Free Open-Source Cinematic LoRA Takes MiniMax H3 From 1K to 4K in ComfyUI
jugernaut126 · reddit · 2026-09-11
A free, open-source Cinematic LoRA for the MiniMax H3 video model has dropped, targeting the plastic "AI look" with better grain, contrast, color, and shadow roll-off. Works for both text-to-video and image-to-video.
The author's real test: using the workflow's latent upscale path, they pushed 1K output to 4K without blowing up VRAM, with highlights and skin tones looking graded rather than smooth-AI.
Quick setup:
- Install nodes into customnodes (restart ComfyUI): ComfyUI-Easy-Use, ComfyUI-KJNodes, comfyui-minimax-h3-blockcache-T8, ComfyUI-H3LatentUpscale-jingchen573, ComfyUI-Custom-Scripts, rgthree-comfy
- Drop the LoRA (Quark cloud link) into models/loras
- Load the shared workflow; check model/LoRA paths, Block Cache params, and Latent Upscale after the sampler
- Tweak prompts and render
Resources include the LoRA download and a detailed Chinese guide on Zhihu. Great for short films, trailers, and mood shots.
More from Multimodal
- 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
- No Camera Needed: Creator Builds an Entire World in CineMap and Directs It Into Cinema — gizakdag · 2026-10-26