Fallout-trained H3 world model demo: WASD control plus LLM-reactive combat
-Ellary- · reddit · 2026-09-08
An early prototype uses MiniMax H3 trained on Fallout: Bakersfield footage as an interactive world model: 5 seconds of WASD/prompt user control followed by 5 seconds of LLM-driven reactions via vision-capable Gemma 4 12b, including dice-roll combat mechanics.
More from Multimodal
- MacPaw unveils GenRA, an LLM-driven framework animating rigged 3D models at ECCV — ducha_aiki · 2026-09-08
- ECCV 2026 MUCG workshop lineup: unified multimodal MLLMs and three new papers — mohitban47 · 2026-09-08
- Free ComfyUI workflow turns any photo into a full character sheet with Krea 2 — solomars3 · 2026-09-08
- Seedance 2.5 Skyline Rampage Video Prompt Shared in Full — techhalla · 2026-09-08
- ECCV 2026 AI4VA workshop reveals 14 accepted papers, Björn Ommer among keynote speakers — ducha_aiki · 2026-09-08
- Local MiniMax H3 generates 29 motion clips to build a talking AiTuber avatar — Hailuo_AI · 2026-09-08