MaskDiffusion: Boosting Text-to-Image Consistency with Conditional Mask （IJCV）

🚩 TODO/Updates

• Basic Code.
• Demo
• Integration with LLM Priors.
• Support for Other Pre-trained Models

---

Setup

Environment

Our code builds on the requirement of the official Stable Diffusion repository. To set up their environment, please run:

conda env create -f environment/environment.yaml
conda activate maskdiffusion

Testing

Testing with the mini-testset

python run_maskdiffusion.py

Acknowledgements

This orignal code is builds on the code from the diffusers library and Prompt-to-Prompt. When reorganizing the code to be public, I also reused code from Attend-and-Excite and Densediffusion, to achieve a more concise implementation.