← Back to the wire

D4RT: Unified, Fast 4D Scene Reconstruction & Tracking — Google DeepMind

AnnouncementResearchJan 22, 2026

Google DeepMind introduced D4RT, a unified AI model for 4D scene reconstruction and tracking across space and time. The model addresses the challenge of recovering volumetric 3D worlds in motion from flat 2D video projections. D4RT tracks every pixel of every object as it moves through three spatial dimensions and time, while disentangling object motion from camera motion. Traditionally, this required computationally intensive processes or multiple specialized AI models for depth, movement, and camera angles.

Receipt № 5761 source · awaiting confirmation ◐

Evidence

1source· awaiting independent confirmation

No score is assigned. Sources and their independence are shown in the citation chain below.

Citation chain · 1 source

AlphaEvolveModelGeminiModelNano BananaCompanyAlphaGoModelWeatherNextModelGemini OmniModelGemmaModelAlphaFoldModelGenie 3ModelGemini RoboticsModelAlphaEarthModelSIMA 2ModelVeoModelLyriaModelGemini AudioModelImagenModelGoogle DeepMindCompany
Canonical: https://deepmind.google/blog/d4rt-teaching-ai-to-see-the-world-in-four-dimensions/