MiniMax Video Workflow Tested: 1-Minute Clip Takes 5 Hours on RTX 4090
KeanuRave100 · reddit · 2026-08-09
A developer deeply tested the MiniMax reference workflow in ComfyUI on a 4090 + 128GB RAM setup, sharing insights from creating a Pathfinder-style animated short.
- Audio References: The model is highly sensitive to audio reference ordering. Unused character audio references can consistently break the first line of dialogue, even if explicitly told to use another. Removing unused references is the current workaround.
- Generation Length & Compute: Longer generations maintain visual continuity better but are computationally heavy. A 60-second clip takes 5 hours, 25 seconds takes 24 minutes, and 12 seconds takes 5 minutes. The author recommends generating shorter clips to reduce prompt iteration costs and editing externally.
- Visual Cohesion: Realistic reference images yield better results than highly stylized ones, and camera orientation descriptions must be extremely precise.
More from Multimodal
- Meituan LongCat's WorldCast benchmark: GPT-Image-2.5 tops at 72.8, open-source lags — meituan-longcat · 2026-10-16
- Three resources to get up to speed with FLUX 3 — ariG23498 · 2026-10-16
- Watch: Claude Opus 5.5 builds a Tesla Roadster from a single prompt — TAbrodi · 2026-10-16
- Midjourney v8.2 turns out striking indigo ink-wash art with sref 1848359068 — azed_ai · 2026-10-16
- "AI television from another dimension": robot grooms a chihuahua on air — BotLordBotflix · 2026-10-16
- 108M-parameter text-to-image denoiser Surjo-Image-Preview lands on Hugging Face — Consistent-Rule-3720 · 2026-10-16