Tsinghua's Stateful Harness: Executing Rules, Rendering Pixels for Consistent Interactive World Models
Tsinghua-IVG · hf · 2026-10-20
A Tsinghua team proposes Stateful Harness, tackling hidden-state consistency in interactive world models where the same frame plus the same action should yield different outcomes.
- 'Execute the rules, generate the pixels': separates rule-based world-state updates from a frozen video rendering model
- Hidden state persists across turns; deterministic rules resolve action outcomes into events that the frozen video model renders
- Video information is recorded back as the next visual anchor; introduces WorldState-Bench to evaluate hidden-state consistency
More from Multimodal
- MOSS Image with 8-step Turbo LoRA tested: 10x faster with minimal quality loss — Aggressive_Collar135 · 2026-10-20
- User turns their AI muse Maki into a punky ska song with Suno V6 — thederbiedone · 2026-10-20
- Grok Imagine parody short 'Wendy' shows off AI-made TV skits — Old_Put1306 · 2026-10-20
- AI-generated "Back to the Future" sequel clip goes viral for Halloween — mitchdeg · 2026-10-20
- Midjourney v8.2 style ref 1631327937 yields a dreamy pastel look — azed_ai · 2026-10-20
- Google Flow hosts hybrid production workshop pairing live-action with AI filmmaking workflows — lmoroney · 2026-10-20