The hosts discuss 3D Gaussian Splatting, a technique for creating photorealistic 3D scenes using photographs or videos. This method differs from photogrammetry by creating a point cloud or radiance field instead of a traditional mesh. Advantages of 3D Gaussian Splatting include faster computation, smaller file sizes, and more predictable results. However, it is not as sophisticated as NeRFs, which use machine learning to create more detailed representations. The hosts explore various applications of 3D Gaussian Splatting, such as virtual production, virtual reality, and medical imaging. Furthermore, the sources cover research efforts to improve the technique, including optimization methods, training strategies, and hybrid approaches. The information suggests 3D Gaussian Splatting is a promising technique that could have a significant impact on various fields.