Using Qwen3-VL-4B to auto-expand prompts for Minimax H3 video generation
CountFloyd_ · reddit · 2026-09-04
The author chains a Generate Text node with the qwen3vl4b CLIP model, feeding it Minimax's prompting guide to auto-expand bland base prompts into rich multi-shot prompts for H3 video generation — with better-than-expected results. Remaining issue: the small model consistently ignores the 5-second per-shot duration cap even when it's in the system prompt.
More from Multimodal
- Creator uses MiniMax H3 as a local product-CG renderer before hitting the API — Hailuo_AI · 2026-09-04
- MiniMax Design shows off Brutalist-style MV generated entirely by text with H3 — Hailuo_AI · 2026-09-04
- AI-generated Terminator 2 remake showcases video model creativity — alanskimp · 2026-09-04
- Workaround for full-head portraits in video models: add a 'floating invisible halo' to your prompt — robertwellesley · 2026-09-04
- Reframe render engine claims ~48x speedup over Octane and Arnold after months of kernel rewrites — D3VAUX · 2026-09-04
- Bringing museum exhibits to life with AI video generation — FallingKnifeFilms · 2026-09-04