Either they miss images that are the same but not really (Eg two different jpg save of an original), or they catch false positive.
Then you find something that works great, but it ends up requiring manual usage to be reliable, which makes my 20+ years of photos archive unusable.
Then you try to find the same for movies and video file, and it's even worse.
I'm writing PhotoStructure, specifically to deal with my last 20+ years of photos, and the fact that I haven't stayed on top of keeping them all organized. The disarray was made (much) worse by my bad luck in picking cloud photo services (most of them which have shut down, or, like Google Photos, left me with a "takeout" that was down-sampled or with metadata loss).
In any event, if you tackle this task, this article should help you consider which set of duplicate files you'd want to consider to be the "winner": https://photostructure.com/faq/what-do-you-mean-by-deduplica...
Frontend alternative to JavaScript
Easier solution than CSS
Building RAD tools on modern web stacks is like building castles on quick sands - no one is quick enough to keep up...