No, honestly, I know this could sound VERY scary in regards to processing times and memory needs (64bit yay!), but maybe it would be half as dramatic as it sounds (or will sound!!!
). Introduction
Currently you have horizontal source and transforms, making it essentially one dimensional.
Source A, B, C, D... - Transform A, B, C, D...
Imagine you could add a 2nd dimension to it.
[a]Source A, B, C, D... - Transform A, B, C, D...
Source A, B, C, D... - Transform A, B, C, D...
[c]Source A, B, C, D... - Transform A, B, C, D...
--------(2nd dimension) Transform [a],,[c]...
The oscillator index would receive a second index for the second dimension.
index X, index Y
This is not some arbitrary "wow, wouldn't that be interesting" experimental request, but a very conscious one in an effort to "strike fear" into the hearts of all other vst synths out there. This kind of thing could turn the already glorious MFO into an MMFO (MonsterMultiFormOscillator. Ok, ok, maybe MassiveMultiFormOscillator, haha, MatrixMultiFormOscillator? Anyway...).
Why would anyone want that?
Because you could make the most breathtaking virtual instruments the world has ever heard via a simple modular system (mux) and this power was open and ready for any user.
For example, in Native Instrument's Kontakt you have scripting, which can enable you to do some pretty amazing stuff, as I've already done myself, too, by the way, but it's something that takes time to learn and a pretty good understanding of programming, of course. Only few products exist like Kontakt that can make such virtual greatness available to their users and 3rd party vendors. But because it's so difficult, even those vendors run into their limits easily and it seems like they're already hitting a wall.
They also have to rely on Gigabytes of sample data only to escape in babysteps from the rigidity of its nature.
MMFO would provide simple access to create highly controlled, highly complex instruments anybody could put together in minutes from scratch. Not to mention that it then may also attract 3rd party vendors to go nuts with producing high quality presets.
Yes, sure, an easier way to go stereo in the sources may be interesting, too, (MMFOMG?) and a proper convolution module would probably be needed to seal the deal, but it sure would be one heck of a revolution!
Practical Techniques
Breaking up massive waveforms into wavetables of essential moments allows you to not only reduce the size of a sample to a ridiculous minimum (e.g. 30mb to 10kb), but provides unimaginable flexibility on what you can do with such a sample while maintaining its very nature.
Since I presently still have to choose which dimension to use to blend along a list of source grains, time or notes (frequency), I naturally go for notes in order to achieve a more natural reproduction of a sampled source such as flutes or choirs and such. But this requires severe compromises on sound changes over time. A lot can be beautifully recreated with the rest of the mux synth modules, but some characteristics are almost impossible to hit right.
Having the ability to both have a time based transformation or blend across source grains as well as a note based blend across those time based blends, one could do things that were never done before.
I leave the rest to your imagination, but I felt like I should mention it at some point and this is as good a moment as any.
