6-step Turbo LoRA merge speeds up MiniMax H3 video gen on low-spec GPUs
hard_gravy_2 · reddit · 2026-09-08
A new 6-step Turbo LoRA merge for MiniMax H3 is out on Hugging Face and Civitai, aimed at faster video generation on modest hardware.
- Recipe: an over-strength DARE merge of lightx2v and larryvrh turbo LoRAs plus 0.4 strength of JonXL's H3 photorealism LoRA, built with ComfyUI LoRA Optimizer nodes.
- Results: audio/video quality holds up better at 6 steps than with single turbo LoRAs; on a 16GB RTX A4000 (Ampere) the author gets 330s per 10s clip at 0.5MP with the SLA attention node.
- Settings: Euler sampler, Simple scheduler, 6 steps, shift unset; ComfyUI workflow included (h3fast.).
- Caveats: the original development workflows are lost so the merge isn't fully reproducible; untested with ref2v/hybrid models and no support offered.
More from Multimodal
- ostris pre-training "Primal VAE": 1-channel latents match modern 8x VAEs via patching — ostrisai · 2026-10-26
- Slopcore: AI-generated meme films turn low-fi slop aesthetics into a subculture — Yamapama · 2026-10-26
- TischEins, a large Midjourney SREF library, relaunches with new design after security outage — tisch_eins · 2026-10-26
- MOSS-Image-SFT 4-step distillation tops 16.5K downloads on Hugging Face — _akhaliq · 2026-10-26
- Claude Code can now edit videos: a 10-prompt workflow from raw footage to final cut — TawohAwa · 2026-10-26
- One-sentence 3D worlds turn cinematic with CineMap and TapNow — gizakdag · 2026-10-26