We (Pascal OVNI and me) kept it very simple. No real big tricks as things are.
It was a 3 hour live set, where we moved from ambient to full-on, warming up for Eat Static.
First half, was on Pascal’s kit, sub-mixed through my A&H Qu24. MPC wasn’t used for that first half.
Then for the more full-on material I loaded projects on-the-fly on the MPC, each project being a ‘song’. Currently, the only way to get STEMs going, is to load the whole audio files into memory. Just about enough room for 3 24 bit stereo STEMs. That also let me choose tracks on-the-fly.
Transitions are/were ambient/fx/pad based on a bunch of synths, rather than something approaching beat-matching. We tend to add to the material, with live percussion, synths and sometimes guitar, rather than mess around with the material.
If I can get more STEMs (if Akai implement streaming from SSD), then I can route individual tracks out to the mixing desk for messing around with - or use the MPC’s own effects. Merv (Eat Static) actually does something similar, in that he has a multi-track HD based playback system, into a big 32 channel analogue desk. He then ‘plays’ the material with the desk, similar to how the dub guys used to (and still do) work - sending to delay, reverb etc. and playing with EQ / filters, panning etc.
I could chop up loops, but we’re not working in the mode of arranging on the fly. The material is a bit too complex synth-wise. Drum-wise, I could do that, but would take a fair bit more work.