Sometimes TCSPC curves are analyzed using the actual photon counts, and sometimes they are normalized with the signal running from 0 to 1, or even from 0 to some value less than 1. What quantity are normalized TCSPC curves normalized to/by? I have searched the literature for an answer to this but have not found a satisfactory answer.

I'm guessing one integrates the area under the curve to obtain the total photon counts, and then each photon count is divided by the total photon counts. But, why would normalizing the curve in this (or any other) manner be desirable?

Also, the square root of the counts are often used as uncertainty weights in fits to TCSPC curves, assuming the photon detections are Poisson distributed. How are these weights properly modified when the curve is normalized?

Thank you for your responses.