Obstacle Interaction with Spermlike Flocking

[shushens]

Hi Softimage Lovers,

I have been trying a not-so-simple ICE particle behaviour where some fishlike creatures would show a certain flocking behaviour and also react in a natural fashion when close to certain obstacles or interest objects. I have set up a few goals and they are flowing along nicely. Now to add some character to it, I am trying to use forces like coagulate so that they repel each other if they come too close and attract each other if they go too far. I know about Kim Aldis's flocking test, but here the movement is not random, there is almost a certain path or collective goal that the fishes follow.

Strangely enough, when the goals are set, several forces, especially Coagulate, are totally ineffective on the particles. No matter what settings I try, nothing is happening. Is there something I am missing here? Attract/repel by surface seems to work fine. Gravity is also working, but no coagulate or turbulence.

Any rule of thumb I should follow here to get things right?

[Hirazi Blue]

You probably tried this already, but anyway:
First guess would be to lower the Weight of the "Move toward Goal" compound and maybe increase "Attraction Strength" in the "Coagulate Force" compound. But maybe the order in which the force and the goal are evaluated within the ICE Tree would make any difference... (I'm guessing: force first, goal second...)

[owei]

Hi..!

Move towards goal is not a "force", there fore it SETs the new particle velocity based on certain parameters. Hirazi is right concerning that the order plays a certain role.
If you are using this compound AFTER you brought in some other forces, the velocities they produce will just be "overwritten".. You could try to change the order of your setup, that "move towards goal" comes first and do some other "force"-calculations afterwards..."goal" first, "force" second ;)
Otherwise you will have to do your own "goal"-force and use this instead of the "move towards goals" compound..

hope this helps,
cheers,
oliver

[shushens]

Thank you very much, Hirazi and Owei
I have another question.

Say, a lot of particles are flowing to a goal in a relaxed manner like a school of fishes.
When they get close, they circle around the goal a few times, but every cycle should not be identical (which is why it cannot be curve based).
I wanted to make several goals around the interest object and make the particles choose them at random using a multi-goal sequencer. But that is looking too mechanical. It's not like living organisms roaming around an object of interest.

Sorry if the question is vague. My interest here is casual interaction with some flocking behaviour.

It reminds me of the 'arrive' mode of the classic particle goal in XSI.

[owei]

...hm, that´s tricky if you will use the "move towards goal" compound, because, as I said already, the will SET new velocities and orientations. You can try to get a more natural look limiting the maximum turn rate and limiting the acceleration. So the actual velocity and orientation will be taken in concern for calculation. This might help. Using forces that can be blended would give a better result, but you will have to set them up on your own...

EDIT: values around 0,1 for the limits seem to work. Also adjust the "Velocity over distance" curve to get a more steady motion...
cheers,
oliver

[shushens]

What if I do not have to use 'move towards goal'? I am just a n00b in ICE and that node seemed to have made sense. But I do not really have to use it.

[shushens]

Here I have tried something different. Could not get rid of MOVE TOWARDS GOAL entirely, but broke things up in states.

State 0: particles rush toward to the goal. As soon as they get close, State 1 is triggered.
State 1: particles ORBIT AROUND CONTROLLER for 15 seconds. Then State 2 is triggered.
State 2: particles move away.

When the particles are shifting from state 0 to state 1, the ICE strands are not bending naturally. It looks artificial. If it was some force, a lot of gradual variation would be possible. But ORBIT AROUND CONTROLLER makes it complicated. I haven't tried to fix it yet because I am not sure if I will be able to keep this node in the final tree, because here they are getting totally scattered and I want them to somewhat hold the line.

So, either the orbit function has to go, or the strands need be modified. Or maybe something else will solve things entirely. Please help me with your insight.

Since I want to put the goal (the ball, which I will eventually replace by a logo) in the midst of other obstacles, I have used a FLOW AROUND SURFACE here on an obstable (shown below).

Clipboard02.jpg

Clipboard01.jpg

[Hirazi Blue]

Move towards goal is not a "force", there fore it SETs the new particle velocity based on certain parameters. (...)
If you are using this compound AFTER you brought in some other forces, the velocities they produce will just be "overwritten"...

I just have been playing with this some more and found out the owei's above statement isn't 100% correct. The "Select Speed Method" parameter of the "Move Towards Goal" compound is set to "Set New Velocity" by default, you can however choose one of the two other options (i.e. "Multiply by Existing Velocity" or "Multiply by Emit Velocity"). "Multiply by Existing Velocity" DOES consider the active forces, although it's no fun to tweak it together with the applied forces just right. ;)

[Tekano]

Hi to get around this in the past ive used states 'inbetween' the states you have and these are used to aid the transition. Cant remember the details off the top of my head but its something like using a node called 'test time in state' to drive the blend a 'linear blend' node which would be blending the point positions of the before and after of where you want the particles to be.

[shushens]

Very valuable piece of advice! I shall look into it right now. Many thanks!