Video World Models · 2026-10-30
Fight World Model: Customizable Combat through Visual Interaction Proxies
Jiahua Li, Zixu John, Tom Zhong, Fuping Wu, Tianhao Xu, Ruicheng Zhang, Jun Zhou, Zunnan Xu
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
Fight World Model (FWM) separates rule-governed combat execution from video rendering using an executable engine and a deterministic compiler that outputs a spatially and temporally aligned visual interaction proxy. The proxy separately encodes attempted actions, brief hit reactions, and persistent health with identity binding, enabling rule edits (e.g., damage scaling, immunity, lifesteal, multi-stage skills) and appearance customization without retraining the renderer.
Why it matters
Video world models can generate plausible combat but may not keep hit reactions and health updates consistent with edited game rules. FWM claims a pipeline where rule edits change only the executed interaction records/proxy, while a shared trained renderer preserves action-reaction-state alignment and identity across side switches and appearance changes.
Method
- Executable engine resolves hit events and applies editable rule configurations; a deterministic compiler rasterizes aligned fighter states/events into a visual interaction proxy.
- Proxy encodes action/skill windows, transient reaction cues, and persistent health independently, with identity colors and identity binding for references across side switches.
- A pretrained video diffusion transformer conditions on the proxy plus character/arena appearance references; only video tokens are decoded; renderer parameters remain unchanged across rule edits.
Abstract (from arXiv)
A playable, programmable game world must maintain persistent state, enforce executable rules, and support extensible gameplay mechanics. Existing video world models can generate plausible combat, yet struggle to keep hit reactions and health updates consistent with game rules. We introduce Fight World Model (FWM), which separates rule-governed combat execution from video rendering. An executable engine maintains state and resolves interactions. A deterministic compiler encodes these outcomes and attempted actions in a spatially and temporally aligned interaction proxy. The proxy aligns action and skill windows, brief hit reactions, and persistent health by fighter identity and timing. These cues can be controlled separately to render interactions under modified rules. Identity-bound references preserve character appearance across side switches. We train FWM on FightWorld Corpus, spanning 52 skill effects, 187 character-appearance variants, and 15 arenas. Without retraining the renderer, FWM supports damage scaling, limited-hit immunity, lifesteal, and multi-stage skill mechanisms beyond the native game. It also allows character and arena customization. Experiments demonstrate the effectiveness of FWM, which outperforms existing game world models in action fidelity, hit reactions, health-state consistency, and character appearance preservation. Our project is available at https://fight-world-model.github.io.
Related papers
- Wizard: Manipulate Any World in Real Time via Agentic Control
- BoxWorld: Let World Engines Control and Video Generators Render
- PlayGen2: Learning Real-Time Playable Worlds with Explicit 3D State
- WorldOracle: Benchmarking Action-Conditioned World Models against Real-World Executions
- ROWBench: Do Video Models Render What the Program Specifies?
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-29 · Indexed by Pub-AI: 2026-10-30 UTC
Fight World Model: Customizable Combat through Visual Interaction Proxies
Preprint posted: 2026-10-29 · Original record ↗
- arxiv
- 2610.40607
Explicit source versions (1)
- v1 · 2026-10-29
- Preprint available
Source date: 2026-10-29 · First detected: 2026-10-30 UTC · Historical baseline
Source evidence (2)
- Official evidence ↗ · RSS guid=oai:arXiv.org:2610.40607v1; arxiv:announce_type=new; item pubDate (official metadata)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
A playable, programmable game world must maintain persistent state, enforce executable rules, and support extensible gameplay mechanics. Existing video world models can generate plausible combat, yet struggle to keep hit reactions and health updates consistent with game rules. We introduce Fight World Model (FWM), which separates rule-governed combat execution from video rendering. An executable engine maintains state and resolves interactions. A deterministic compiler encodes these outcomes and attempted actions in a spatially and temporally aligned interaction proxy. The proxy aligns action and skill windows, brief hit reactions, and persistent health by fighter identity and timing. These cues can be controlled separately to render interactions under modified rules. Identity-bound references preserve character appearance across side switches. We train FWM on FightWorld Corpus, spanning 52 skill effects, 187 character-appearance variants, and 15 arenas. Without retraining the renderer, FWM supports damage scaling, limited-hit immunity, lifesteal, and multi-stage skill mechanisms beyond the native game. It also allows character and arena customization. Experiments demonstrate the effectiveness of FWM, which outperforms existing game world models in action fidelity, hit reactions, health-state consistency, and character appearance preservation. Our project is available at https://fight-world-model.github.io.
