360PanT: Training-Free Text-Driven 360-Degree Panorama-to-Panorama Translation
| Link | Description |
|---|---|
 |
Google Colab of 360PanT [which can be run on Nvidia T4 (16G)] |
|
Google Colab of 360PanT (F) [which can be run on Nvidia L4 (24GB)] |
| Datasets | 360PanoI-Pan2Pan and 360syn-Pan2Pan |
| basis.zip | semantic bases of indoors and outdoors for FreeControl and 360PanT (F) |
360 panoramic images viewer: It could be used to view the synthesized 360-degree panorama.
CLIP-score: It contains Google Colab to calculate the CLIP-score.
DINO-score: It contains Google Colab to calculate the DINO-score.
If you find the code helpful in your research or work, please cite our paper:
@InProceedings{Wang_2025_WACV,
author = {Wang, Hai and Xue, Jing-Hao},
title = {360PanT: Training-Free Text-Driven 360-Degree Panorama-to-Panorama Translation},
booktitle = {Proceedings of the Winter Conference on Applications of Computer Vision (WACV)},
month = {February},
year = {2025},
pages = {212-221}
}We thank MultiDiffusion, Plug-and-Play, FreeControl and StitchDiffusion. Our work is based on their excellent codes.