AI summaryⓘ
The authors introduce QIRF, a new way to shrink 3D Gaussian Splatting models, which are used for detailed 3D scene rendering but can be very large and slow. Instead of just cutting down the number of Gaussian pieces individually, they look at how nearby Gaussians overlap and work together, using math inspired by quantum mechanics. This method picks the most important Gaussians while keeping the details, leading to a much smaller file size and faster rendering without losing image quality. Their tests show QIRF can compress these models over three times smaller and speed up rendering by about a third. Overall, the authors highlight that understanding overlaps in the Gaussians is key for better compression.
3D Gaussian Splattinganisotropic Gaussian primitivesfunction-space compressionnon-orthogonal basisGaussian overlap matrixradiance-response density matrixgeneralized eigendecompositionprimitive reductionPSNRMip-NeRF 360
Authors
Shizeng Jiang, Hao Zhang, Xuerui Ma, Ying Hu, Tao Zhang
Abstract
3D Gaussian Splatting (3DGS) achieves high-quality real-time rendering by representing a scene with a large collection of anisotropic Gaussian primitives. However, complex scenes often require millions of Gaussians, resulting in substantial storage and rendering costs. Existing compression methods mainly reduce redundancy through primitive-wise pruning, attribute quantization, clustering, or neural coding, while redundancy caused by strongly overlapping and non-orthogonal Gaussian basis functions remains largely unexplored. We present QIRF, a quantum-inspired non-orthogonal function-space compression method for 3D Gaussian Splatting. QIRF models neighboring Gaussian primitives as a local non-orthogonal basis and formulates primitive reduction as a subspace-aware selection problem. Specifically, an analytic Gaussian overlap matrix and a radiance-response density matrix are constructed to characterize functional redundancy and rendering relevance. Generalized eigendecomposition is then used to identify the dominant local subspace and select representative Gaussian primitives. An RRDM-based response model and detail-aware safeguarding further preserve visually important high-frequency structures under aggressive pruning. Experiments on 13 scenes from Mip-NeRF 360, Tanks and Temples, and Deep Blending show that QIRF reduces the Gaussian count and raw PLY storage by 71.7 percent on average, corresponding to approximately 3.54 times compression, while maintaining reconstruction quality comparable to 3DGS and achieving a marginal average PSNR improvement of 0.10 dB. QIRF also improves the average rendering speed over 3DGS by 34.3 percent. These results suggest that non-orthogonal function-space redundancy is an important yet underexplored source of representational redundancy in explicit Gaussian radiance fields.