pixworker

中文文档

Note: This project is in early development. Features are experimental and may be incomplete or unstable. Use for learning and research purposes only.

A video enhancement tool powered by ONNX Runtime, supporting frame interpolation and upscaling.

Features

Frame Interpolation (VFI): Enhance video frame rate using GIMM-VFI models for smoother playback
Upscaling: 4x or higher resolution enhancement using Real-ESRGAN models
Hardware Acceleration:
- macOS: CoreML acceleration
- Linux: CUDA / TensorRT acceleration
- Windows: CUDA / TensorRT acceleration
NTSC Video Processing: Deinterlacing support for interlaced video content

Build Requirements

Prerequisites

Rust: Latest stable version (install via rustup)
Make: For build automation

Platform-Specific Requirements

macOS

xcode-select --install

Linux

# Ubuntu/Debian
sudo apt install build-essential

# Optional: CUDA and TensorRT for GPU acceleration

Windows

Visual Studio 2019 or later (with C++ build tools)
Or MinGW-w64 via MSYS2

Building

Quick Start

# Clone the repository
git clone <repository-url>
cd pixworker

# Build release version (recommended)
make

# Or build debug version
make debug

Cross-Platform Compilation

# View all available targets
make help

# Build for specific platforms
make macos-x64      # macOS x86_64
make macos-arm64    # macOS Apple Silicon
make linux-x64      # Linux x86_64
make linux-arm64    # Linux ARM64

# Build all platforms and package
make dist

Note:

Windows binaries cannot be cross-compiled from macOS/Linux (due to ring crate limitations)
Windows builds must be done on a Windows machine
Cross-compiling Linux targets may require additional linker configuration

Binary Locations

Release build: target/release/pixworker
Debug build: target/debug/pixworker
Multi-platform package: dist/<target-triple>/pixworker

Usage

Model Download

On first run, you need to download ONNX model files. Place models in the following directories:

~/.cache/pixworker/models/vfi/       # Frame interpolation models
~/.cache/pixworker/models/upscale/   # Upscaling models

Or the program will automatically download them from huggingface.co.

Basic Commands

# Frame interpolation - increase video frame rate to 60fps
pixworker enhance --vfi 60fps --vfi-model gimm-vfi-f-p-hf -i input.mp4 -o output.mp4

# Video upscaling - 4x resolution
pixworker enhance --upscale 4.0 --upscale-model realesr-animevideov3-hf -i input.mp4 -o output.mp4

# Show help
pixworker --help

Development

# Run build
make

# Clean build artifacts
make clean

License

Source Code

The pixworker source code is licensed under the MIT License. See LICENSE file for details.

AI Models

This project uses AI models with different licenses:

GIMM-VFI (Frame Interpolation)

License: S-Lab License 1.0 (Non-Commercial)
Link: https://github.com/GSeanCDAT/GIMM-VFI/blob/main/LICENSE
Restrictions:
- ✅ Personal, non-commercial use
- ✅ Academic research
- ✅ Educational purposes
- ❌ Commercial products or services
- ❌ Profit-generating activities
Commercial Use: Contact the contributors for permission

Real-ESRGAN (Upscaling)

License: BSD 3-Clause License
Link: https://github.com/xinntao/Real-ESRGAN/blob/master/LICENSE
Restrictions:
- ✅ Commercial use allowed
- ✅ Modification allowed
- ✅ Distribution allowed
- ⚠️ Must include copyright notice and license text

Usage Guidelines

Important: Due to GIMM-VFI's non-commercial license:

If you use VFI (frame interpolation): The software cannot be used for commercial purposes
If you only use upscaling: Commercial use is permitted under BSD 3-Clause terms

The software will prompt you to accept the respective model license when downloading models for the first time.

universonic/pixworker