If you have several tracks all routed to a common Gigapulse instance, how do you maintain discreteness? Each track is routed to a separate output, but GP is routed to only one. So even though my strings are routed to say "7,8", if I send it to GP, the wet portion is routed to "1,2", or whatever the aux send is set to.
The only say I can figure out how to do it is have a GP instance for each track and route the aux output to the tracks output, or bounce one at a time changing the output of the send effect each time. Of course this is impractical.
Is there a way to bounce multiple tracks each having a unique wet/dry mix?
Nothing wrong with mixing all your GP instances into one reverb ouput track unless you need to manipulate them separately. I sometimes do that with lead vocal tracks... often find I need to filter a female vocal's reverb to prevent it from worstening sibilance. But I do like to keep the reverb separate from the dry source. That way I can envelope it and do other stuff.
If you want to maximize imaging clarity, try using a separate GP instance fed by it's own audio pair whose source setup is matched to the instrument(s) panning. I usually do this with a daw send-bus for each pan position, each feeding it's own audio out.
Also nothing wrong with doing it one track or send-bus at a time. Just tedius. Little choice if you run out of horsepower. Or if you're like me and can afford one really good reverb box. You either get more horses or take more passes. One approach that's not too bad is to pair each instrument with it's own reverb track. That way you only need to re-render reverb tracks that go with instrument tracks you edit.
You can get simple panning within one normal gigapulse (at least some settings: i use 'medium hall' mostly). Just route from the midi mixer to an input, set the position and width controls in the dsp input, then route THAT to a dsp group with your gigapulse. For instance, I use a large number of separate tracks for first violins, and another large number of tracks for second violins. In the midi mixer, all the firsts are routed to "input 1,2" and all the seconds are routed to "input 3,4". Then on the "input 1,2" page of the dsp mixer, the first violins are positioned somewhat to the left and the "input 3,4" page the second violins are positioned closer to the center. Both inputs route to "group 1,2", where a gigapulse instance is loaded with no panning or microphone adjustment, but with the level and perspective that suits me.
Since my computer can handle three gigapulses, I route the violas and cellos through 2 more dsp input pages to another group with a different gigapulse, and the woodwinds to a third gigapulse through several more inputs with position adjustments. (Brass and bass reverb settings are handled entirely separately). Not perfect, but seems to be a pretty good separation with little wasted time in re-adjusting things constantly.
A single gp with impulse sources set to 15 or 30 degrees left and right is what I'll probably be using for piano. It might be a while before it'll be possible to run 44 instances so each key position can be individually focused... though I've toyed with the idea of splitting the keyboard into a handful of separately processed zones. Treating each zone like an individually positioned instrument.
For example, one could fire up 3 gp instances. Since each one has 2 impulse source positions to manipulate that would accomodate 6 different zone panning positions if they were routed to the gp's in pairs.