DTAM (Dense Tracking and Mapping) is a real-time camera tracking and reconstruction system that relies on dense, pixel-level methods rather than feature extraction. As a handheld RGB camera moves over a static scene, DTAM estimates detailed textured depth maps at keyframes to produce a surface model with millions of vertices. The system uses a novel non-convex optimization framework to improve the quality of photometric data terms and minimize a global spatially regularized energy functional. Interleaved with this, the camera's 6DOF motion is tracked by aligning the entire image against the dense model at frame rate. DTAM achieves real-time performance using current commodity GPU hardware. The dense model enables superior tracking performance under rapid motion compared to feature-based methods and enhances real-time scene interaction in augmented reality applications. The paper discusses the method, including dense mapping, dense tracking, and model initialization, and presents evaluations and future work directions.DTAM (Dense Tracking and Mapping) is a real-time camera tracking and reconstruction system that relies on dense, pixel-level methods rather than feature extraction. As a handheld RGB camera moves over a static scene, DTAM estimates detailed textured depth maps at keyframes to produce a surface model with millions of vertices. The system uses a novel non-convex optimization framework to improve the quality of photometric data terms and minimize a global spatially regularized energy functional. Interleaved with this, the camera's 6DOF motion is tracked by aligning the entire image against the dense model at frame rate. DTAM achieves real-time performance using current commodity GPU hardware. The dense model enables superior tracking performance under rapid motion compared to feature-based methods and enhances real-time scene interaction in augmented reality applications. The paper discusses the method, including dense mapping, dense tracking, and model initialization, and presents evaluations and future work directions.