Running Minimax H3 Video Generation on 16GB VRAM: Workflow & Prompts
Fabulous-Snow4366 · reddit · 2026-08-10
This post demonstrates a practical case of generating videos locally using the Minimax H3 model, featuring a speaking portrait of Hollywood actor Brad Pitt.\n\n- Hardware: 16GB VRAM (RTX 5060ti) + 48GB DDR4 RAM.\n- Workflow & Models: Utilized minimaxh3fl2vaprunedint8convtot combined with qwen3vl32b for prompt processing, and loaded the minimaxh3turbov4600 LoRA.\n- Generation Parameters: 0.6 MP resolution, 9:16 aspect ratio, 5 seconds duration, using Euler - beta sampler, 8 steps, taking about 2 minutes to render a single clip.\n- Post-processing: Final cut was made in DaVinci Resolve Studio and upscaled to 1080p using its built-in AI Upscaler.
Related event: MiniMax H3 Local Tests: 24-Min 2K Video on 16GB VRAM(12 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