neural radiance fields (NeRFs) produce state-of-the-art view synthesis
results. However, they are slow to render, requiring hundreds of network
evaluations per pixel to approximate a volume rendering integral. Ba
Neural Radiance Field (NeRF) is a framework representing a 3D scene with MLP, achieving state-of-the-art photorealistic image renderings for view synthesis and expanding capabilities to generate views from dynamic scene representations with fewer images for training.