The field of 3D Gaussian Splatting is moving towards more efficient and generalizable methods for 3D reconstruction and novel view synthesis. Researchers are exploring new approaches to address limitations such as redundancy and geometric inconsistencies in long-duration video sequences, and to improve the quality and completeness of 3D reconstructions. Notable advancements include the use of hierarchical compression, mixture of experts, and re-activation mechanisms to improve the efficiency and accuracy of 3D Gaussian Splatting. These innovations have the potential to enable more widespread adoption of 3D Gaussian Splatting in applications such as immersive communication and dynamic scene reconstruction. Noteworthy papers include: SaLon3R, which introduces a novel framework for structure-aware long-term 3DGS reconstruction, and PFGS, which proposes a pose-aware 3DGS framework for complete multi-pose object reconstruction. HGC-Avatar and MoE-GS also demonstrate significant improvements in compression efficiency and rendering quality, while Re-Activating Frozen Primitives for 3D Gaussian Splatting addresses the challenges of over-reconstruction artifacts and primitive frozen phenomenon.