search

visit-dav / visit

VisIt - Visualization and Data Analysis for Mesh-based Scientific Data
https://visit.llnl.gov
BSD 3-Clause "New" or "Revised" License
438 stars 116 forks source link

Index Select on Particles #12107

Closed brugger1 closed 3 years ago

brugger1 commented 4 years ago

From: Allen Sanderson allen@sci.utah.edu Sent: Friday, March 13, 2020 11:12 AM To: Tony Saad tony.saad@chemeng.utah.edu Cc: Biagas, Kathleen biagas2@llnl.gov; Maguire, Alister Owen maguire7@llnl.gov Subject: Re: Index Select on Particles

Hi Tony,

I just tried using the index select on some particle data and while it does do the index select the scalar values are lost and the particles are black. I have attached the original and with the index select. The legend is the same but the values go from 0 to 1. Which tells me the point/cell data is not getting on to the points.

I have not looked at this filter in quite sometime. I am cc’ing two of the VisIt developers to see if they have any immediate ideas of what may be the issue. Otherwise a bug report will need to be submitted.

Cheers,

Allen

Allen Sanderson SCI Institute University of Utah

On Mar 12, 2020, at 9:41 PM, Tony Saad tony.saad@chemeng.utah.edu wrote:

Hi Allen,

In the past, I used to be able to add an "Index Select" operator to a particle plot so that I can skip every two particles for example... I can't seem to do that anymore. Any ideas?

-- Tony Saad Assistant Professor of Chemical Engineering University of Utah 50 S Central Campus Drive, MEB 2286 Salt Lake City, Utah 84112 Tel: 801-585-0344 www.tonysaad.net

indexselect_bug1

indexselect_bug2

brugger1 commented 4 years ago

I'm not familiar with that issue. I'll look into it. It will probably be Monday.

markcmiller86 commented 4 years ago

I will try to take a quick look

markcmiller86 commented 4 years ago

@ARSanderson I am unable to reproduce with 3.0.1. Here is what I tried...

  1. Open noise2d.silo (one of our silo_hdf5_test_data files)
  2. PC plot PointVar
  3. Add index select, set to 1D and incr to 2.
  4. Draw. It looks good. No all black
  5. Try switching point type to sphere. It still works

Honestly what it looks like in @ARSanderson example above is that the point density is so high, maybe there is some other odd effect going on. If point density is reduced substantially, does the all-black behavior go away?

Untitled

brugger1 commented 4 years ago

What version of VisIt are you using and what did you use for the point glyph type? It looks like it was run on a Mac.

markcmiller86 commented 4 years ago

What version of VisIt are you using and what did you use for the point glyph type? It looks like it was run on a Mac.

Is this a question for @markcmiller86 or @ARSanderson?

brugger1 commented 4 years ago

For Allen. You stated you were using 3.0.1 and from the image I can see it was running on a mac.

ARSanderson commented 4 years ago

I am running on 3.1 on OS X. The original image was with point glyphs.

I just switched to spheres and got the following image. The square in the middle is perspective bounding box. When I have seen gooblygook like this in the past we figured out that double arrays were being stuffed into float arrays or visa-versa. This issue has nothing to do with the index select though I though it got fixed.

Cheers,

Allen

On Mar 13, 2020, at 1:31 PM, Eric Brugger notifications@github.com wrote:

What version of VisIt are you using and what did you use for the point glyph type? It looks like it was run on a Mac.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/visit-dav/live-customer-response/issues/14#issuecomment-598879840, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGHBOIKXP4PNE72XXK5KELDRHKCYTANCNFSM4LHIMURA.

ARSanderson commented 4 years ago

I figured out the issue. From: https://vtk.org/doc/nightly/html/classvtkMaskPoints.html

vtkMaskPoints https://vtk.org/doc/nightly/html/classvtkMaskPoints.html is a filter that passes through points and point attributes from input dataset. (Other geometry is not passed through.)

Emphasis added. This issue is the same as what we discovered within VisIt proper, that is when it comes to a mesh that is a VTK_VERTEX, the data can be either point (node) or cell (zone) centered. It does not matter. Regardless both point and cell attributes should be passed through. The vtkMaskPoints is not passing through cell data. That is the issue and is a VTK bug.

Given that I have fix for Uintah reader I am going to change vertex data to node center to get around this issue.

Cheers,

Allen

Allen Sanderson SCI Institute University of Utah

brugger1 commented 4 years ago

Thanks for the update. I'll close this issue since you have a work around.