This report, titled "A survey on 3D Gaussian Splatting Compression Methods," provides a comprehensive overview of techniques aimed at optimizing the storage and memory usage of 3D Gaussian Splatting (3DGS) scenes. 3DGS is a state-of-the-art technique for real-time radiance field rendering, offering high-quality and fast performance. However, its extensive use of Gaussians leads to significant storage demands, making it impractical for mobile devices and headsets. To address this, the report categorizes existing methods into two main categories: compression and compaction. **Compression** focuses on reducing file size by techniques such as vector quantization (VQ), which clusters similar Gaussian attributes to reduce redundancy. Structured representations, like octrees, hash-grids, and self-organizing grids, also help by organizing Gaussians spatially and reducing the need to store individual attributes independently. **Compaction** aims to minimize the number of Gaussians by techniques like Adaptive Density Control (ADC), which dynamically adds or removes Gaussians based on their contribution to the scene. Other methods include attribute pruning, which reduces the number of spherical harmonics (SH) coefficients, and various structured representations that exploit spatial relationships among Gaussians. The report discusses the mathematical foundations, implementation details, and key design choices of these methods, providing a detailed comparison of their effectiveness using common datasets. It also highlights the advantages and disadvantages of each approach, establishing a consistent framework for evaluating 3DGS compression techniques. The goal is to advance the practicality of 3DGS in computationally constrained environments, making it more widely applicable in fields such as gaming and film.This report, titled "A survey on 3D Gaussian Splatting Compression Methods," provides a comprehensive overview of techniques aimed at optimizing the storage and memory usage of 3D Gaussian Splatting (3DGS) scenes. 3DGS is a state-of-the-art technique for real-time radiance field rendering, offering high-quality and fast performance. However, its extensive use of Gaussians leads to significant storage demands, making it impractical for mobile devices and headsets. To address this, the report categorizes existing methods into two main categories: compression and compaction. **Compression** focuses on reducing file size by techniques such as vector quantization (VQ), which clusters similar Gaussian attributes to reduce redundancy. Structured representations, like octrees, hash-grids, and self-organizing grids, also help by organizing Gaussians spatially and reducing the need to store individual attributes independently. **Compaction** aims to minimize the number of Gaussians by techniques like Adaptive Density Control (ADC), which dynamically adds or removes Gaussians based on their contribution to the scene. Other methods include attribute pruning, which reduces the number of spherical harmonics (SH) coefficients, and various structured representations that exploit spatial relationships among Gaussians. The report discusses the mathematical foundations, implementation details, and key design choices of these methods, providing a detailed comparison of their effectiveness using common datasets. It also highlights the advantages and disadvantages of each approach, establishing a consistent framework for evaluating 3DGS compression techniques. The goal is to advance the practicality of 3DGS in computationally constrained environments, making it more widely applicable in fields such as gaming and film.