ComfyUI Ref2VA turns mushy? Only two LoRA/step combos produce clean output
Reasonable-Aide1259 · reddit · 2026-10-30
A Reddit user found that with the latest ComfyUI and Ref2VA, most recent model merges (including the Singularity × 10Eros PRISM merge) output noisy, mushy results — except two combos:
- A model without turbo merged in + DMD LoRA at 8 steps
- LightX2V turbo LoRA at 12 steps
- Most reliable so far: DaSiWa v2 + DMD LoRA at 8 steps
- Turbo-merged models are especially prone to never fully cleaning up
A full workflow screenshot is included (MiniMax H3 ref2va nodes, euler/simple sampler, steps=8, denoise 1.00). Useful reference for anyone hitting the same issue.
More from Multimodal
- Video generator adds sequence continuation, saving model/render/RNG state for cheap seed testing — pixlpa · 2026-10-30
- Pika H3 Flex Turbo LoRA cuts video sampling from 6 steps to 4 with a single drop-in file — Illustrious_Row_9971 · 2026-10-30
- ComfyUI Updater adds one-click Core and Frontend updates with rollback from the Settings dialog — Inuya5haSama · 2026-10-30
- Ideogram 4.5 image editing on fal now edits regions at original resolution without downscaling — JenniferHli · 2026-10-30
- After AI artist 'Young Soul' goes viral, 'Old Soul' emerges as a new AI musician persona — bennash · 2026-10-30
- Creating an anime scene with AI video: keeping the same location across shots is still hard — Alive_Winner_8440 · 2026-10-30