60 seconds of AI video fully local on an RTX 3050 6GB with MiniMax H3
No_Cow26 · reddit · 2026-11-11
A Reddit user shares a complete pipeline for generating 60 seconds of coherent video fully local on an RTX 3050 (6GB VRAM):
- MiniMax H3 reference-to-video via ComfyUI's native H3 nodes, int8 quantized model + Qwen3-VL text encoder
- Two character reference images reused across shots; prompts follow the official six-section reference format (subject definitions, summary, retention analysis, detailed description, soundscape, music); no negative prompts (H3 doesn't use them)
- Six shots of 260 frames (10.8s) at 864×480, 25 steps, resmultistep sampler + simple scheduler
- Continuity: each shot starts from the last 22 frames + audio of the previous via the MiniMaxH3AddGuide node at frame 0
- A small Python script chains shots through ComfyUI's API, trims overlaps, and encodes one 60-second MP4 with audio, unattended
Known issues: soft detail and some flicker; still tuning.
Related event: User Generates 60-Second AI Video Fully Locally on RTX 3050 6GB(2 posts)→
More from Multimodal
- AI-Generated Series Episode 1: The Anomaly Deleted Her Parents and Is Rewriting the City — Crazy-Baker-8828 · 2026-11-12
- Runway launches Praxis-1, a 27B MoE world action model, with a 73ms Fast variant — ChongZzZhang · 2026-11-12
- MiniMax H3 Image generates 3D game scenes 3x faster than GPT Image — HeyToha · 2026-11-12
- Ditto launches swap-anything video tool to replace any video's main character — yoobinray · 2026-11-12
- Fable 5.2 impresses with pixel art gaussian splats, posts go viral — Flomerboy · 2026-11-12
- Reddit user shares AI-animated 'Bubblegum Adventures' Halloween episode — Zard-Oz · 2026-11-12