About the number of NeRF

[xiaobiaodu]

Hi

Thanks for your excellent work! I have some details that make me confused and I may need your further explanation if you are available.
You mentioned in your paper that you use N NeRFs to disentangle different foreground and background. But I find that you only use one NeRF(named "decoder" in your code) to achieve this. That's different from your paper, right?

By the way, could you please explain why NeRF can disentangle the objects and background and achieve the segmentation result?