[kj911]

Use the delta algorithm at the beginning of the chain. It may improve the ratio, it may not.

Use 3.92 from SREP! (The default value of the "l/c" parameters is 512, if not defined separately.) You can play a little more with the a0/a1/a2 parameters, although it does not multiply much. (v3.93 contains decompression bug.)

Did you run MSC with only WAV=1? With the RAW=1 option, how many audio streams does it find? True, unpacking can be a little more problematic this way, but usually it works. What can be a problem if there is no error during packing, but when unpacking it does, because problematic RAW/PCM audio files can appear that run with errors.

It is interesting to note that the small test results show that with MSC there is only an extra 100MB gain at the end of the compression chain, after the initial 800MB size reduction.

In the case of LOLZ, in the case of compression on 1 thread as described above, be prepared for it to take up to ~2-3 days! It might only bring ~5%, but it might be more, since you've run it on 10 threads so far, it might even be a bigger profit.

ps: Hint, try use DLZ replacement from LZMA, from faster compression times testing than LOLZ. Use 256-512MB dictionary.

My preferred switches: -slp -d512m -lp0 -fb337 -mc8192 -cu3 -cm43