Do I need to normalize unspliced reads as long as I've normalized spliced reads?

[dg520]

Hi,

Thanks for developing and maintaining this powerful tool!
I have a question about normalization. I need to regress out some batch effects before running velocyto.R, similar to this tutorial provided by Seurat team.
However, I'm quite confused about the fact that they only applied normalization (via SCTransform function) on the "spliced" assay but not on the "unspliced" assay. How can the Velocyto result hold in this way? What is the reason for not normalizing the "unspliced" assay?
Any insight is really appreciated!

Best,
Dadi