Normalization in 2d subtraction

Normalization in the 2d subtraction seems to off. For instance, for the example in PIPE2D-735, the flux in the whole image is 65K, while in the subtracted image the whole flux is still 14K, i.e., it is very far from 0. The figure is showing what fraction of flux is still in the residual image, after subtraction. As we can see, for some fibers more than 30% of flux is still present after subtraction. Possibly this is due to poor centering, but I believe I saw this issue also before.

Robert also mentioned that we saw some weird stuff with normalization, but I am not sure if this is connected.