实景三维是新型基础测绘标准化产品，为智慧城市建设提供了统一的空间基底。目前，实景三维模型存在生产过程复杂、可视化效果失真等问题，限制了进一步推广应用。人工智能技术发展日新月异，开始渗透到了三维重建领域。回顾了三维重建技术的发展，对多视图三维重建、神经辐射场和3D高斯散射3种建模技术进行了对比和分析。采用3D高斯散射技术生产了实景三维模型数据，并对数据进行编辑，验证了3D高斯散射技术建模的高效性和便捷性，展望了3D高斯散射技术未来可能的研究方向。

3D real scene is a new standardized product for basic surveying and mapping, providing a unified spatial foundation for the construction of smart cities. At present, 3D real scene models have problems such as complex production process and distorted visualization effect, which limit their further promotion and application. The development of artificial intelligence technology is advancing rapidly and has begun to penetrate into the field of 3D reconstruction. We provided a review of the development of 3D reconstruction technology, compared and analyzed the multi-view 3D reconstruction, neural radiance field, and 3D Gaussian splatting three modeling techniques at first. Then, we used 3D Gaussian splatting technology to produce 3D real scene model data, and edited the data, verifying the high efficiency and convenience of 3D Gaussian splatting technology modeling. Finally, we looked forward to the possible research directions of 3D Gaussian scattering technology in the future.