the extraction of the files is trivial, the problem is their name and the directory tree which is more boring/complex.
so if you don't care about the names (they will have a sequential numeric name) the following script does the job:

*edit* use the subsequent scripts

I worked on this format some days ago (there is a thread here http://forum.xentax.com/viewtopic.php?f=21&t=3899) but there are some fields missing or that require additional checks/reversing mainly in the part regarding the chunks of compressed data because the rest should be complete. for refer...

for anyone, the latest version of the script is 0.2.1 (thanx to chrrox and brienj for the testing eh eh eh)

problem solved, it was simply that the data containing the files isn't compressed with gzip.
I have updated the script: http://aluigi.org/papers/bms/fbrb.bms

eh eh eh I made all my tests with the demo, that's why I was already working on this format while you wrote the first post :)

fixed both the bug of MEMORY_FILE (because one of the file had size 0) and the compatibility with the EPK8 archives (names of 8 bytes instead of 4)
released script 0.1.2

chrrox pointed me to those "EMD " fields of the EPAC archives which must be skipped, so I have updated the script to version 0.1.1

uhmmm, the last thing I can try is watching the extractionlog.
so launch: then send me log.txt or attach it to your post.

well, I have decided to do a complete job including support also for PACH files and EBP chunks (so not only those included in the EPAC archives but also other provided as "stand-alone" files):
http://aluigi.org/papers/bms/yuke.bms

the following quickbms script should be able to extract (and unpack automatically) any EPAC archive:
*edit* use the yuke.bms code posted below (it's EXACTLY like the one I posted plus separate handling of standalone PACH and BPE files)

just as reference I have continued again to work with this format but it really sux so I give up

I have verified the structure of the file table and it has nothing strange or fields too big (nothing that can't be handled by 4gb or ram + virtual memory eh eh eh). I have noticed that you are using an old version of quickbms because at line 5111 there is no myalloc in the current one. so can you m...

I have worked on this game just today but there is something strange in the compressed chunks.
the algorithm is lzo but for some unknown reasons it doesn't work properly, even using the dumped function used in the game

this particular type of compression (that I have called "yuke_bpe") is now included in QuickBMS 0.3.9a

how much ram you have? it's possible that you don't have enough ram and/or virtual memory because it must decompress over 1gb of data in your memory. so I'm highly sure that this is the problem (the files you provided were enough small, that's why I opted for the in-memory decompression). in this ca...