Tiny Sliders Steer MiniMax H3 Motion and Emotion Without Training a LoRA
MatlowAI · reddit · 2026-11-01
A Reddit user built tiny ComfyUI sliders that control motion, emotion and more in MiniMax H3 video generation without training a LoRA, demonstrated in a video comparing steered vs. unsteered outputs.
More from Multimodal
- A Halloween pumpkin generated with NB2.1: a face covered by dozens of hands — fofrAI · 2026-11-01
- AI-generated Dune visions: Arrakis eye and blue-within-blue rendered near photoreal — egeberkina · 2026-11-01
- Viking Funeral Ad Made with ElevenLabs Music 2.5 and Magnific Move — charis_ai · 2026-11-01
- LuCiD VRP: Open-Source ComfyUI Node Suite for Regional Prompting and In-Canvas Editing — EBEB__ · 2026-11-01
- Mage launches Magejutsu to recast any characters into any video, in three tiers — itsmnjn · 2026-11-01
- Fully local Halloween music video: 19 scenes via ComfyUI Video Builder with MiniMax H3 — Cheap_Credit_3957 · 2026-11-01