Revel filters out duplicates based on file byte comparisons (aka. hashing or checksumming). Revel will not filter images that are edited copies of originals.
Revel doesn't look at file names when determining whether images are duplicates, but considers the binary contents of the files. If the file contents are exactly the same, then Revel flags them as duplicates regardless of filename and other properties.
When there are two photos with the same pixel data--but one has some metadata added, or is of a different file format, then images are not considered as being duplicates.
When you upload the same picture from 2 different devices, Revel will determine it as a duplicate