Abstract: Existing 3D Gaussian Splatting (3DGS) methods employ fixed spherical harmonic (SH) degrees and static density control strategies, which restricts their adaptability and performance in scenes with different levels of complexity. This paper proposes two enhancement schemes to improve the adaptability and efficiency of 3DGS. Firstly, an adaptive SH degree control mechanism is proposed, which dynamically adjusts the SH degree for each Gaussian based on reconstruction error and viewpoint changes, achieving more accurate color representation while optimizing computational resources. Secondly, a dynamic density control strategy is developed, which adaptively adjusts the density threshold according to local scene complexity, taking both depth information and image gradients into account to achieve a more reasonable point cloud distribution. Experimental results demonstrate that our method achieves PSNR improvements of 0.59 dB, 0.49 dB, and 0.57 dB on the MipNeRF360, 3DGS, and an image sequence of the Comprehensive Experimental Building of Nanchang Hangkong University, respectively, compared with the baseline method. The advantages of the proposed strategies are particularly significant in challenging scenes with dense vegetation and complex occlusions, such as the “Garden” scene. Furthermore, ablation studies verify the individual effectiveness and synergistic gains of each proposed module.