ChristopherC Posted May 6, 2015 Share Posted May 6, 2015 Heya, I'm currently trying to simulate a bunch of smoke trails flying at high speed in many directions and I would like to split the simulation into different DOP Networks to avoid computing a large container at once (my laptop is slow). It seemed intuitive to put my DOP Network inside a `ForEach` node but it doesn't seem to work and all I could find on the internet was a mention that DOP Networks obviously can't work within `ForEach` nodes. Why is that? How would I go to procedurally apply the same DOP Network a variable number of times as I've tried to do in the attached scene? Cheers! split_simulation.hipnc Quote Link to comment Share on other sites More sharing options...
Annon Posted May 6, 2015 Share Posted May 6, 2015 It's not going to work in a foreachSOP, have a look at clustering: http://www.sidefx.com/docs/houdini14.0/dyno/clustering Quote Link to comment Share on other sites More sharing options...
ChristopherC Posted May 7, 2015 Author Share Posted May 7, 2015 Thanks Christian, this link has been really helpful! I can't believe that I didn't stumble on it after all the time spent in the docs! I've attached the resulting scene—it allows to generate a variable number of animated spheres, each starting to emit smoke from a user-defined frame. I ended up not using directly the `Clustering Points` node since I intended to manually define the clusters, which means that I only had to update my graph to feed the actual feature that I was after: the smoke object instancing. Also, and if someone wouldn't mind explaining, I'm still curious about what's wrong with the `ForEach` approach? Could it theorically work if the object and data names were unique amongst all the iterations? split_simulation.hipnc Quote Link to comment Share on other sites More sharing options...
ChristopherC Posted May 7, 2015 Author Share Posted May 7, 2015 Ooops, the source velocities don't make it into the sim even though they seem to be fine at the SOP level. I'm a bit clueless as of why, any idea? Quote Link to comment Share on other sites More sharing options...
Annon Posted May 7, 2015 Share Posted May 7, 2015 I don't have time to look into your hip file at the minute, but make sure they're vel.x_[0-numcluster] etc... Each source corresponds to a different simulation domain. As for the forEachSOP, I don't know the technical answer and I might be completely wrong on this thought process, but the forEach would have to remain open in the session to work sequentially as a simulation. Each frame it would have to return to the iteration, pull the data from the last frame, progress forward etc... Don't think it really works like that. It just iterates and closes at the end, next frame it iterates then closes, so sequential solvers won't work. Or that's how I think about it anyway, might be completely wrong. Quote Link to comment Share on other sites More sharing options...
ChristopherC Posted May 7, 2015 Author Share Posted May 7, 2015 You were right—I manually named the velocity volumes “vel_#.x/y/z” instead of “vel.x/y/z_#”. Everything is working well now. There's still one weird thing which makes me wonder if I set up everything properly—Houdini crashes when I plug the `Gas Resize Fluid Dynamic` node in the pre-solve plug of the smoke solver, and the `Source Volume` in the post-solve, like it seems to be the case with the shelf tools. But I'll leave that one alone for now, I need to move on. Thanks again Christian! split_simulation.hipnc Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.