Hi,

I have been trying very hard to achieve this simulation in Popnet but to no avail.

http://www.youtube.com/watch?v=9U5_A4eNzCE [youtube.com]

Basically i want to:
1)Make the particle flow exactly along the net shape curve
2)The particle is being attracted to the intersection points
(instead of Creep>UVposition where no forces is involved)
3)Everytime the particle hit the intersection, it will die. Subsequently a new set of particles will birth from the spot and each of them take one path, being attracted to the next nearest intersection.
4)It is exactly as what is shown in the video.

I hope the solution will not involved VOP POP as it i am not very good at it and probably cant understand.

Wonder if this is too complex to be done? Will appreciate any help even if you could only get step 1 done.

Thanks in advance.

Try an option where you don't use particles.

Build your curves. Use Carve SOPs and extract the points.

Hi Jeff. Thank for the reply

I wanted to do this with particles into order get information (eg.numhit) of the intersection. In other words, it is like testing a network/traffic to see if the particle could reach the end of the net and to see which intersection point is more “popular”.

The information will be translated into something else later on.

And there is animation for the particles which makes the presentation of information more appealing i guess.

Here's a start using particles.
Unfortunately my neural net seems to resolve to find primitive 0. Trivial logic in the Attribute Transfer POP.

Need to add more logic to the attribute transfer or some other mechanism to index randomly each point in to what edge.

Without using VEX/VOPs you will have your work cut out for you.

Thanks Jeff that was a great start.

I am trying to solve the $EDGE_DIR problem using Python Script for now.

Do you mean you know how to use VOP to get the result?

Yep.

With VOPs you can pass the neural net in as a point cloud in to your VEX POP. As your particle locates a node (bunch of points), you figure out how many points are there, exclude the one that sends you back to where you came from (dot product of v vs. the direction vector equals 0) then randomly choose one of the points using a random function, mark it as “done” with a custom attribute so it doesn't get processed by the same node over and over then send your particle on it's way along the new edge.


-jeff