Question about the "estimate_cell_number_RNA_reads" function

[DDDoGGie]

Dear Author, I reviewed the function "estimate_cell_number_RNA_reads" for estimating the number of cells in each spot, but I couldn't understand the rationale behind the algorithm. It seems that the total UMI counts in each spot are first normalized and then log-transformed. A curve is fitted from this to estimate the cell number for each spot. What I can't understand is: after normalization, the UMI counts in each spot are the same. Wouldn't this step eliminate the differences in cell number? Could you please explain how the proportion of genes in each spot can be used to estimate the cell number? Thanks a lot~