ETCHR Decouples Image Editing from Understanding for Visual Chain-of-Thought Reasoning

ETCHR addresses two gaps in using image editors as reasoning assistants — language-side failure to map abstract questions to visual transforms, and generation-side edit correctness degradation — by decoupling a dedicated editing model from an understanding model.