I compose mainly for TV and Film and my mothership-monster Cubase template with literally almost everything I have has almost 4000 (yes 4k) midi tracks. Those are grouped into sections of instruments Strings, Woodwinds, Brass, Percs, FX's, Sound Design, Voices, Pianos etc. and are hosted by many instances of Vienna Ensemble Pro all on one powerful machine, no slaves. All audio outputs are properly named and routed to proper subgroups, FX channels, main groups and finally to the Stereo or Surround output. This gives me the freedom to export audio tracks as audio stems for further mix in different formats, depending on the clients wishes. Some need stereo, some 5.1 or 7.1 mixes.
The fun part is this - using tracks presets, export selected tracks function, VST instruments rack presets etc. I am able to create smaller versions of my big template for the times where I don't need everything under my disposal. Sometimes I only write for strings so I load only the string part of the library and go from there. It is much quicker to load but the main logic behind this system/construction the naming and routing stays constant.
It took veeery long time to create this whole system but it pays off in the long run because now it is just a matter of a few clicks and whatever sound I need for a project - it's there ready to go!
Adding midi controller surfaces, touch sensitive monitor, iPad with Lemur on it and some long hours of programming and mapping Generic Remote functions made composing fun again and not - like it was before - 15% composing and 85% loading, searching, tweaking, troubleshooting, coursing ... ;)
Every bigger project is stored in one project folder for this project only and inside there are folders for Reels, episodes, parts and so on all depending on the nature of a project.
Saving and naming different versions of a session is also very important since often months later I need to re-open particular Cues for whatever reason (another DVD version, Surround Mix, shorter version, version without percussions etc) and I should be able to find myself in the session quick and with no extra pain.
Using notepad, screenshots and backing up stuff saved my butt many times so far and those are the things we don't really like to do. It is worth the time invested.
:P This post makes me jealous. :P
I think this is the most ideal situation.
But if i am allowed i have some question:
- how many cores you need for that? 48 ?
- You use one instance (as indicated as preffered by vienna) or many instances with your own core assignment ?
- And the output routing - how do you prevent your individual tracks to suffer from getting masked when so many signals are on the same signal path when doing this on a single machine?