Far cry 4 (PS3) *.DAT/*.FAT by Ultrafighter at 4:18 AM EDT on October 9, 2018
Hi guys, I'd like to get some assistance with properly unpacking bigfiles from the game (mainly audio ones) so that consistency of data is preserved & resulting streams / banks / index files remain playable / readable / usable. I believe in your abilities and hope that FC4 (and some other installments like FC3 or Blood dragon) won't remain unrippable way too long!
As mentioned it's using Ubi 'file' BAO. Here is some info for adventurous rippers:
BAO comes in two flavors, external .pk+.spk and 'file' BAOs inside bigfiles. 'File' BAOs reference each other by ID, and ID is in the bigfile index (.FAT), so they must be named by ID like 0000289C.bao (extension/0x is optional) so that vgmstream can find them by filename.
Once properly extracted, support must be added in vgmstream (this is the 'easy') part. Extracting those bigfiles is no joke though.
For "forge" bigfiles (Assassin's Creed, Prince of Persia 2008, etc), first must extract the various .forge (there is a bms for this). Some contain raw BAOs, but others you must uncompress some of the resulting files (also has bms), then extract header BAOs from some of the uncompressed files (no bms, complex format), and put together with raw BAOs.
For "yeti" bigfiles (Beowulf) the bigfile index seems to simulate folder/files, I couldn't figure out exactly nor find bms.
For "FAT" bigfiles (Far Cry), I think it's somewhat like "yeti", maybe existing .bms could be modified to output file IDs.
Good luck!
by Ultrafighter at 2:18 AM EDT on October 13, 2018
I found another way of unpacking FC4 archives (Gibbed.Dunia2.Unpack.exe) and filenames seem correct (like 65FD7E399CCAE0DB or F9AFA07C9CBAE616, 001b8b44.sbao or 001b8b46.sbao) but apparently the files themselves are the same (I compared extracted content to 6 "MP3" samples I provided and they're byte-wise identical). Does it mean that this program doesn't do its job properly (like Luigi's fat2_fat3.bms) or bigfile format was probably changed for the 4th installment? Or (to keep it short): the files unpacked by either that BMS or this prog can't be implemented in plugin anyway (or you possibly need more examples)?
And can you please elaborate on your quote from VGMstream topic: "I tried to rip Blood Dragon before but the bms part was just too complex and time consuming"? Is there a 3rd method of extraction, the one requiring not a single script or tool but multitude of scripts? I'd like to check out if it's going to work with FC4.
BAO data can be repeated, but you should check with a hex editor if the same file is included in the FAT 6 times.
Blood Dragon simply I didn't have a program or BMS to extract the .FAT, and making one myself would be hard.
by Ultrafighter at 9:21 AM EDT on October 14, 2018
Hi, thnx for clarifying!
It seems you misunderstood me a bit: I didn't mean to say there're any duplicate files among those (mostly extensionless) files extracted by Gibbed.Dunia2, I merely wanted to say that I was lazy to use fat2_fat3.bms on SOUND.DAT + SOUND.FAT again just to compare every single file extracted with that BMS to every single file extracted using Gibbed.Dunia2. That's why I grabbed those 6 "MP3" samples I uploaded for you and searched for dupes; it turned out these six files are byte-wise identical (of course they have different names but actual files are the same), it was reason enough (IMHO) to report that both BMS by Luigi & EXE by Rick Gibbed seemingly output exactly the same files.
I hope this info was useful, cheers!
by Ultrafighter at 7:06 AM EST on November 15, 2018
Anyone interested in more samples from FC4 or FC2? The latter proves to be just as difficult as the former because all 3 versions (PC, X360, PS3) crash Gibbed.Dunia2, I'm not even sure anymore it was used to get this rip done. Best regards!