Collected sources and patterns will appear here. Add from search or the patterns library.
(LatentTensor, Mask, DiffusionModel) -> LatentTensor
Enforce mask consistency during diffusion sampling by applying multiple iterative correction passes ('thinking steps') to the latents before final denoising.
Problem it solves
Standard training-free inpainting methods rely on simple blending or partial denoising, which causes boundary artifacts and limits content generation quality.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.