Implicit Surface Codec

Tao Jin1, Shengxi Wu1, Tianshu Huang1, Mallesham Dasari2, Srinivasan Seshan1, Anthony Rowe1
1Carnegie Mellon University, 2Northeastern University
Paper volumetric_kit_gfx volumetric_kit_recon
Overview of the implicit-surface predictive codec: captured RGB-D, mesh and point cloud input is fused into a sparse TSDF volume, compressed with block-wise DCT transform coding and motion-predictive coding, then decoded back into meshes and point clouds. A bottom row compares reconstruction quality and bitrate against MPEG-GPCC, Draco, Tang et al. 2018, MPEG-VPCC and MPEG-VDMC.

Overview of our implicit-surface predictive codec for captured dynamic geometry. Our method targets 3D data produced by 3D capture/reconstruction pipelines, including RGB-D capture sources, captured meshes and point clouds. These inputs are represented as a sparse TSDF volume, which serves as the coding domain. We compress the dynamic TSDF sequence using block-wise transform coding and motion-predictive coding across frames to exploit spatial and temporal redundancy. The decoder reconstructs the TSDF volume and extracts meshes and point clouds through surface extraction algorithms.

Abstract

The rapid adoption of volumetric capture technologies has created a pressing need for efficient storage and streaming of dynamic 3D content. Unfortunately, current compression standards often treat dynamic sequences as independent frames or rely on computationally expensive non-rigid registration, making them unsuitable for real-time applications or large-scale environments. In this paper, we present a novel end-to-end compression framework for dynamic Truncated Signed Distance Field volumes derived from captured 3D content, leveraging a representation that is temporally stable, easily parallelizable, and already widely used in scene reconstruction and volumetric fusion pipelines. We then adapt classic 2D video coding paradigms such as spatial coding via Discrete Cosine Transform and temporal coding using a real-time motion compensation pipeline to provide robust, real-time, and training-free encoding and decoding for 3D content. Extensive evaluations on human performance captures demonstrate that our codec achieves ~35% bitrate savings at equal distortion while operating in real time at 30 FPS, while stronger temporal coherence in large-scale synthetic environments yields up to 12× bitrate reduction at equal distortion.

BibTeX

@article{jin2026implicit,
  title={Implicit Surface Compression—with Good Old Discrete Cosine Transform and Motion Compensation},
  author={Jin, Tao and Wu, Shengxi and Huang, Tianshu and Dasari, Mallesham and Seshan, Srinivasan and Rowe, Anthony},
  journal={ACM Transactions on Graphics (TOG)},
  volume={45},
  number={4},
  pages={1--14},
  year={2026},
  publisher={ACM New York, NY, USA}
}

Acknowledgments

This work was supported by Bosch Corporate Research and in part by the National Science Foundation under Award IIS-2544317. We thank the anonymous reviewers for their constructive feedback. We are also grateful to Ioannis Gkioulekas for an insightful discussion, Yoshiki Takashima for suggesting the use of the Blender Film sequences to demonstrate the scalability of our codec, and Ankush Jain for saving the day with last-minute help in generating the .bbl file before the submission deadline.

This project page is built on the Nerfies template by Keunhong Park et al. Many thanks to the authors for releasing their source code.