Volumetric neural rendering methods like NeRF [34] generate high-quality view synthesis results but are optimized per-scene leading to prohibitive reconstruction time. On the other hand, deep ...