Back to Model
M

MOSS-Image-Edit-HQ

Model Oct 27, 2026

An unofficial instruction-editing LoRA for MOSS-Image-SFT with public checkpoints, comparisons and ComfyUI workflows.

Visit primary link 2 sources attached

What happened lately

MOSS-Image-Edit-HQ timeline

MOSS-Image-Edit-HQ is a community adapter for the MOSS-Image-SFT image model. It targets instruction-driven edits and publishes experimental checkpoints, training recipes and author-generated comparisons. It is not an official OpenMOSS release. The adapter requires its base model and supporting pipeline components; downloading a LoRA alone does not provide a complete image-generation system.

The English card distinguishes native ComfyUI ai-toolkit-format loading from the Chenkin-x plugin’s moss-edit-lora format. Each route uses different files and model folders. The maintainer recommends armV5, retains armV6 as an experiment, and documents a separate optional DMD speed adapter. Hardware guidance and visual quality judgments are the author’s reports, not independently reproduced results. The repository was created October 22, 2026; later artifact uploads and recommendation updates do not reset that origin date.