create_info, nullptr, instance);
    if (result != VK_SUCCESS) {
-        spdlog::error("Failed to create Vulkan instance.");
+        video2x::logger()->error("Failed to create Vulkan instance.");
        return -1;
    }

@@ -20,7 +20,9 @@ static int enumerate_vulkan_devices(VkInstance *instance, std::vector<VkPhysical
    uint32_t device_count = 0;
    result = vkEnumeratePhysicalDevices(*instance, &device_count, nullptr);
    if (result != VK_SUCCESS || device_count == 0) {
-        spdlog::error("Failed to enumerate Vulkan physical devices or no devices available.");
+        video2x::logger()->error(
+            "Failed to enumerate Vulkan physical devices or no devices available."
+        );
        vkDestroyInstance(*instance, nullptr);
        return -1;
    }
@@ -28,7 +30,7 @@ static int enumerate_vulkan_devices(VkInstance *instance, std::vector<VkPhysical
    devices.resize(device_count);
    result = vkEnumeratePhysicalDevices(*instance, &device_count, devices.data());
    if (result != VK_SUCCESS) {
-        spdlog::error("Failed to retrieve Vulkan physical devices.");
+        video2x::logger()->error("Failed to retrieve Vulkan physical devices.");
        vkDestroyInstance(*instance, nullptr);
        return -1;
    }
@@ -96,7 +98,7 @@ int list_vulkan_devices() {

 int get_vulkan_device_prop(uint32_t vk_device_index, VkPhysicalDeviceProperties *dev_props) {
    if (dev_props == nullptr) {
-        spdlog::error("Invalid device properties pointer.");
+        video2x::logger()->error("Invalid device properties pointer.");
        return -1;
    }
Author	SHA1	Message	Date
k4yt3x	996b0bfa78	docs(readme): update readme for 6.3.1 Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	709cf08760	fix(encoder): decreasing PTS precision with increasing video duration (#1269 ) Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	bf5917b084	docs(readme): add French to the list of GUI languages Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	96ebd860e3	docs(readme): add sample clips for Real-CUGAN and Real-ESRGAN Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	2afbeab2aa	fix(video2x): remove redundant newline statements Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	5c4d23bc97	docs(readme): update readme for 6.3.0 Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	20e43a9724	chore: update changelogs for version 6.3.0 Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
k4yt3x	486c23d9f4	fix(video2x): add missing Real-CUGAN help messages Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 00:00:00 +00:00
K4YT3X	127d9e0019	feat(realcugan): add support for Real-CUGAN ncnn Vulkan (#1268 ) Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-21 02:58:19 +00:00
k4yt3x	e1e8d64056	chore(deps): update spdlog to v1.15.0 Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-20 00:00:00 +00:00
k4yt3x	7b8e186207	feat(logger): improve logger initialization Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-20 00:00:00 +00:00
k4yt3x	f3999a431c	feat(logger): improve FFmpeg logging format Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-20 00:00:00 +00:00
K4YT3X	6676cd2439	feat(logger): add logger manager to provide unified logging (#1267 ) Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-20 04:46:10 +00:00
k4yt3x	b8eb6de59b	docs(readme): add licensing information for RIFE ncnn Vulkan Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-19 00:00:00 +00:00
K4YT3X	ae2d5d32e4	feat(ns): improve optimization flags and add namespaces (#1261 ) Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-17 16:24:51 +00:00
k4yt3x	5884dd1ba4	fix(encoder): always use the calculated PTS with corrected math Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-16 00:00:00 +00:00
k4yt3x	33ee43a088	docs(citation): add citation information Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-15 00:00:00 +00:00
k4yt3x	7ee9d60fcf	docs(readme): update readme for 6.2.0 Signed-off-by: k4yt3x <i@k4yt3x.com>	2024-12-12 00:00:00 +00:00