Generating 30s Videos from a Single Image: MiniMax H3 Local Test & Prompt Breakdown
Tight_Organization54 · reddit · 2026-08-07
A developer successfully used the MiniMax H3 model to generate a coherent 30-second video locally, using only a single static image of John Wick.
- Hardware & Settings: Running on an RTX 4070 Ti Super (16GB VRAM) and 32GB RAM, each clip took about 6 minutes. The generation used 6 steps, 0.4 MP resolution, and a Turbo LoRA for acceleration.
- Workflow: The author skipped RTX upscaling to avoid facial tearing, directly using the built-in ComfyUI Ref2V workflow.
- Prompt Engineering: The author leveraged Grok to automatically generate a structured, lengthy prompt detailing 6 sequential actions over 30 seconds (waking up, drinking coffee, playing soccer, shooting, brushing teeth, sleeping), successfully maintaining character consistency.
Related event: Developers Test MiniMax H3 for Local 30s Video Generation(2 posts)→
More from Multimodal
- Opus 5.5 turns heads with surprisingly polished vertical video demo — FinanceYF5 · 2026-10-16
- 'Celestial Ocean Reverie' prompt template blends starry skies with dreamlike seascapes — LudovicCreator · 2026-10-16
- 'Thicc Lil' Red': an AI-made musical meme reimagining Little Red Riding Hood — bennash · 2026-10-16
- CoWorld Reasoner uses an evolving video world model as a visual reasoning engine — ModalityDance · 2026-10-16
- PixelLab tackles ComfyUI pain points: model dedup, image triage, workflow repair — SmartRick137 · 2026-10-16
- Everything is instantly a playable game: browser Roadster Rally demo — bennash · 2026-10-16