RealFlow Official Forum

Hello,

I know it's faster to simulate fluids via command line so I thought it was the same for the mesh generation process, but my tests show a different result: Meshing from within the interface using the Build Mesh Sequence button generates a new mesh in around 22-24 seconds, while using the command line takes around 2:02 minutes. Is this the expected behavior? or maybe I'm missing some settings to speed up the process via command line?

I'm meshing a Dyverso fluid with a VDB Mesh. I'm using Relaxation and Tension mesh filters, and Smooth volume filter.

Thanks for any help,
Orlando.

No, that's not the expected behaviour. I can only guess, but I seems that your CL simulation does not only do the meshing, but also the Dyverso part. Open the "Command Line Options" and check the "Mesh" otpion:

http://support.nextlimit.com/display/rf ... imulations and jump to "Meshing".

Thanks for answering Thomas,

Unfortunately I'm not simulating the Dyverso particles: I switched to cache the domain and also set the arguments to meshing. Is there something else that could be causing this behavior?

Orlando.

Very difficult to tell from a distance without having access to the scene.

Btw, you don't have to switch the domain to cache when you're just meshing. Meshing is kind of a playback job. Justing thinking loudly: what happens when you set the domain to "Active" again and try to mesh again via the CL?

I switched the domain to cache to ensure it would be ignored (to avoid any user error hehe).

I think I found the culprit: I had a second mesh in my scene. This one had visibility off, build set to no. I thought setting build to off the meshing would be ignored but seems like this setting has no effect when meshing using the command line (at least in my scene). I wasn't actually using this mesh at all, it was a previous test that I wanted to keep around just in case. I deleted the mesh and now I'm getting almost the same time per frame when meshing: command line 44 seconds (consistent in the 3 frames I tested) and from the GUI (disabling the viewports and using the "Build Mesh Sequence" button) test first frame took 1:15 minutes, second frame 43 seconds and third frame 40 seconds.

I wouldn't expect this behavior if one of the meshes was set to not build, but at least I have some similar results now.

Sorry for the noise,
Orlando.

Thanks for letting me know. In fact you're right: an inactive/disabled mesh shouldn't affect the meshing process and it sounds like a bug. I'll try to reproduce this error and file a bug report then.

Happy Simulating,
Thomas