PBRv3: Struggling with confusion of the importance function of perspective camera

[Yang]

Hey all, I really love this book and trying to digest most parts of  it. Thanks for this masterpiece.

One thing that has been making me struggling for months is the importance function of perspective camera when used in BDPT:

1) on one hand, when we are generating camera paths, the weights of the perspective camera is always 1:

https://github.com/mmp/pbrt-v3/blob/13d871faae88233b327d04cda24022b8bb0093ee/src/integrators/bdpt.cpp#L81

2) on the other hand, when we are connecting the subpaths and sampling the camera for t == 1 case, the importance function is computed according to the "We" function of the perspective camera, which is not 1.

see

https://github.com/mmp/pbrt-v3/blob/13d871faae88233b327d04cda24022b8bb0093ee/src/integrators/bdpt.cpp#L470

****

While I understand how we get the importance function of the perspective camera for (2), I do not really get why we can simply set the importance function to be 1 for (1) though.

I wonder if case (1) is just an approximation?

Thanks a lot.

Yang