T-Code: Simple Temporal Latent Code for Efficient Dynamic View Synthesis
This work addresses the problem of efficient dynamic view synthesis for computer vision applications, offering incremental improvements in storage and speed over existing methods.
The paper tackles efficient dynamic view synthesis by introducing T-Code, a decoupled latent code for the time dimension, which reduces computation and storage costs. Experiments show that HybridNGP achieves high fidelity with top processing speed and less storage, and DNGP-T attains state-of-the-art quality and high training speed for monocular reconstruction.
Novel view synthesis for dynamic scenes is one of the spotlights in computer vision. The key to efficient dynamic view synthesis is to find a compact representation to store the information across time. Though existing methods achieve fast dynamic view synthesis by tensor decomposition or hash grid feature concatenation, their mixed representations ignore the structural difference between time domain and spatial domain, resulting in sub-optimal computation and storage cost. This paper presents T-Code, the efficient decoupled latent code for the time dimension only. The decomposed feature design enables customizing modules to cater for different scenarios with individual specialty and yielding desired results at lower cost. Based on T-Code, we propose our highly compact hybrid neural graphics primitives (HybridNGP) for multi-camera setting and deformation neural graphics primitives with T-Code (DNGP-T) for monocular scenario. Experiments show that HybridNGP delivers high fidelity results at top processing speed with much less storage consumption, while DNGP-T achieves state-of-the-art quality and high training speed for monocular reconstruction.