Refmod nodes lose clothing details in video generation, plain image references work better
spacemidget75 · reddit · 2026-10-28
A Reddit user reports that popular ComfyUI refmod nodes underperform plain image references in video generation: clothing details went missing and faces drifted across several videos. Swapping back to direct clothing and face reference images restored accuracy on small details like belt buckles and buttons. They used Adudeguyman's refmod nodes (with bundled images in the safetensor) and ask whether others see the same regression. A useful counter-intuitive data point for anyone chasing character consistency: refmods aren't automatically better than raw references.
Related event: ComfyUI Test Finds refmod Node Loses Detail vs Plain Image Reference(2 posts)→
More from Multimodal
- Seedance 2.0 Mini demo: bubble-trick penguin AI wildlife clip — Burak_Antakya · 2026-10-28
- MOSS meets the ComfyUI team as MOSS-Image lands ComfyUI support — Physical_Hat4022 · 2026-10-28
- Blind test of 152 listeners: Suno matches human composer, but almost no one can tell which is AI — teropa · 2026-10-28
- Dalet Labs raises $7M pre-seed for generative personalized shopping media — eptwts · 2026-10-28
- What's trending in AI: Seedance 2.5, Kling 4.0, Suno, and world models next — Kyrannio · 2026-10-28
- AI-generated 70s funk MV 'After Two – TAREVI' — caponx · 2026-10-28