MiniMax H3 Video Morphing: 4-Step Turbo LoRA Workflow Shared
Dohwar42 · reddit · 2026-08-09
A developer shared a ComfyUI workflow for achieving video morphing effects using the MiniMax H3 model. The pipeline generates a silent video through a two-pass process (first pass and second pass) combined at the end.
Technical Details:
- Acceleration: Utilizes a 4-step Turbo LoRA, requiring a custom node ComfyUI-MiniMax-H3-Turbo.
- Structure: Features subgraphs; the first frame of the second pass is extracted from the preceding video itself to ensure a smooth transition.
- Target Audience: The workflow is complex and not beginner-friendly, intended for ComfyUI veterans to learn and test multi-image morphing.
The author has open-sourced the complete workflow, including prompts, original assets, and test videos.
Related event: MiniMax H3 ComfyUI Acceleration Workflows Shared(2 posts)→
More from Multimodal
- Recreating a Real Car Crash Accurately Using Minimax Video Model — Puzzled_Dot2324 · 2026-08-09
- Minimax H3 Tutorial: Achieving Seamless Long Videos via Tail-Frame Continuation — rcscs · 2026-08-09
- Minimax H3 Flaw: Why Does Character Swap Always Alter the Background? — throwaway0204055 · 2026-08-09
- Treating Minimax H3 as a 'Dumb Cameraman' to Stitch a 2-Minute Dialogue Scene — niechta · 2026-08-09
- ByteDance's Seedance 2.5 Video Model Reveals Custom Prompt Markup Language — DavidmComfort · 2026-08-09
- Exploring Double Exposure Art with Midjourney — hewarsaber · 2026-08-09