Paints-Undo is a project aimed at providing base models of human drawing behaviors with a hope that future AI models can better align with the real ne

Search code, repositories, users, issues, pull requests...

submited by

Style Pass

2024-07-11 08:00:03

Paints-Undo is a project aimed at providing base models of human drawing behaviors with a hope that future AI models can better align with the real needs of human artists.

The name "Paints-Undo" is inspired by the similarity that, the model's outputs look like pressing the "undo" button (usually Ctrl+Z) many times in digital painting software.

Paints-Undo presents a family of models that take an image as input and then output the drawing sequence of that image. The model displays all kinds of human behaviors, including but not limited to sketching, inking, coloring, shading, transforming, left-right flipping, color curve tuning, changing the visibility of layers, and even changing the overall idea during the drawing process.

(If you do not know how to use these commands, you can paste those commands to ChatGPT and ask ChatGPT to explain and give more detailed instructions.)

The inference is tested with 24GB VRAM on Nvidia 4090 and 3090TI. It may also work with 16GB VRAM, but does not work with 8GB. My estimation is that, under extreme optimization (including weight offloading and sliced attention), the theoretical minimal VRAM requirement is about 10~12.5 GB.