Duplicate Ignore Transparency

To reduce the filesize of many of my images that are stupidly large, I am converting them to jpg, this eliminates the transparency, in my case, making the formerly transparent area black.
I import the files to hydrus so I can give them a once over saying yes, the quality difference is ok though duplicate search, and remove.

now the issue is, when a file has a large enough transparent section it doesn't pick up that they are duplicates, forcing me to manually do the duplicate tagging. I have also seen that while i'm importing these down converted files, some of them will have several duplicates attached to them that again, hydrus didn't pick up on in the first place due to transparency.

I think ignoring transparency area would be a good idea, or assigning it a black/white color if you can't just duplicate search with an ignored area.

The issue does not name any files, tests, or entry points. Start by locating the duplicate-search implementation and its handling of transparent image areas, then determine how converted JPGs should compare with their transparent originals. Done means duplicate search reliably identifies the described pairs without requiring manual duplicate tagging.