Video Upscale uses AI to increase the resolution of any video — up to 4K or 10x the original size — while recovering fine detail, sharpening faces, and removing noise. It supports five different AI models tuned for different content types and quality budgets.
Standard upscaling tools simply stretch pixels and get blurry. AI upscaling reconstructs detail that wasn't in the original, making old footage look crisp, AI-generated clips look cinematic, and compressed social videos look broadcast-quality.
What you can do
- upscale_video — increase resolution using your choice of AI model, up to 10x factor or a specific target resolution like 1080p, 2K, or 4K
- list_models — browse available models with descriptions to choose the right one for your content
Who it's for
Filmmakers restoring old or low-res footage, content creators polishing AI-generated video clips, social media editors bringing compressed downloads back to full quality, and archivists upgrading historical video for modern display.
How to use it
- Use list_models to see the available options and pick one suited to your content
- Run upscale_video with the video URL and your chosen model
- Set target_resolution (1080p, 2K, 4K) or an upscale_factor (2x, 4x, etc.)
- For AI-generated footage, use the aigc enhancement preset; for old film, use the old_film preset; for faces, choose the seedvr2 model
Getting started
Have a publicly accessible video URL ready. For general content, the default model works well. For premium results on faces and detailed textures, choose seedvr2. For the fastest turnaround, try flashvsr.
