Marionette: Predicting World States, Rendering Geometry, Painting Appearance

Hugging Face Daily Papers Papers
world-model interactive-games video-diffusion 3d-graphics controllable-synthesis

Summary

Marionette introduces a world model that explicitly predicts 3D articulated states for interactive games, decouples geometry rendering with a zero-parameter renderer, and uses diffusion for appearance synthesis, improving controllability and long-horizon consistency.

Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose, geometry, and occlusion to be implicitly maintained by the same generative sequence. Over long horizons, errors in these latent world properties accumulate, making consistency and controllability fragile. We explicitly model the evolving world state, delegate exact geometric computation to a fixed, zero-parameter renderer, and leave the neural model to synthesize appearance. We instantiate this idea as Marionette, a world model for interactive games with articulated characters. First, a two-stage autoregressive dynamics model predicts an explicit and interpretable 276-dimensional 3D world state comprising multi-entity articulated skeletons, metric root trajectories, and rotations. Second, a zero-parameter graphics bridge converts the predicted state into pose-control videos, computing world-space geometry and occlusion in closed form. Third, a control-conditioned video-diffusion observation model synthesizes photorealistic RGB observations from the resulting structured controls. Our experiments establish two properties of Marionette. First, the predicted world state is directly controllable. Forcing a mismatched action stream changes root-aligned joint error by 31% across 48 held-out segments. Second, long-horizon behaviour is determined in the state, and can be repaired there. Left free, the two generated characters drift to 21.2 m apart (recorded sessions stay near 5 m) and a third of frames show ground penetration. Two rules imposed on the explicit state, a terrain collider and a separation cap, cut penetration by 66% and keep the pair engaged, with no change to the observation model. Routing appearance through the predicted state costs no fidelity we can detect, at an FVD of 831 against 799 for recorded pose.
Original Article
View Cached Full Text

Cached at: 08/17/26, 03:45 AM

Paper page - Marionette: Predicting World States, Rendering Geometry, Painting Appearance

Source: https://huggingface.co/papers/2608.14530

Abstract

Marionette predicts explicit 3D articulated world states for interactive games, uses a fixed renderer for geometry, and synthesizes video via diffusion, enabling direct state-level control and long-horizon consistency repair.

Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose, geometry, and occlusion to be implicitly maintained by the same generative sequence. Over long horizons, errors in these latent world properties accumulate, making consistency and controllability fragile. We explicitly model the evolving world state, delegate exact geometric computation to a fixed,zero-parameter renderer, and leave the neural model to synthesize appearance. We instantiate this idea as Marionette, a world model for interactive games with articulated characters. First, a two-stageautoregressive dynamics modelpredicts an explicit and interpretable 276-dimensional3D world statecomprising multi-entityarticulated skeletons, metric root trajectories, and rotations. Second, a zero-parameter graphics bridge converts the predicted state into pose-control videos, computing world-space geometry and occlusion in closed form. Third, acontrol-conditionedvideo-diffusion observation modelsynthesizes photorealistic RGB observations from the resulting structured controls. Our experiments establish two properties of Marionette. First, the predicted world state is directly controllable. Forcing a mismatched action stream changes root-aligned joint error by 31% across 48 held-out segments. Second, long-horizon behaviour is determined in the state, and can be repaired there. Left free, the two generated characters drift to 21.2 m apart (recorded sessions stay near 5 m) and a third of frames show ground penetration. Two rules imposed on the explicit state, a terrain collider and a separation cap, cut penetration by 66% and keep the pair engaged, with no change to the observation model. Routing appearance through the predicted state costs no fidelity we can detect, at anFVDof 831 against 799 for recorded pose.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2608\.14530

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.14530 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.14530 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.14530 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

CG-World: A Large-Scale World-State Dataset and Protocol for World Models

arXiv cs.AI

CG-World is a large-scale world-state dataset and protocol derived from industrial computer graphics pipelines, explicitly recording multimodal world states, interventions, and counterfactual branches to support world model research. It demonstrates improvements in geometry-conditioned video generation, action prediction, and closed-loop transfer of vision-language-action policies.

Programmable World Model

Hugging Face Daily Papers

The Programmable World Model introduces a framework that decouples world-state evolution from visual generation, enabling persistent and controllable environments through executable programs and 3D bounding boxes.