ComfyUI VDN-H3 24GB v1.1.0 fixes adapter mapping, boosts prompt following at ~4% speed cost
NoMouse9610 · reddit · 2026-09-11
The author released v1.1.0 of the ComfyUI-VDN-H3 24GB node after spotting an issue in the released adapter mapping. Fixing it noticeably improved prompt following, alongside better memory handling for longer clips.
Key changes:
- Restored the complete token-refiner adapter mapping
- Improved temporary memory handling for longer generations
- Fixed CUDA stream lifetime handling for prefetched weights
- Kept AutoMemory / AutoLongCache behavior from v1.0.0
Tested on an RTX 3090 24GB with 5s/10s/15s/20s clips at 0.4MP plus 10s at 0.8MP; character/style LoRAs work normally. The fix costs about 4% sampling speed, which the author considers worth it. A same-prompt/same-seed comparison video is attached, with GitHub and HuggingFace checkpoint links.
Related event: ComfyUI VDN-H3 24GB Plugin v1.1.0 Boosts Prompt Following(2 posts)→
More from Multimodal
- Adobe Firefly video editor walkthrough: one-sentence Quick Cut turns 32s into 20s — LudovicCreator · 2026-10-26
- Hi3D V1.0 open-sourced with 2M training assets and 90% slicing success rate — kimmonismus · 2026-10-26
- Claude Can Replace Your YouTube Thumbnail Designer in 10 Minutes — thedankoe · 2026-10-26
- Multimodal AI Is Not Useless: A Practitioner Pushes Back on Growing Skepticism — thoma_gu · 2026-10-26
- One prompt, full Blender animation: dev says he never touched Blender once — iamfakhrealam · 2026-10-26
- Dense instance segmentation via Muse Wonder + SAM 3.1 on Meta's Model API — nikhilaravi · 2026-10-26