vadbeg / diffusers-inpainting

Diffusers pipeline for inpainting with any available finetune

Diffusers Inpainting

This is a repository for image inpainting with a Stable Diffusion finetunes which weren't trained on inpainting task. Code is based on pipeline from huggingface 🤗 Diffusers library.

It is a simple learning project, it is better to use StableDiffusionInpaintPipeline or StableDiffusionInpaintPipelineLegacy from 🤗 Diffusers library.

Installation

1. Create a virtual environment:

virtualenv -p python3.9 .venv && source .venv/bin/activate

2. Install all requirements:

pip install -r requirements.txt

3. Use the project 🎉

Usage

To run use command below:

python run_simple_inpainting.py --device cuda:1 --prompt "Face of a yellow cat, high resolution, sitting on a park bench" --strength 0.95 --seed 0

Diffusion inpainting process

This gif was created by decoding latent features at each step of the diffusion process.

Built With

• 🤗 Diffusers - Huggingface diffusion models library
• Typer - CLI framework

License

This project is licensed under the MIT License - see the LICENSE file for details.

Authors

• Vadim Titko aka Vadbeg - LinkedIn | GitHub