I've just bought a second-hand (legal, approved by Gary) copy of the Garritan Orchestral Strings. I'm pretty thrilled as I got them cheap and I think this library still sounds fantastic by today's standards.
I have a dedicated machine waiting for them, stuffed with RAM and running Giga 3. I've just upgraded to this and haven't got around to using it yet. I'd like as far as possible to set GOS up in one standard template, and have as many articulations as possible all there all the time.
One thing I'm thinking is that I'd like to combine some of the instruments so that they can be triggered from a single track in Sonar, rather than needing loads of tracks each triggering a separate articulation. But I'm not sure what's the best way to do this, whether to:
a) use program changes
b) use the "stacking" facility, or
c) reprogram the .gig files themselves to contain more articulations.
The last solution seems the most thorough and customisable, and I get the impression from the manual that Gary only didn't do more of it himself because of the limited dimensions available in Giga 2.5. But I wonder about things like the Maestro Tools facility. For example, the LEG patches all have the special triggering of transition samples in them. Wouldn't this become problematic if I try to introduce another dimension, crossfading them using a different controller with another set of samples that don't have the transition samples? I feel like one could easily get into hot water.
Has anybody done anything like this? I mean, reprogrammed GOS to take advantage of the capabilities of Giga 3, that weren't around when the library was released? If so, I'd be interested in hearing what it was possible to achieve, and what the problems and limitations were, if any.
Assuming the legato patches work like VSL's, which if memory serves they more or less do, then yes, I think you're quite right: you can't add more dimensions and still have them work. At least in VSL (and I'm pretty sure GOS) the MIDI layer in the middle (VSL performance tool or GOS's tool) remaps the keys as well as playing different samples or parts of samples in context. So for example for an up- and down-bow alternation the upbows are at the bottom of the keyboard and the downbows are at the top... or probably vice versa... and each successive key plays either the low or high note. Which would happen when you switched.
HOWEVER-- that's going to happen no matter what you do: program switching, whatever.
But I would think that those cases are rare: that you can probably afford to dedicate one channel to violin legato, for example, and leave it alone... just use keyswitching, program changes, or stacking (and of the three, the first is preferable given how giga handles program changes and note offs, and that stacking plays all patches simultaneously without keyswitching) for instruments that don't rely on the performance tool or whatever GOS calls its MIDI interpreter.
I have reprogrammed gos for giga 3, doing almost exactly opposite to your proposed approach: divided every articulation, every dynamic level, into separate programs. It then really does offer much that other libraries do not: you can distinguish off-the-string from on-the-string short bowings; you can make an on-the-string staccato (slur over dots) that sounds pretty realistic, in the violins anyway. The legato implementation is not that special: overhang a release sample of the first note on top of the sounding of the second note; but this can be programmed, too. Though gigastudio 3 is very capable, and loads a lot of samples at once, I still run out of memory first. Therefore not loading the ff samples for a movement mostly in p-mp makes sense.
Beware that the tuning, especially violas, can be pretty far afield.
If you end up with specific questions, post more! But I'd advise against heavy combinations with lots of layering, you'll end up missing the tremolo or pizzicato when you need it in one piece, or the marcato in another piece.