Spaces:
Runtime error
Runtime error
File size: 1,458 Bytes
c05148c 3111947 dcbf5e3 3111947 dcbf5e3 3111947 dcbf5e3 3111947 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 |
---
title: Composable-Diffusion
sdk: gradio
sdk_version: 3.12.0
app_file: app.py
pinned: true
---
# Composable Diffusion
**Compositional Visual Generation with Composable Diffusion Models (ECCV 2022)**
**[Webpage](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/) | [GitHub](https://github.com/energy-based-model/Compositional-Visual-Generation-with-Composable-Diffusion-Models-PyTorch)**
## Overview
We propose to use **conjunction and negation** (negative prompts) operators for **compositional generation with conditional diffusion models in test time without any training**.
For more details, please refer to our paper:
[Compositional Visual Generation with Composable Diffusion Models](https://arxiv.org/abs/2206.01714).<br>
[Nan Liu](https://nanliu.io)*\, [Shuang Li](https://people.csail.mit.edu/lishuang)*\, [Yilun Du](https://yilundu.github.io)*\, [Antonio Torralba](https://groups.csail.mit.edu/vision/torralbalab/), [Joshua B. Tenenbaum](https://mitibmwatsonailab.mit.edu/people/joshua-tenenbaum/), **ECCV 2022**
## Citation
If you find our paper useful in your research, please cite the following paper:
``` latex
@article{liu2022compositional,
title={Compositional Visual Generation with Composable Diffusion Models},
author={Liu, Nan and Li, Shuang and Du, Yilun and Torralba, Antonio and Tenenbaum, Joshua B},
journal={arXiv preprint arXiv:2206.01714},
year={2022}
}
```
|