The first public release of Vision Studio-X - a professional, local-first AI image and video generation desktop app. Everything runs on your own GPU. No cloud, no subscription, no data leaving your machine. Now fully open source under the MIT License.
Download (Windows x64)
| File | What it is |
|---|---|
| Vision Studio Setup 3.0.0.exe | Windows installer (recommended) - creates Desktop + Start Menu shortcuts |
| Vision Studio-3.0.0-win.zip | Portable build - unzip and run Vision Studio.exe |
First launch downloads the PyTorch / model runtime (~2-3 GB) for on-device generation. An NVIDIA GPU with 8 GB+ VRAM is recommended; CPU works but is slow.
These binaries are unsigned, so Windows SmartScreen may warn "unknown publisher" - click More info -> Run anyway. Signed builds are planned.
Highlights
- Image generation - FLUX.1, Stable Diffusion XL, SD 1.5
- Video generation - LTX Video, Stable Video Diffusion, AnimateDiff
- Canvas-native control layers with reusable inspector, rail, and mask editing
- Script to storyboard to timeline pipeline with scene-linked continuity metadata
- Real timeline playback plus silent local MP4 export with progress tracking
- Refinement pipelines with a visual node builder and 6 built-in presets
- Batch processing, social presets (YouTube / TikTok / Instagram), and a professional editor
Requirements
- Windows 10 / 11 64-bit
- 8 GB RAM minimum (16 GB recommended)
- ~10 GB free disk (plus storage for downloaded models)
- NVIDIA GPU with 8 GB+ VRAM recommended (optional)
Tech stack
Electron 33, React 19, TypeScript, Tailwind CSS v4, Python FastAPI, PyTorch
Full details in the CHANGELOG. Found a bug? Open an issue.