Community LoRA armV5 for MOSS image editing ships v5 with 256-token reference budget
_akhaliq · x · 2026-10-26
The open MOSS-Image-Edit-HQ project released v5 (armV5), an instruction-based image-editing LoRA for OpenMOSS-Team/MOSS-Image-SFT that aims to beat Moss-Image-Edit, with every experiment and checkpoint published.
v5 adds full-scene rich captions (Fibo/Klein-style, folded from ConceptEdit metadata) and a 256-token reference budget (vs MOSS's 144) for better identity and fine detail, on top of masked loss, preservation contracts and text-edit data.
Weights ship in ai-toolkit format and load natively in ComfyUI v0.42.1+ (PR #17296), plus a plugin-format LoRA and a demo Space with Quality and 8-step Fast modes. All showcased edits are real instructions on real photos (50 steps, CFG 3). Unofficial community effort, not affiliated with OpenMOSS.
More from coding & agent
- Fewer, well-designed agent harnesses beat sprawling agent fleets, argues vixsheikh — annetgriffin · 2026-10-26
- ConformWeb: LLM-generated web apps pass only 11.7% of behavioral contract checks — algo_diver · 2026-10-26
- Agent memory explained in 12 minutes: working, procedural, semantic and episodic — _AustinCalvert_ · 2026-10-26
- Grok Bot v0.75.5 leak touts stored cards, email forwarding, 3x automation limits — mark_k · 2026-10-26
- Every time my agent breaks my UI, I add a rule to its skill — 10 design laws — _AustinCalvert_ · 2026-10-26
- Turn nouns into verbs: the 'Range' trick to unlock cross-domain AI output — petesena · 2026-10-26