Bioshock2 Fsb's by pepper at 10:22 PM EST on February 9, 2010
while using fsb_mpeg for bioshock 2 fsb files, it tends to be hit and miss. it might be the SP fsbs are different from the MP fsbs, but what i'm seeing is that with the single player fsb's a value of 2 for padding works partially.. it'll go through a number of files, then come to one and have a sample count mismatch. either the fsb is messed up and has different amounts of padding at different times, or i'm not doing it right. i'll upload one.
oh and the pkg files in the MP sound folders are basically just raw data according to goldwave, not sure if anyone else knew that yet, but the audio is one sound right after another.
.. the upload's still going... http://www.sendspace.com/file/alr8hr
I put up fsb_mpeg 0.7 which will seek back into the padding to find a sync frame (if -p is used). I think this should clear up the issue. I don't know why the padding is so unusual in this game. I still suggest using -p 4 everywhere, if it doesn't work for some files let me know.
edited 7:05 AM EST February 10, 2010
perfect by pepper at 9:54 AM EST on February 10, 2010
amazing, i guess -p 2 was more of a temp fix for some of the files, but -p 4 works perfect now.
Lost sync is normal, as long as it says "Success!" all is well. However in MaterialSFX.fsb there are a few strange values, where there are only 1152 samples in MPEG-2 Layer III streams, the FSB header says that there are 1728, which throws off my check (which is otherwise very valuable for redundancy). I added a very special case to allow this.
I also fixed the issue with the EOF, trying to read past the file end.
I really appreciate this, i dont want to come off greedy or needy, i merely aim to improve the tool.
they all work now (except one) and i found one more, probably similar to the materials in nature (which now extracts perfectly) but i can only assume slightly different. i already found what i was looking for, but i assume its best that everything works fully for future things and whatnot.
the machines_objects.fsb is the only one still giving any errors, it gives a sample count mismatch.
On the contrary, I appreciate your error reports. It gives me comfort to know that it works across all the files.
Your hunch was correct, it is the same sort of bug that I'd encountered last time with MPEG-2. It seems like when building the fsb something screws up with 1152 sample files, and it counts one extra frame. In the MPEG-2 Layer III case this means it expects 1728 samples, in the MPEG-1 Layer III case it means 2304 samples. I modified the special case to check for this, all should be well in 0.9. Thanks for testing it so thoroughly!