Neural rendering is a rapidly emerging field that can compactly represent scenes by learning to render from existing observations through neural networks. The main idea of neural rendering is to combine the insights of classical (physics-based) computer graphics with the latest advancements in deep learning. Similar to classical computer graphics, the goal of neural rendering is to generate photorealistic images in a controllable manner. For example, new viewpoint synthesis, relighting, scene deformation, and synthesis. Section 15.1 introduces the original NeRF theory, Sect. 15.2 discusses acceleration methods for NeRF, including the AutoInt method and the Plenoxels model, while Sect. 15.3 discusses rendering techniques for dynamic scenes, Sect. 15.4 analyzes relighting methods, Sect. 15.5 introduces the generalization problem of NeRF, Sect. 15.6 introduces the latest quality improvement methods, Sect. 15.7 analyzes the latest Gaussian Splatting technology, and, finally, Sect. 15.8 introduces the application of NeRF and GS in autonomous driving.