See how useful dynamic side chains can be? There’s also so much more that can be done. I hope these ideas help people in developing their own uses and inspire creativity.
Honestly, that is all really good stuff and great examples of ways that sidechains on dynamic processors can be used. I guess I was hoping for a bit more creativity form the way it started out. You said "dynamic side chains", which read to me like "dynamic processing on the side chain signal". You also talked about how you can do whatever you want to the sidechain signal, but then didn't really do anything to it. Yes, you used one source to control the dynamics of another, but you didn't do anything to that source on its way to the detector.
So, I've never actually tried it, but it might be worth a try to actually put a dynamic processor on the sidechain signal itself, before the detector.
Say you want to duck one thing with another, but that other thing swings a lot further than you want the ducking to go. Compress (or limit) it on the way to the detector so that the gain reduction only ever goes so far.
If you split the signal, run one copy through a compressor/limiter and then to the detector, and the other through the compressor's audio input, you can emulate that thing that some vintage circuits do where they "open back up" and start to apply less gain reduction than expected as the signal goes further and further above the threshold on a more standard comp and with more precise control over how it does that.
Likewise, if you put an expander in the detector path, it could emulate a soft-knee response in a hard-knee compressor.
OR Put other things in that detector path.
Filters are pretty common, and many comps have them built in nowadays. Be careful with that, though. A filter in the side side-chain is not the same thing as a multi-band compressor. An MBC splits the signal into frequency bands and applies discrete compression to each band based on its own content. A side chain filter causes the compressor to compress the entire frequency spectrum based on the content of the filtered signal. That is, if you run the HPF up so that it's only responding to high frequencies (typical for de-essing), it will still turn down the low-end too when the high end gets loud. Likewise, if you use an LPF it will still turn down the highs when the lows get loud. (Course, that's almost redundant with the attack/release/RMS time "filters".)
I have done things in the past where I wanted to key an expander on one track from another, but I wanted it to kind of "smear", and there wasn't attack/release controls on the expander, so I put a reverb in the detector path. Worked out kind of cool.
A delay effect might create a sort of "bouncing gate" effect that could be cool.
I'm not sure how much use most modulation effects would be, but I guess a relatively resonant flanger or phase shifter could do...something...
Ooo... A very fast delay makes a strange sort of resonator which could be used to either accentuate or attenuate the signal when a given note comes into the detector.
Things can get pretty whacky fast, but it's all fun to play with.
Just one more quick thing - if you set all of the time constants (including RMS window) to 0, a compressor turns into a distortion. If you then run a second input into the detector path, it works almost like a ring modulator.