Thanks, but my remark was also meant as more of a general observation:
posting ICE Trees that are unreadable is not very useful IMHO...

Its just a POC (proof of concept)...but that's how it will work. The null will trigger the state change...then whatever effect we want to do will be in State1..State2..State3..etc.

Easy as pie from there.

Probably it would it be possible to grab keyframe animation also - Like having SRT animation defined for frames 1-100 of a reference object, then look how much an individual instance is affected by the null. If it's 33% for example, it spools it to frame 33 of the reference animation. This would be nice for some great effects although I'm not sure right now whether motionblur would work with that.

yep Its basically keyframe transfer with offset from reference object

Not sure if it's even possible to get the animation data from a certain frame. Probably using cache arrays but it would be better obvioulsy to have a live operator here.

Best way would be to have direct access to the animation fcurve in ICE -- Is this possible? -- a good start would be to be able to create an fcurve programatically in ICE.

something like this?

guys thanks a lot for effort and help, would it be possible to have a screenshot of the tree cause i cant really figure out how to make it to work, i've tried with the nodes and what iamvfx said but it doesnt work.
Sorry ice beginner here :/

thanks, Max

@Tekano it coming close, built-in bezier nodes are great ("Fit Bezier" even does interpolation through knot points!). But having full native fcurve support (including access to all scene fcurves) would still beat it I guess - a setup like shown above would be much easier to create then.

I'll be posting several video tutorials this weekend. I brought my laptop to work so I can finish up some example scenes. There's some missing pieces from that screenshot, so you wouldnt be able to doing anything with JUST that.

Eating lunch, working on laptop, figured out how to do his last trick with the disappearing pieces....we can use the TimeInState property, so as the null [or some other trigger] switches the particle from its current state to a new state, we can evaluate the TimeInState property to determine when the effect should trigger, in this case, a disappearing effect.

Dude, I'm telling you, ICE is sweeet!