Using Signed Distance Fields to Achieve Temporal Compression of Mesh-Based Volumetric Video

Volumetric Video is an exciting medium that enables visualization of 3D time-varying data, often as sequence of Point Clouds or Textured Meshes. Despite over a decade of focus, there still exist several problems that make it challenging to employ. This paper focuses on one aspect of deploying Volumetric Video — data compression. Reducing the data storage size is necessary for streaming applications due to the large size of sequences consisting of even a few seconds of footage. This work explores the use of temporal compression of time-varying textured meshes using Signed Distance Fields and compare with the state-of-the art techniques. The goal is to produce a novel compression method to maximize a size-to-error compression ratio while maintaining a decompression speed of 30 frames per second. The results reflect that as file sizes decrease, the proposed method’s error increases at a slower rate than the state-of-the-art.