UGG: Unified Generative Geometry Model

This paper introduces UGG, a unified framework for understanding 4D scenes that jointly captures geometry, appearance, and dynamics from partial observations. By assigning different noise levels across modalities and time, UGG formulates visual generation and geometric prediction as conditional completion within a single generative model. This approach allows the model to learn from diverse datasets with heterogeneous annotations and enables inference-time search over multiple plausible scene hypotheses using geometric evidence from observed RGB frames. The authors demonstrate UGG's effectiveness over non-unified baselines in camera-controlled video generation, depth estimation, and camera pose prediction, including in sparse observation settings. Code and visualizations are available at https://ugg-model.github.io/.

Generated from attributed reports · Updated 1 hours ago

There is not enough continuous observation data to show a trend.