Are PSF rip parameters inaccurate? by jimbo1qaz at 9:42 PM EDT on August 7, 2016
I read on http://www.neogaf.com/forum/showthread.php?t=116857 that PSF/PSF2 rips sometimes have reverb/channel-volume parameters set by ear (instead of being extracted directly from the game). Specifically, they mentioned some Chrono Cross track as having incorrect bass.
This seriously worries me. All along I've assumed that xSF files are exact representations of the data stored in-game, not a set of arbitrarily picked and inaccurate fudge factors, chosen to sound "somewhat" like the actual song. And now to think that the music I've been listening to all these months/years... is WRONG?!
Is that correct, that ripped PSF tracks actually contain fudge factors determined by ear, and not accurate to the game?
People actually often borrow drivers, typically from similar games, same developers, etc. There's also a lot of sets that use generic drivers (for standard VH/VB/SEQ data). So unfortunately, yes, it's true that many sets we have are not accurate, and not just in regards to reverb even. I ripped the original driver of Digimon World for this reason -- the rip that was floating around (which used Davironica's generic PSF driver) even had some inaudible notes in some tracks.
Is the difference due to different drivers interpreting data differently, or different drivers setting/interpreting reverb/volume differently, or incorrect state stored directly in the PSF file?
Using different drivers, guessing parameters because knowing the real values would require deep disassembly of the entire game or playing through every possible moment manually and capturing save states to find what it uses in every situation, etc.
Let's pick Chrono Cross. Are volume parameters stored as part of the ripped sequence files, manually compiled into custom sequence files, or stored as PSF state? (someone mentioned that volumes are wrong).
They'd be stored as function parameters somewhere in the game's code, which isn't in the PSF. Guessed parameters have been substituted in their stead.
To find the parameters, one must debug the game while playing through it completely, and locate which parameters are used where. They're not in the songs themselves, so they could be anywhere else in the data.
There is a fundamental flaw in believing that a 100% faithful PSF rip/set exists simply for the fact that original drivers require having retail/machine code to be altered. The r3000a mips language isn't too extensive, but people make mistakes, overlook certain variables, etc.
Yes, games definitely use static/variables with playback of track/sfx specifics... but not as many as you'd think. In my experience, they typically have a global reverb for all music and sound effects often use parameters located in a simple array somewhere in the exe (sometimes, the table is located in an entirely different, non-suspect file).
Of course, there are always exceptions - the tracks I've encountered that have variable tempo/reverb/etc were always "premiere" tracks, such as something that may be played during an FMV, etc.
Data should always be 1:1 faithful representation. As kode mentioned, it's really a matter of debugging the exe and ensuring that each track uses the same global variables, noting which do not and proceeding to make the appropriate rip.
...and this is exactly why I offer the option for user-friendly parameter modification in my generic driver - reverb, tempo, channel muting and so on can easily be customized via psf-o-cycle, etc. I presume that Chrono Cross is AKAO (unsupported), but still. I'm not going to pretend it's the best solution, because it's not and probably never will be... but it is rather nice to have customization available to the end-user for this exact issue.
All said, one would have to be pretty dedicated to complete accuracy... but even then, you'd have to look past the fact that you're using a hacked executable or non-official driver coded by fans ;P