POP Streams Questions

[catchyid]

Hi,

I've watched https://www.sidefx.com/tutorials/particle-streams-1/ by Peter Quint on Streams, and I have the following questions:

-Streams are just a way to describe particle "states"? In the tutorial above, they are placed only at the top of different chains and then merged and connected to a POP Solver

-What's the use of the input connection of the Stream node, either way somehow it has access to the entire Particles in the DOP network?

-POP Group and POP Stream nodes have very similar interfaces, what makes them different? The only difference I can find is that the output connection of a Stream is those particles that are in this Stream, whereas the output connection of a Group is still the entire input particles?

-The order in which we when we merge multiple streams into POP Solver matters? Meaning, the POP Solver will execute the Streams branches in the same order it's fed in?

Thanks,

 

[Peter Quint]

Streams are essentially a convenience method. You could set up a group and then apply a following series of nodes to that group by setting the group parameter on each node and have the same effect as the stream. The stream is neater, that's all.

One use is as a state system, as the tutorial shows, but there are other uses. 

If I remember right the input comes into its own when you have more than one emitter in the sim. 

Yes the order of connection to the merge node does effect the order of execution. So if a particle is in two streams the ordering can have a significant impact. 

 

[catchyid]

Wow, I was not excepting Peter to reply  Thanks Peter for taking the time to reply. Sorry for the poorly written questions, I guess I was super tired when I wrote them

The dependency of execution on the order of connections is a little bit "strange" in my opinion, I "think" there should be other ways to make results less sensitiveness to order. Maybe using "Streams" as states make it look that way, because if a particle in a state, then its next state should be the same regardless of the order of execution. On a second thought, I guess it depends on how one designs his/her network, i.e. if transitions are independent then we should not have this order of connection dependency issue... 

Once more, thanks for your reply

[Peter Quint]

Yes the dependency on connection order seems odd, but actually it's true throughout Houdini, wherever a merge is used, or multiple nodes are connected to a single node the order of execution depends on the order of connection. So at least it's consistent with the rest of Houdini.