Isosurface rendering artifacts with HDF5 data (VisIt 3.1.2)

[yuheng-guo]

Hi VisIt Developer Group,

I’m seeing persistent rendering artifacts in isosurface plots from HDF5 data. In some views, parts of the expected surface appear to be missing, and regenerating the plot does not resolve the issue. Issue seems only happen in certain frames. I don’t see any obvious errors in the VisIt output logs. I’m currently using VisIt 3.1.2.

Two images are attached for reference. First one is normal and the second is problematic (you can see through the yellow density region). Any suggestions on what might be causing this, or what I should try next, would be greatly appreciated.

Thanks!

[JustinPrivitera]

Hi @yuheng-guo. When you say different views, do you mean after rotating the viewer window? Or at different timesteps? How are you generating these plots?

You could try turning on depth cueing in the advanced rendering settings.

You may also try using a newer VisIt. The current version is 3.4.2; 3.5.0 will be releasing shortly.

[yuheng-guo]

Hi Justin. "views" mean different timestamps. I'm generating these plots with python scripting like: visit -cli -nowin . Can you be more specific on how depth cueing can help? Since from my view this looks like visit is failing to generate some isosurfaces. Can you suggest other debugging options other than using a newer version of visit?

[JustinPrivitera]

Hi @yuheng-guo. Thanks for clarifying. I was wondering about any specific steps you might be taking to make these specific plots? It's hard to know what tips to suggest without more information. Depth cueing may have an effect; it may not. It may be that different depths are competing for what is rendered in front. It's worth trying, but I can't guarantee it will solve your issue.

VisIt 3.1.2 is pretty old and there is a chance that whatever issue you are facing has been resolved in the years since. But I understand that it may not be convenient to upgrade.

You could share your data and instructions on reproducing these plots with me and I could take a look in the newest VisIt version.

[JustinPrivitera]

@yuheng-guo any update on this issue?

[yuheng-guo]

Hi Justin. Sorry for the delayed reply. We did narrow down the issue to a run script that launches visit, and by some experiments with that script we were able to fix the issue without moving to a newer visit. Thanks for the help and suggestions!

[JustinPrivitera]

Excellent; I'm glad you got things working!