You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

By requesting access, you agree to the GenLit Non-Commercial Research License
(https://genlit.is.tue.mpg.de/license.html). This license restricts use to
non-commercial scientific research, education, and artistic projects.

GenLit: ControlNet weights for object relighting

Three checkpoints are released, details below. Paper · Code on GitHub

Subfolder	Base model	Resolution	Frames	Use case
`single_object/`	`stabilityai/stable-video-diffusion-img2vid`	512×512	14	single object
`mit/`	`stabilityai/stable-video-diffusion-img2vid`	768×512	14	for MIT Multi-Illumination dataset
`multi_object/`	`stabilityai/stable-video-diffusion-img2vid-xt`	640×448	25	multiple objects scattered in the scene

Usage

Install the genlit package, then run:

pip install -e .  # from the genlit GitHub repo
huggingface-cli login  # gated repo — accept license first
python -m genlit.inference --mode single --img_json examples/single.json --output_dir out/

Weights download automatically from this repo on first run.

Citation

@inproceedings{genlit:sigasia:2025,
  title  = {{GenLit}: Reformulating Single Image Relighting as Video Generation},
  author = {Bharadwaj, Shrisha and Feng, Haiwen and Becherini, Giorgio and
            Abrevaya, Victoria Fernandez and Black, Michael J.},
  year   = {2025},
  booktitle = {SIGGRAPH Asia Conference Papers '25},
  doi    = {10.1145/3757377.3763970},
}

License

Non-commercial research only. See https://genlit.is.tue.mpg.de/license.html.

Downloads last month: -