VGGT-World: Transforming VGGT into an Autoregressive Geometry World Model

TL;DR

VGGT-World introduces a novel geometry-focused world model that predicts the evolution of frozen GFM features, outperforming baselines in depth forecasting with high efficiency and fewer parameters.

Contribution

It repurposes frozen GFM features as a predictive state and develops a new training approach to improve high-dimensional feature forecasting.

Findings

Outperforms strong baselines in depth forecasting.

Runs 3.6-5 times faster than existing models.

Uses only 0.43B trainable parameters.

Abstract

World models that forecast scene evolution by generating future video frames devote the bulk of their capacity to photometric details, yet the resulting predictions often remain geometrically inconsistent. We present VGGT-World, a geometry world model that side-steps video generation entirely and instead forecasts the temporal evolution of frozen geometry-foundation-model (GFM) features. Concretely, we repurpose the latent tokens of a frozen VGGT as the world state and train a lightweight temporal flow transformer to autoregressively predict their future trajectory. Two technical challenges arise in this high-dimensional (d=1024) feature space: (i) standard velocity-prediction flow matching collapses, and (ii) autoregressive rollout suffers from compounding exposure bias. We address the first with a clean-target (z-prediction) parameterization that yields a substantially higher signal-to-noise ratio, and the second with a two-stage latent flow-forcing curriculum that progressively conditions the model on its own partially denoised rollouts. Experiments on KITTI, Cityscapes, and TartanAir demonstrate that VGGT-World significantly outperforms the strongest baselines in depth forecasting while running 3.6-5 times faster with only 0.43B trainable parameters, establishing frozen GFM features as an effective and efficient predictive state for 3D world modeling.