ComfyUI help wanted: MiniMax H3 ref2video with checkpoint and LoRA loaders
aanabobby31 · reddit · 2026-10-05
A Redditor is trying to build a MiniMax H3 ref2video workflow in ComfyUI with two reference images, swapping the default diffusion model loader for a checkpoint loader plus LoRA loaders, but can't wire the nodes correctly and asks for a working workflow. No answers yet, but it highlights a common ComfyUI loader-integration pain point.
More from Multimodal
- Creator releases radio edit remastered with Suno and Perplexity Computer — inductionheads · 2026-10-05
- Crowd-sourced Westminster photos dropped at exact spots using Google 3D Tiles, no scanning — MonaJalal_ · 2026-10-05
- HeyGen launches Video built on MiniMax H3, generating 720p 15-second clips — HeyGen · 2026-10-05
- Opus vs Astra 'jelly world' islands: better water costs $15.54 vs $10.51 — i_dg23 · 2026-10-05
- Image model A/B tests are back, hinting at a new version in hidden testing — flowersslop · 2026-10-05
- 50k DALLE-2 images used to fine-tune SD1.5, recreating the retro AI art look — cephaloform · 2026-10-05