In the end you'll have instances not meshes being driven by particles?
The other way around, now I have Meshes, not Intances, driven by particles. Each Agent is a duplicated Mesh. They all have kinda similar Topologies, but each is procedurally.. more or less.. generated based on Parameters passed from the Initial ICETree at time of birth of the particle.
What kind of data you want from objects that doesn't go already?
Typically if you animate something in original object it is replicated in instance with Control Instance Animation node.
Well typically all you can control about crowds when using Particles is:
Transform (of the whole object)
Local Animation Frame (just an integer for scrubbing the per-particle timeline)
Color(only 1 color for the shader.... what if you want to randomize 2 colors? like pants and shirts...)
The reason for this limitation is optimisation. The Renderer just caches the geometry and reuses for each Particle.
But if I understand animation correctly, the instance will look exactly like the original animation. What if additionally to the walk cycle that you animated using key-frames, you want each Instance to:
- Turn their head towards a null?
- Have different number of arms sticking different places on the body
- Have a difference in behaviour. Like a "interested" parameter which controls how much that guy is willing to turn his head towards the null to see it, if at all
- If it has legs, what if you want the legs to actually stick to the ground?
you get my point
