CompGS: Efficient 3D Scene Representation via Compressed Gaussian Splatting This paper proposes Compressed Gaussian Splatting (CompGS), a novel method for efficient 3D scene representation. CompGS leverages compact Gaussian primitives to achieve significantly reduced data size while maintaining high rendering quality. The method introduces a hybrid primitive structure that captures predictive relationships between Gaussian primitives, enabling efficient representation of 3D scenes. A small set of anchor primitives is used to predict the majority of primitives, which are represented in compact residual forms. Additionally, a rate-constrained optimization scheme is developed to further improve the compactness of the primitives by minimizing the rate-distortion loss. CompGS outperforms existing methods in terms of compression efficiency, achieving a compression ratio of up to 110× on prevalent datasets. The method achieves superior compactness in 3D scene representation without compromising model accuracy and rendering quality. Experimental results show that CompGS achieves significant improvements in compression efficiency compared to existing methods, with a notable reduction in bitstream size and bitrate consumption. The method is evaluated on three view synthesis datasets: Tanks&Templates, Deep Blending, and Mip-NeRF 360. Results demonstrate that CompGS achieves high-quality rendering with significantly reduced data size, with compression ratios ranging from 45.25× to 110.45×. The method also shows superior performance in terms of rendering quality and bitstream size compared to existing compression methods. The proposed method is implemented using PyTorch and CompressAI libraries, and is optimized using Adam optimizer with cosine annealing learning rate decay. The method is also evaluated in terms of computational complexity, showing that it is efficient and practical for real-world applications. The results demonstrate that CompGS achieves a remarkable reduction in data size and bitrate consumption while maintaining high rendering quality, making it a promising approach for efficient 3D scene representation.CompGS: Efficient 3D Scene Representation via Compressed Gaussian Splatting This paper proposes Compressed Gaussian Splatting (CompGS), a novel method for efficient 3D scene representation. CompGS leverages compact Gaussian primitives to achieve significantly reduced data size while maintaining high rendering quality. The method introduces a hybrid primitive structure that captures predictive relationships between Gaussian primitives, enabling efficient representation of 3D scenes. A small set of anchor primitives is used to predict the majority of primitives, which are represented in compact residual forms. Additionally, a rate-constrained optimization scheme is developed to further improve the compactness of the primitives by minimizing the rate-distortion loss. CompGS outperforms existing methods in terms of compression efficiency, achieving a compression ratio of up to 110× on prevalent datasets. The method achieves superior compactness in 3D scene representation without compromising model accuracy and rendering quality. Experimental results show that CompGS achieves significant improvements in compression efficiency compared to existing methods, with a notable reduction in bitstream size and bitrate consumption. The method is evaluated on three view synthesis datasets: Tanks&Templates, Deep Blending, and Mip-NeRF 360. Results demonstrate that CompGS achieves high-quality rendering with significantly reduced data size, with compression ratios ranging from 45.25× to 110.45×. The method also shows superior performance in terms of rendering quality and bitstream size compared to existing compression methods. The proposed method is implemented using PyTorch and CompressAI libraries, and is optimized using Adam optimizer with cosine annealing learning rate decay. The method is also evaluated in terms of computational complexity, showing that it is efficient and practical for real-world applications. The results demonstrate that CompGS achieves a remarkable reduction in data size and bitrate consumption while maintaining high rendering quality, making it a promising approach for efficient 3D scene representation.