Vapourkit is a free, open source program for video upscaling and enhancement using VapourSynth and AI models. It provides a user-friendly interface for video processing with support for both NVIDIA TensorRT and DirectML (AMD/Intel/NVIDIA) backends.
- Download and extract/install to your desired location
- On first launch, click "Start Setup" when prompted to install dependencies
- Select or drag-and-drop a video file
- Choose an upscaling model
- Configure output location and format
- Click "Upscale Video" to process
- Use "Preview Output" or "Compare Videos" to review results
For advanced features like custom filters and workflows, see Advanced Mode.
- AI Video Upscaling: Process videos with high quality AI upscaling models
- Dual Backend Support: TensorRT (NVIDIA) or DirectML (AMD/Intel/NVIDIA)
- Real-time Preview: See results while processing
- Video Comparison: Built-in side-by-side viewer
- Batch Processing: Upscale multiple videos sequentially
- Pre-made Filters: Dozens of ready-to-use filters (thanks pifroggi!)
- Custom VapourSynth Filters: Write and chain custom video processing filters
- Templates & Workflows: Save/share filter configs (
.vkfilter) and complete workflows (.vkworkflow) - Custom Models: Import your own ONNX models
- Enhanced Batch Processing: Process multiple videos sequentially with custom workflows
See Model Support for included models, custom model requirements, and model licensing details.
- OS: Windows 10/11 (x64)
- RAM: 8GB+ recommended
- Storage: 5 GB Minimum, 10 GB recommended free space for application and dependencies
- GPU:
- Minimum 6 GB VRAM
- NVIDIA 16xx series or newer (for TensorRT) AND at least driver version 580.x!
- AMD/Intel GPU with DirectX 12 support (for DirectML)
See Development for more information.
GPL 3.0 - See LICENSE file for details
Chat here about Vapourkit!
- VapourSynth & vs-mlrt
- tepete's work on the filters & his plugins
- video-compare
- Sirosky's models
- the database's models
- vs-jetpack for additional VapourSynth filters
- tepete/pifroggi, Bendel, leobby, Princess, and Hermes for beta testing!