Questions regarding the example illustrating the supervision conflict in Figure 1

[JIANG-CX]

Hi, great work! I have some questions regarding the example illustrating the supervision conflict in Figure 1. I'm curious why in the case of NeRF with constant opacity, the points before the surface are assigned a high opacity instead of the points behind the surface (as indicated by the points in the green rectangle). If the points behind the surface were assigned a high opacity, it seems that such a conflict would not arise. I would greatly appreciate it if you could provide an explanation for this. Thank you!