Nebula doesn't default to a mode as far as I know.
The library you load into it is set (by default) to a certain mode by the library-developer.
And as a rule, you should not touch it (they did their tests if it sounds OK or not).
Remember I explained in some other thread that there are two algorithms? FREQD and TIMED? And that TIMED maybe better but requires more processing?
Well, (correct me if wrong somebody) SPLITH mode allows the developer to use both. Normally (specially for reverbs) they use SPLITH to apply 'TIMED' algorithm to the initial part of the effect, and then switch to FREQD mode for the long tail a program might have. That way you have high-quality TIMED mode for the start and most important part, and FREQD mode for the stuff after.
For short programs like preamps / EQs and stuff, it might be that the developer set the mode to CLASSIC because they don't have to work in hybrid, they can have the whole program TIMED / FREQD mode.
So to me, the same rule applies here as in your other thread about tweaks: Yes, go ahead and play with it. But I don't see the point, since those parameters are set by the library developers and they know better what they are doing and can compare it to the original sampled device .
no you shouldn't use splith all the time. splith mode is mostly as jorismak described, but also: a) you can't use smooth2 smoothing when using splith, because each use an extra kernel (this is why 10k programs say 11k, it's not because they actually have 11 sampled kernels, but 10 and then one for smoothing or splith), and only one of them can be used at a time because they share that kernel, or something. b) you need smoothing for pretty much anything with dynamics, except for longer reverbs because they use longer program rates where smoothing isn't necessary c) so that means that splith mode is only something you can use with programs without dynamics, like most EQs, because they don't need smoothing d) i'm not sure but i don't think you can use it with reverbs, or at least not ones over 100ms
but so basically, if you take a preamp or compressor or any program that's not using splith, and set it to use splith, chances are pretty high you are going to be getting artifacts. unless it's something without dynamics, like the vast majority of equalizers.
Normally (specially for reverbs) they use SPLITH to apply 'TIMED' algorithm to the initial part of the effect, and then switch to FREQD mode for the long tail a program might have.
c) so that means that splith mode is only something you can use with programs without dynamics, like most EQs, because they don't need smoothing d) i'm not sure but i don't think you can use it with reverbs, or at least not ones over 100ms
It would be nice to have a table with all kinds of categories like reverb, EQ, consoles, etc.. and general good setup for PROGRATE, DSPBUFFER, TIMED FREQD MODES, for the best sound possible in a good machine, that is possible?
dynamic programs should not use splith, they should use either classic with freqd or timed and have smooth2 smoothing. without smooth2 smoothing the program will almost surely cause artifacts.
nondynamic programs like eqs can use splith because they dont need smooth 2 smoothing. they can also use classic with either freqd or timed. timed is always best quality, but uses more cpu. splith is mid quality, and freqd lowest (but still really good, and the differences are really subtle anyway).
its that simple.
by default, eq programs usually or always use splith. by default dynamic programs never use splith (and shouldn't). you could change an eq from splith to either classic with freqd or timed, but you would want to make sure the program rate doesn't change because if it does, that could also lead to artifacts. so the safest thing would be to just leave it on splith. you should never change a dynamic program from classic to splith. ever. but again, safest thing is to just leave this stuff alone.