Search papers, labs, and topics across Lattice.
This paper introduces QuARC-GS, a novel quantization-aware framework for dynamic scene reconstruction that significantly reduces storage requirements while maintaining high-quality rendering and reconstruction speed. By leveraging a single canonical frame and compressing per-frame residuals through motion-aware anchor deformation and change-gated densification, QuARC-GS effectively addresses the challenges of online free-viewpoint video streaming. Experimental results show that this approach achieves up to 11脳 reduction in per-frame storage compared to existing methods, making it a promising solution for sustainable dynamic scene streaming.
Achieving up to 11脳 reduction in storage for dynamic scene streaming without sacrificing quality or speed could revolutionize online video applications.
3D scene representation techniques such as neural radiance fields (NeRFs) and Gaussian splatting have made substantial progress in novel view synthesis, achieving high-quality renderings from arbitrary view angles. More recently, such techniques have been extended to dynamic 3D scenes; however, achieving sustainable online free-viewpoint video (FVV) streaming remains challenging, especially for longer videos, due to significant storage demands of detailed scene representations and high reconstruction/rendering speed needs. To address these challenges, we propose Quantized Anchored Residual Coding Gaussian Streaming (QuARC-GS), a quantization-aware 4D scene optimization framework for online dynamic scene reconstruction that achieves ultra-high compression while maintaining reconstruction speed and quality. QuARC-GS represents a scene using a single canonical frame and highly compressed per-frame residuals. Specifically, we compress each residual through two complementary strategies targeting motion, appearance, and densification. We introduce quantization-aware anchor deformation, which suppresses insignificant motion updates while preserving meaningful deformations, maintaining reconstruction quality under low-storage streaming. Furthermore, we design a change-gated densification strategy that allocates new Gaussians only in regions exhibiting genuine temporal changes, effectively eliminating redundant appearance updates and reducing storage overhead. Extensive experiments on widely used datasets demonstrate that QuARC-GS enables competitive reconstruction quality and training speed while cutting per-frame storage by up to 11$\times$ compared to the state-of-the-art.
