When would you add your own library size in edgeR?

Question

[Tota]

In edgeR there is an option to add your own library size for each sample instead of letting edgeR sum the total reads for each sample. You could set your library size to total reads, uniquely mapped reads etc.

In what scenarios would you want to add your own library size than to let edgeR sum the reads for you?

Say if you were inputing your own library size that is larger than the automatically summed reads by edgeR, wouldn't the reads relative to your own library size be smaller? And you could lose some information because you've decreased its proportion?

I'm trying to find information but to no avail. Any help, clarification or correction of my logic will be appreciated.