Training MiniMax H3 on 341 stills alone: does motion really break? A before/after test
shootthesound · reddit · 2026-09-19
A controlled before/after test shows a LoRA trained on just 341 stills doesn't wreck MiniMax H3's motion quality: same prompt, seed and settings, fully T2V, with the workflow available via open-source Fizgig v6.2.0.
More from Multimodal
- Free course covers the 4 hard problems of production generative media agents — Al_Grigor · 2026-11-04
- PixelBench: testing multimodal models pixel-by-pixel instead of with QA — bdsqlsz · 2026-11-04
- Kitten-TTS 2 offers free voice cloning on laptops, no GPU needed — solyarisoftware · 2026-11-04
- "We'll Take Your RAM": a fully local-hardware cyberpunk meme from 2030 San Francisco — Wonderful-Yak-3348 · 2026-11-04
- Openfilm: open-source AI video agent turns coding agents into video makers — solyarisoftware · 2026-11-04
- Microsoft reopens VibeVoice: 10-second voice cloning, 90-minute local TTS — solyarisoftware · 2026-11-04