vivalapanda / conceptual-image-to-image-1.5

Conceptual image-to-image model for Stable Diffusion 1.5

  • Public
  • 1K runs
  • GitHub

This model lets you condition on image content when performing image-to-image with Stable Diffusion.

Combine both structural and conceptual guidance from the initial image for the best results!

A model for SD 2.0 is also available here: https://replicate.com/vivalapanda/conceptual-image-to-image However, our initial test prompts generally did better with SD 1.5 - so you might need to do some more tinkering to figure out where SD 2.0 shines!