The question is also what additional benefits you would get from bringing all these images into DT. You might, for instance, want to use linking to them directly from within DT, and that’s a good use case. You could also replicate select image sets from a single-truth repository to other folders that are dedicated to, say, a single research output, a grant application, whatever. That’s another good use case where DT would shine.
In my own research use (somewhere between literature, history, book history, and more!), however, I have a large database of images (mostly of manuscripts and rare books), about 65 GB. But because I never really invested in creating good metadata and most images have the camera’s default image numbering for a file name, it’s the one database that I use where DT offers me very little benefit over, say, Finder or a shared Dropbox folder. I never search it and can only get to what I need by navigating folder hierarchy (Library / Collection / Shelfmark). It suits my needs, but isn’t often used in the ways my other databases get used.
Once more, a lot depends on the nature of your materials. If your archival materials contains, e.g. typescripts you can OCR the lot and unlock DT’s powerful search and AI. In my case, I am mostly collecting images of medieval and renaissance handwritten documents and there is, as yet, no reliable automated paleography solution!
Welcome to the forum, and good luck!