SPARE-GS: Structural Parsimony and Resource Efficiency for 3D Gaussian Splatting
Computer Vision and Pattern Recognition
Summary
The authors address a problem in 3D Gaussian Splatting (3DGS), a method for creating detailed 3D images quickly, where too many building blocks (primitives) make training slower and the model less efficient. They treat the problem like a budget to optimally distribute these building blocks across the 3D space, ensuring resources are used where most needed. Their method, called SPARE-GS, dynamically adjusts the number of these primitives during training to better match the complexity of different areas. Experiments show that SPARE-GS cuts down on both the number of primitives and training time while improving image quality. It also helps with faster processing and better compression of the 3D models afterward.
Authors
Zhang Chen, Shuai Wan, Fuzheng Yang, Jiazhi Xia, Weiyao Lin, Junhui Hou
Abstract
3D Gaussian Splatting (3DGS) achieves high-fidelity novel view synthesis in real-time; however its training efficiency and representation compactness are hindered by excessive primitive proliferation. To address this challenge, we formulate the structural evolution of 3DGS as a global budget-constrained optimization problem and derive an optimality condition, which requires the marginal utility of structural resources to be balanced across spatial regions under a finite primitive budget. Based on this formulation, we propose SPARE-GS, a general plug-and-play framework that dynamically aligns the distribution of 3D Gaussian primitives with regional representational demand. SPARE-GS estimates capacity-normalized regional demand, assigns adaptive target quotas, and uses regional budget deviations to coordinate densification, pruning and adaptive termination toward a more balanced structural allocation. Extensive experiments across standard, accelerated, and structure-enhanced 3DGS pipelines demonstrate that SPARE-GS reduces the Gaussian count and training time by an average of 30.38% and 23.81%, respectively, while improving the average PSNR. Moreover, the resulting compact representations reduce downstream processing time and improve the rate-distortion performance of diverse compression and pruning methods, demonstrating the broad applicability of global structural budget regulation.