Hi
Could you upload a pi-psm file, containing the process icon with the PixelMath instance?
Did you correctly set the image identifier of the sinthetic flat on PixelMath? How are you applying it to the original image?
Meanwhile, try this:
Calculate the median value of the sinthetic flat image. Now, desactive the "rescale" check box. Put the median values calculated for the flat as the scaling factor for the result. This would yield a result very similar to the original image, in terms of the mean brightness.
I guess that you have some zero valued pixels in the flat imagen, and then, when you divide, those pixels in your original image become totally saturate, with great value, and when you rescale, all the information is hidding at very low intensity values. Please verify this. To avoid that, create more sample boxes in the DBE instance.