v0xie / sd-webui-incantations

Enhance Stable Diffusion image quality, prompt following, and more through multiple implementations of novel algorithms for Automatic1111 WebUI.
GNU General Public License v3.0
120 stars 7 forks source link

TextCenGen #49

Open v0xie opened 1 month ago

v0xie commented 1 month ago

Very very WIP implementation of "TextCenGen: Attention-Guided Text-Centric Background Adaptation for Text-to-Image Generation" https://arxiv.org/abs/2404.11824

There are some really interesting techniques that we can use to manipulate attention.

v0xie commented 1 month ago

Results for a region mask that roughly covers the left half of the frame. Test with a prompt + seed that covers the whole left half of the image and low margin force. xyz_grid-0007-1716326508