SUPIR by XPixel Group
verifiedFreexpixel.group
Open academic research model (SUPIR) for high-fidelity, text-guided AI image restoration and upscaling.
What it does
SUPIR, from the XPixel Group, is a research project for large-scale, diffusion-based image restoration that uses text prompts to guide intelligent repair of low-quality images. It targets restoration across landscapes, portraits, animals, architecture, games, and vintage film/photos, with an associated paper and code release.
Core features
Text-driven, diffusion-based image restoration
Specialized handling for faces, landscapes, animals, and architecture
Restoration presets for gaming screenshots and classic film footage
Vintage photo revival for old/degraded images
Published research paper and GitHub codebase
Commercial spinoff product (SupPixel AI) for restoration and upscaling
Best for
→Researchers experimenting with diffusion-based restoration techniques
→Developers building on the open SUPIR model via GitHub
→Users wanting to restore old or degraded photos through the SupPixel product
Reviews
Big-picture takes: what it's for and whether it delivers. High-engagement YouTube videos — not sponsored.