MiniMax H3 roundup: CrossView-Warp and four more LoRAs land
optimisticalish · reddit · 2026-09-21
A Reddit roundup of September 21 MiniMax H3 community releases:
- CrossView-Warp LoRA (Cseti): given a video and a camera offset (azimuth, elevation), generates the scene from the new viewpoint. It reads two videos — a depth-warp clip carrying geometry and the reference clip carrying identity/appearance. Trigger word crosswarp; nodes and workflow are open-sourced.
- Day-Night Slider LoRA: adjusts day/night look in video.
- Video Age Slider (civitai): controls how 'old' footage looks.
- Rough 2D Cartoon Illustration LoRA (prithivMLmods): trained on free-flow motion illustration styles; trigger rough 2D cartoon illustration; no demo yet.
- H3-Shadow-Negative-Nodes (Leon1000): adds negative prompting support to workflows.
ComfyUI Portable is now at 0.37.0; Kijai's proposed fix for the tile seam problem was well received but not yet merged into Portable or Nightly.
More from coding & agent
- Boubou: open-source CLI turns your laptop into a fully local voice AI hub — alexcovo_eth · 2026-11-12
- User automates entire video clipping workflow with Nous Research's Hermes agent — Teknium · 2026-11-12
- As models improve, writing good evals gets harder and more important — amankhan · 2026-11-12
- Prime agent ships with day-one multi-agent communication, runs in Minecraft unmodified — sarahcat21 · 2026-11-12
- Autonomous agent edits just 1,024 weights to lift Qwen3.5 0.8B from 7.4% to 37.4% on MATH-500 — VictorSanh · 2026-11-12
- Omarchy 4.6 lets you share local AI models across your tailnet, says DHH — alexcovo_eth · 2026-11-12