H3 REF2VID Demo: Video Generation Without Masking
AthleteEducational63 · reddit · 2026-08-30
A demo of the H3 model's REF2VID capability, showing video generation from a reference image without the use of masking techniques.
More from Multimodal
- Midjourney releases version 8.2 update — azed_ai · 2026-08-30
- GPT-6 generates stunning Pagoda Voxel art — ChrisGPT · 2026-08-30
- Seedance 2.5 outperforms Grok 1.5 in art direction test — ChrisGPT · 2026-08-30
- Dev builds custom ComfyUI node for continuous, length-unlimited MiniMax H3 video generation — rynaleopard · 2026-08-30
- HR Endless Sampler: open-source ComfyUI node renders any-length H3 videos on 16GB VRAM — rhradec · 2026-08-30
- MiniMax-H3-Longvideos hits Hugging Face trending: long-form text-to-video with synced audio — Smite79 · 2026-08-30