Of course, it may seem obvious, but those particular file types have to be selected in Photorec's options, I don't think that they are by default... (JPEG is, on the other hand, as it's much more common – and it was Photorec's original target, as its name implies.) As a general rule, it's best to select as few file types as possible, by unchecking those which are definitely not to be found on a given volume, because some legitimate files can be truncated by random false signatures of another file type (for instance, “ÿØÿ” or “FF D8 FF” can appear randomly in the middle of a video file, if it's at a cluster boundary it will be interpreted as the begining of a JPG file, and the video file will be truncated to half its length, even if it's not fragmented and could otherwise have been extracted sequentially). In that regard, Photorec's approach is quite rudimentary, despite the fact that it has been compared positively with very expensive forensic sofwares in professional tests. (Two documents I've found on the subject : “A comparative analysis of file carving software”, Timothy Courrejou & Simson L. Garfinkel, 2011, at
www.dtic.mil ; and “Advanced file carving : How much are you ignoring?”, Bas Kloet, Hoffman Investigations, 2010, at digital-forensics.sans.org.)
But, as it has been said already, if the bad areas are scattered all over the surface, it should be / should have been possible to at least partially recover the MFT, hence get a partial directory structure, which is much more convenient than a bunch of folders with 500 files in each of them, sorted by the order in which they were extracted instead of their original location, and named after their first occupied sector (which can be very useful for the recovery process but not so much for the end user !). In a case like this both approaches are complimentary : recover as much files as possible using what can be salvaged of the filesystem, then do a thorough file carving, and then remove the duplicates, or let the client take care of that as it can be a painstaking process. (There are quite a few duplicate files finders, my favorite are DoubleKiller, streamlined and efficient, and AllDup, which has some nice extra features but a rather cluttered and unintuitive interface.) R-Studio recognizes both of those RAW file types, so it can search for them as “Extra found files” while scanning for directory structures, thus making those extra steps unnecessary, as long as the detection accuracy for those files is just as satisfying as Photorec's.
And of course, never do that kind of analisys directly on a problematic HDD, that should be obvious for a professional...