AI Face Restoration

AI Face Restoration

Enlarges group photos, portraits and crowd shots two, four or eight times and lets you compare any face before and after. A static demo for people who work with old or small photos.

Made by MegaPotatoAI 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: AI Face Restoration is a static web page, called Face Enhancer, that enlarges group photos, portraits and crowd shots by 2, 4 or 8 times. Once it finishes, you can choose a face and look at the original next to the enlarged one, at the same place and size. The upscaling itself runs on a proprietary model at MegaPotato, not in your browser.

Who it suits: It suits people with small or old photos of groups, such as family photographers, archivists and anyone who wants a closer look at faces in a crowd.

Industry | Who uses it | How they use it | Example
Photography | Photographers | Enlarge a group photo and check each face | Run a team picture at 4 times and compare faces
Archives and history | Archivists | Look closely at faces in old scans | Enlarge a portrait and compare before and after
Family use | Individuals | Improve small faces in a shared photo | Mark one relative and view the result

If you need a download of the full-size result in the demo itself, you can skip it, since that step happens on the MegaPotato site.

Where to use it:

• A web browser, on the Space's page on Hugging Face
• A web browser on Windows, macOS or Linux, from a local copy of the page files

Use case | Who | What they get back
Enlarge a photo | Anyone | A 2, 4 or 8 times larger version, with the output size shown
Compare faces | Photographers and archivists | A before and after pair for each face you mark
Try the examples | Visitors | Real results from the bundled examples, with no login

How to set it up:

• A web browser, which is all the online page needs
• A Hugging Face login, to enhance your own photos in the demo
• Python with pip, only if you want the page files on your own computer

1. Open the Space's page on Hugging Face to use it online. Nothing needs installing.
2. Log in to Hugging Face if you want to run your own photo.
3. For a local copy, install the Hugging Face hub package with pip.
4. Download the Space's files into a folder named Face-Enhancer.
5. Move into that folder and open index.html in a web browser.

pip install -U huggingface_hub
hf download MegaPotatoAI/Face-Enhancer --repo-type space --local-dir Face-Enhancer
cd Face-Enhancer

How to use it: Add a PNG, JPEG or WebP photo, choose a scale, mark the faces you care about and run the enhancement. Then compare each face before and after.

1. Choose, drop or paste your photo.
2. Pick 2, 4 or 8 times. The page shows the exact output size.
3. Select each face to examine by clicking or tapping, and the page numbers its box. Boxes can be placed ahead of the run.
4. Run it with the Enhance faces button. Each box then shows a pair of images, and you can move a box or flip the full photo between its two states.

Tool | What it does
Scale choice | Sets the enlargement to 2, 4 or 8 times
Face boxes | Marks the faces you choose, and you place every box yourself
Enhance faces | Runs the model and turns each box into a Before and After pair
Before and After switch | Flips the whole photo between the original and the result

Example: Paste a group photo, pick 4 times and mark three faces. After you press Enhance faces, each box shows the face before and after, and you can drag a box to inspect a different spot.

For newcomers and experts: Try the bundled examples first, because they work without a login. Then log in and run one of your own photos at 2 times to see the demo's behavior.

The page is a plain HTML file. Your photo travels to a MegaPotato service hosted on Cloudflare. That service keeps the API key, enforces the daily limits and forwards the image to the MegaPotato API. The model behind it is proprietary and runs on MegaPotato's own GPUs. The page never finds faces by itself, so each box is one you drew.

You add a photo and mark faces
The page sends it to the MegaPotato service
The service checks limits and the image
The proprietary model enlarges the photo
The page shows Before and After for each face

Good to know: The model weights are not published, and the demo is limited to 2 runs a day per network address over a rolling 24 hours.

• License: MIT for the demo code, while access to the proprietary model and API follows the MegaPotato service terms
• Cost to run: the demo is free within its daily limit, and a full-size download needs an account on the MegaPotato site, where new accounts get free chips
• Data it reaches: your uploaded photo, which goes to the MegaPotato service and is reviewed automatically before processing
• Not for: images that may break the content rules, which are refused, or photos you do not want public in the Community tab

On Hugging Face, 13 November 2026

• Likes: 4
• Built with: static
• License: mit

Links

• Space: https://huggingface.co/spaces/MegaPotatoAI/Face-Enhancer
• Files: https://huggingface.co/spaces/MegaPotatoAI/Face-Enhancer/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.