Thanks! So Point Sop is a no-no? (Tutor says VEX is 10 times faster than PointSOP) Every single tutorial im watching uses point sop with expressions, why is that? Wranglers / VOP SOP basically make Point sop obsolete (because they are faster)? Wouldn't it be better for future versions of H to reduce the redundant stuff like pointsop?
Piotrek Marczak Thanks! So Point Sop is a no-no? (Tutor says VEX is 10 times faster than PointSOP) Every single tutorial im watching uses point sop with expressions, why is that? Wranglers / VOP SOP basically make Point sop obsolete (because they are faster)? Wouldn't it be better for future versions of H to reduce the redundant stuff like pointsop?
Tutorials are for learning the software and point sop makes it easier to learn Houdini.
This is how Ive done it in ICE, but obviously more atomic (lack of hi level nodes). This is how I like to work. On the lowest level possible and in one graph. Thanks Scott
sakeating If you're interested in a pure POPs example of this type of particle simulation, I've created a .hip file with a few sticky notes to help explain what is happening. This uses the NewPops setup developed for H13.
I attempted to replicate what I thought you were describing ( Clusters breaking apart and leaving streams ).
As was mentioned earlier, there are many different approaches to solving these types of problems in Houdini.
With regard to ‘high level nodes’, you'll find that with the NewPops implementation of particles, just about every PopNode is actually a Digital Asset which contains the lowest level nodes, generally Geometry VOPs for forces and Sop Solvers ( Geometry creation nodes ) for generators.
So, if you're interested in how things are implemented at the lowest level, dive inside the PopNodes themselves and dig around! This way you're easily able to make custom adjustments or learn how to create your own high-level nodes from scratch.
sekow This is how Ive done it in ICE, but obviously more atomic (lack of hi level nodes). This is how I like to work. On the lowest level possible and in one graph. Thanks Scott
sakeating If you're interested in a pure POPs example of this type of particle simulation, I've created a .hip file with a few sticky notes to help explain what is happening. This uses the NewPops setup developed for H13.
I attempted to replicate what I thought you were describing ( Clusters breaking apart and leaving streams ).
As was mentioned earlier, there are many different approaches to solving these types of problems in Houdini.
Let me know if you have any questions!
—————- Scott Keating Product Designer SideFX Software —————-
sakeating No problem, I'm glad you found it useful.
With regard to ‘high level nodes’, you'll find that with the NewPops implementation of particles, just about every PopNode is actually a Digital Asset which contains the lowest level nodes, generally Geometry VOPs for forces and Sop Solvers ( Geometry creation nodes ) for generators.
So, if you're interested in how things are implemented at the lowest level, dive inside the PopNodes themselves and dig around! This way you're easily able to make custom adjustments or learn how to create your own high-level nodes from scratch.
sekow This is how Ive done it in ICE, but obviously more atomic (lack of hi level nodes). This is how I like to work. On the lowest level possible and in one graph. Thanks Scott
sakeating If you're interested in a pure POPs example of this type of particle simulation, I've created a .hip file with a few sticky notes to help explain what is happening. This uses the NewPops setup developed for H13.
I attempted to replicate what I thought you were describing ( Clusters breaking apart and leaving streams ).
As was mentioned earlier, there are many different approaches to solving these types of problems in Houdini.
finally found some time revisiting scotts example. I must say I am a bit confused. It seems that the point clusters is created every frame. by copying the initial point location and deleting the source point.
that confuses me, as I am under the assumption that the cluster has to be created every frame. Guess I have difficulties on how the data stream is working.
In ICE I've created a cluster on (for example) the first frame. So I had a fixed number of points per cluster. Than, similar to the i@active branch, forced the detached points, after a random timevariable, into a different force field.
So you've got it right! If you look at the ‘location’ Pop at the top of the network, you'll notice that the ‘impulse activation’ parameter is set to ‘$FF==1’, which just means that it is only active on the first frame.
On that frame, it creates 20 particles.
Then Pop Replicate takes each of those particles and copies of a cluster of 1000 particles. So, for each of the original 20 particles on frame one, you get 20 clusters of 1000 particles. The original 20 particles are removed by Pop Replicate, because ‘kill original particle’ is turned on.
Now, because the original ‘PopLocation’ node only cooks on the first frame, the replication only happens on that frame as well. This gives you a single burst of clusters on the first frame and that's it. This is how you get consistent numbers of particles from frame 1 onward.
Make sense?
I just realized that my original file had an error, which might be confusing you. On the ‘popreplicate1’ node ( in the @active chain ), the connection should be on the RIGHT input, the reference input, not the left. This was causing an increase in points.
—————- Scott Keating Product Designer SideFX Software —————-
I get that the location node is evaluated just on frame 1. But the other nodes, I am not that clear. The point Id's don't change so we are getting the same value for the activate variable. But is this also set just on the first frame? Or on every frame, but it does not matter because it is not time depended?
Also the second replicate node confuses me. Diving in, I see that the stream is flowing into a null object (ref input). So in my understanding it is just a pass through. But somewhere before that and invisible to me, the points get filtered by the active group. It is obvious that this second stream is for the dissolved points. Again confusion about why this is actually working.
And at last, it seems that the points in the active group are solved two times. One time in their separated filtered stream after the second replicate node, and than again directly after the group node. Or am I wrong?
I am pretty sure that I have to get used to some fundamental concepts, but for now its hard to read. The Masterclasses are fantastic btw.
Yes, you're right. There seems to be an issue with the way ‘kill original particle’ is being handled by the Point Replicate. The cook order seems to be incorrect.
If you look at the file I've included, you'll see it fixes the issue by doing the first replicate/merge manually ( instead of the implied merge of the first input of the pointreplicate SOP ).
You'll see the correct number of points are created. The first merge node's order of operations is important.
I'll follow up with the developers to see exactly what is happening.
Thanks! -scott
sekow
sakeating Huh, curious.
I'll take a closer look tonight.
In the meantime, if you'd like a more detailed breakdown of the new Pops, you can check out this Masterclass by Jeff Lait.
finally found out how to generate and feed fields with particles. see 1st attached hip file, where I generate a velocity field for later point advection. problem is that it the field generation won't work anymore as soon there is a second pop network..
see attached files, the first shows a functional particle to field setup the second how I think it should work
