Jump to content
Sign in to follow this  
bazuka

Pyro clustering problems

Recommended Posts

Hi people, i have an urgent problem to solve and i have searched everywhere couldnt find anything about the problem i have, so:

i have an "rocket launcher" setup, simple circle as an emitter going up
i did a cluster setup, and simulations starts without problem, but in some moment when the circle emitter is, lets say 3 cluster container, for some reason pyro wont source the fuel from the object, and that 3rd cluster container just dissapear??? But when the circle emitter gets to the 4th cluster it starts to emit normally??? Im not sure if im doing something wrong or what, but help would be appreciated :)

 

p.s. rocket starts to go up >220frame

rocket.rar

Share this post


Link to post
Share on other sites

Sorry low on time but I hope I can get you a direction. (I didn't really see the setup much.

 

So I believe there's multiple things. Still figuring out the debugging process myself.

Your sources are likely clipping into your container when they are on an instance frame, and there isn't enough volume to what I presume is likely the voxel grid. so ex frame 300, your volume technically transferred to your cluster sourceframe and instantiated the obj for that cluster. But there was no data seen. so Frame 301 you have no grid and it's gone and refuses to look again for your fields. I'm not 100% probably but like an unsafe 80% lol sure if I'm right here, cause I don't know if increasing the padding to high amounts would help or not - certainly wouldn't be as optimized, though you could skew an axis in this example.

Basically I want to think your volume clipped when its got into the grid, so clipped to the point it had little data to be seen. ( I could be so wrong though , just has been my observations )

 

The solution consistently found is using the tracking object, which if I recall won't work if you're using all your obj containers/grids in the same sim.

 

So you'll need to wedge your sim for each cluster, so tracking object works properly.

on each cluster you'll want to get the current bounding box of where your objects are.

Which can be as simple as making a line of your path, and attr transferring where your rockets currently are and have been on that line, and blast wherever they aren't (just Cd comparison). And do a bounds on that path or whatever you're using, and that bounds you can add as much padding as you want to ensure to find your volume source, and still be optimized with gas resizing. This bounds of your path where your object has been will be your tracking object - for that cluster.

 

Also fun fact, that took me awhile to figure out I wasn't crazy but viewing multiple grid clusters, depending on your setup will re-adjust your visualization as containers are added and will see odd data of how your temp/density etc are behaving - or at least this was case for me.

 

Hopefully this is enough to help at least a little bit.

Edited by MaximumsProductions

Share this post


Link to post
Share on other sites

thx for replay mate,

 

if you have the time plz check the scene, i was trying to debug the cluster that is not sourcing the volume but i couldnt find what exactly is the problem... I was experimenting with cluster box sizes but same thing, sometimes cluster capture the source sometime not... And in the all tutorials related to clustering everything is so simple :) and working as expected....

Share this post


Link to post
Share on other sites

The problem I see with your setup is that your cluster areas are changing size. The approach that worked for me, in pre H17 was to create a line out of the points in motion. Then use that line to create cluster segments. Then generate points along the line as single frame bursts to activate the cluster when the fuel enters the boundary region.

Here is an old H16.5 setup that show particles leveraging clusters. Maybe this setup can help you fine tune yours?

Untitled-1.jpg

ap_particle_cluster_020719.hiplc

Edited by Atom

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×