At 05:37 PM 7/15/01 -0400, Stuffed Crust wrote:
>On Sun, Jul 15, 2001 at 05:08:49PM -0400, Ravi wrote:
>> Hmm, I didn't know there were multiple versions. Could you give me the
>> names of a couple games in both versions?
>
>Not quite multiple versions per se, but older games only have 48
>instruments, while newer ones have a second block of 48 instruments
>deliniated by the 0xABCD after the first block. LSL2 has only 48,
>while SQ3 has all 96, for example.
Ah, cool.
>> Indeed, but this should (I think) be limited to fields that aren't actually
>> operator specific. The alrogithm field, for example, is a channel setting
>> that only needs to be included once. The second operator stores other data,
>> but I haven't probed that too deeply. I'd be happy to take another look
>> when I have the time (probably next weekend).
>
>For example, the second operator has freqmod values of 0x06, 0x26, 0x3e,
>and 0xf6, well outside the 4-bits that it actually uses. Similarly, the
>feedback value ranges all the way up to 0x0f.
Oops. I'm just guessting here, but the lower bits might still be valid.
Register values for the second operator have to come from somewhere, after
all. Perhaps the upper bits contain extra information that couldn't be
packed in elsewhere.
>Interestingly enough, things actually sound better if I flip the
>operators. It sounds like the carrier's correct, but the modulator is
>wrong. (or it the other way around? My ears are almost bleeding after
>a couple of days at this..)
All for the cause.
>Also, we have 96 (or 48) instruments; how do they map to the
>original midi instruments? there doesn't seem to be any kind of index
>in the patch. So it's statically mapped. On a per-version or per-game
>basis?
I always assumed that they're already in correct order. Since Sierra built
their own instrument sets for most games, they could arrange instruments
however they pleased; no need to do any mapping to another layout. In any
case, I don't remember seeing space for a lookup table in the Adlib driver.
Cheers,
Ravi.
