AI · AI model
SVFR
Open-source AI framework for video face restoration and colorization
SVFR is an AI model developed by Xiamen University and Tencent Youtu Lab designed for blind video face restoration. It enhances low-quality, blurry, or vintage face footage and provides colorization without introducing temporal flickering.
What it's for
- Restore blurry and degraded facial details in archive video footage
- Colorize black-and-white videos featuring faces
- Inpaint damaged or occluded facial regions across video frames
Features
- Blind face restoration (BFR) for video
- Video face colorization
- Combined restoration and colorization pipeline
- Temporal consistency to prevent flickering across frames
- Interactive Gradio web demo interface
Pricing
Open source. Free and open source
Details
- Type
- AI model
- Category
- AI
- Topic
- Video Enhancement
- Open source
- Yes
- Platforms
- Web
- For
- Video editors, archivists, and AI researchers working with vintage footage
- Alternative to
- CodeFormer, GFPGAN
- Tags
- face restoration, video enhancement, colorization, huggingface, open source
Why it was saved
- @nathanhodgson.ai: Restores, sharpens, and colorizes blurry or vintage video footage without temporal flickering.
Related
Open-source zero-shot text-to-speech and voice cloning model
Open-source expressive desktop companion robot for AI experimentation and coding
Browser that installs and runs open-source AI apps with one click
Open-weights fine-tune of Llama 3.1 built for agentic function calling.
Open-source, self-hosted AI agent with persistent memory and messaging integrations.
AI agent skill for deep social research across Reddit, X, and more.
Updated