RelayVSR Video Upscaler

RelayVSR Video Upscaler

Raises the resolution of video with the RelayVSR diffusion model in a Gradio page on Hugging Face. For editors and developers who want sharper footage with textures that stay steady between frames.

Made by Upsampler on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: RelayVSR Video Upscaler is a Gradio app that raises video resolution with a diffusion model called RelayVSR. A Wan transformer restores a few keyframes at four times their size. A light Flash decoder then passes that detail to the frames between them, which keeps textures steady from frame to frame.

Who it suits: Editors and developers who want sharper footage from low-resolution clips, and researchers who want to see how diffusion video super-resolution runs.

Industry | Who uses it | How they use it | Example
Video production | Editors | Sharpen low-resolution footage before cutting it | Raising the resolution of an old clip
Archives | Archivists | Restore detail in stored video | Trying a short archive clip in the browser
Research | Vision researchers | Study keyframe-based video super-resolution | Reading the refvsr_inference folder

You can skip it if you never work with video, or if you need a tool with a fixed, documented workflow.

Where to use it:

• A web browser, on the Space's own page at Hugging Face
• A copy on your own computer, using Python 3.12 on Windows, macOS or Linux

Use case | Who | What they get back
Upscale a clip | Editors | A higher-resolution version of the video
Study the method | Developers and researchers | The bundled inference code to read and run

How to set it up:

• A browser, if you only use the Space online
• Python 3.12 and pip, if you run a copy yourself

1. Open the Space's page on Hugging Face to use it online. Nothing needs installing.
2. For a local copy, install the huggingface_hub package and download the Space's files.
3. Move into the new folder, then install Gradio and the packages in requirements.txt.
4. Start the app with python.

pip install -U huggingface_hub
hf download Upsampler/relayvsr --repo-type space --local-dir relayvsr
cd relayvsr
pip install gradio==5.49.1
pip install -r requirements.txt

How to use it: Open the Space's page and use the controls it shows to load a video and start the upscale. The source gives no further steps. On your own computer, start the app first and open the page it serves.

python app.py

Tool | What it does
python app.py | Starts the upscaler on your own computer

Example: You load a short low-resolution clip into the page and start the upscale. You get back the clip at a higher resolution, with the detail carried across the frames.

For newcomers and experts: If you are new to this, try a short clip first. Video upscaling is slow work, so a few seconds of footage shows you what it does without a long wait.

For experts, the method restores sparse keyframes with a Wan transformer and relays their detail to the in-between frames with a light Flash decoder. That trades some work per keyframe for steadier textures across the clip. The inference code from the kopperx/RelayVSR project, under Apache-2.0, sits unchanged in the refvsr_inference folder.

Load a video
A Wan transformer restores keyframes at 4x
A Flash decoder relays detail to other frames
The upscaled video comes back

Good to know:

• Built with Gradio 5.49.1 and Python 3.12, with app.py as the main file
• License: apache-2
• Cost to run: none named
• Data it reaches: none named

On Hugging Face, 13 November 2026

• Likes: 3
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/Upsampler/relayvsr
• Files: https://huggingface.co/spaces/Upsampler/relayvsr/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.