brookman64k

The paper says that training / reconstruction takes 4 hours. While inference runs at 60fps reconstruction is far from real-time.

Great work! Having tried the code from the original NeRF paper I found the inference time (generation of new views) to be rather slow because the network had to be queried multiple times per ray (pixel). The paper said that there is still potential to speed this up. Did you improve inference speed and do you think that it will be possible to get it to real-time (>30 fps) in the foreseeable future?