We extend pix2pix-Turbo and CycleGAN-Turbo with a plug-and-play warp–unwarp strategy.

Specifically, this framework supports three tasks:

• Human relighting
• Driving scene relighting
• Driving scene translation (weather & time-of-day)

Setup and environment installation is in install.md.

Dataset download, preparation and warping instructions are in dataset.md.

Model training and testing instructions are in train_test.md.

Key warp–unwarp code insertions are marked with ✅.

The main insertion points are:

• Data loading → src/my_utils/training_utils.py (PairedDataset, UnpairedDataset)
• Pix2Pix-Turbo → src/train_pix2pix_turbo.py
• CycleGAN-Turbo → src/train_cyclegan_turbo.py
• Warp utilities → src/warp_utils/

• Add Gradio demo
• Add arXiv link

This project builds upon the following excellent open-source works:

• img2img-turbo
• Instance-Warp
• Two-Plane Prior