This new model looks impressive.
"Spatial gradients of the discrete rasterization are approximated by the novel concept of ghost geometry. After rendering, the neural image pyramid is passed through a deep neural network for shading calculations and hole-filling. A differentiable, physically-based tonemapper then converts the intermediate output to the target image," states the abstract. "Since all stages of the pipeline are differentiable, we optimize all of the scene's parameters i.e. camera model, camera pose, point position, point color, environment map, rendering network weights, vignetting, camera response function, per image exposure, and per image white balance."
The team thinks their new approach can synthesize sharper and more consistent novel views than existing approaches "because the initial reconstruction is refined during training". Sadly, no code is available yet but you can now check out a paper from the team available here.