Spaces:
Running on Zero
Running on Zero
A newer version of the Gradio SDK is available: 6.24.0
metadata
title: MirrorPPR
emoji: 🪞
colorFrom: purple
colorTo: indigo
sdk: gradio
sdk_version: 5.50.0
app_file: app.py
short_description: Exemplar-based portrait photo retouching
python_version: '3.12'
startup_duration_timeout: 1h
MirrorPPR: Exemplar-Based Portrait Photo Retouching
This Space demonstrates MirrorPPR, a novel framework for exemplar-based portrait photo retouching. Given an exemplar pair (an original face image and its retouched version), the model infers the retouching operations and applies them to a new query image.
Usage
- Upload a query face image you want to retouch.
- Provide an exemplar pair: the original and retouched version of another face showing the desired retouching style. You can quickly load a pre-packaged exemplar pair from the template gallery.
- Click "Retouch" to apply the exemplar's retouching operations to your query image.
Model
This Space uses the SJTU-DENG-Lab/MirrorPPR-Face model, built on top of Qwen-Image-Edit with a Retouching Operation Extractor (MAE + R-Former + Connector) and LoRA adaptation.