View Issue Details Jump to Notes ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0016745ParaView(No Category)public2016-06-13 13:022016-08-11 13:52
ReporterKen Sheldon 
Assigned ToT.J. Corona 
PrioritynormalSeverityminorReproducibilityhave not tried
StatusclosedResolutionfixed 
PlatformOSOS Version
Product Version5.0.1 
Target VersionFixed in Versiongit-master 
Summary0016745: Render error in unstructured grids
Description
See attached data for PVSM and PVD files. This is an unstructured 2-D grid with cell-centric properties IntValue and FloatValue.


Paraview 5.0.1 and 5.1.0-RC1 render incorrectly: cells are not colored solid; instead they are bisected and shown as two different colors.

Paraview 4.3.1 and 4.4.0 render correctly.
TagsHackathon-July-2016
ProjectTBD
Topic Name
Typeincorrect functionality
Attached Fileszip file icon Subset.zip [^] (20,450 bytes) 2016-06-13 13:02
? file icon Sample_0_cleaned.vtu [^] (24,933 bytes) 2016-07-13 16:01

 Relationships
related to 0016746closedKitware Robot Cell centers calculation error 

  Notes
(0036362)
Andy Bauer (developer)
2016-07-13 11:35

It looks like these cells that are rendering weird have duplicate points. I checked this by extracting one of the cells that is rendering weird and then looked at the extracted points. If you create the cells without duplicate points do you get the same behavior?
(0036413)
T.J. Corona (developer)
2016-07-13 16:04

I agree with Andy. Running Extract Surface, Clean, and Append Datasets performs the following: 1) converts unstructured grid to polydata, 2) cleans polydata, 3) converts polydata to unstructured grid. I used this pipeline to generate Sample_0_cleaned.vtu. It has 30 fewer points and renders correctly.
(0036472)
Ken Martin (reporter)
2016-07-14 14:51

I dug into this and the issue is that there are triangles that have three vertices but where two of the vertices have different ids but the same exact coordinate making them degenerate (as opposed to say a two point "triangle" or a triangle with duplicate point ids) It seems that OpenGL does not increment gl_PrimitiveID for such triangles or it does so inconsistently. Checking for and eliminating these when building the IBO solves the issue but does require analyzing the points when building the IBO making that process slower.

I'll push a topic with that change up and we can discuss.
(0036526)
Ken Martin (reporter)
2016-07-18 14:44

See https://gitlab.kitware.com/vtk/vtk/merge_requests/1714 [^]
(0036549)
Alan Scott (manager)
2016-07-22 19:19

Asked Utkarsh to test and if appropriate close. I don't know enough about what it should look like to do so.
(0036691)
Utkarsh Ayachit (administrator)
2016-08-11 13:52

Tested using attached dataset (Sample_0_cleaned.vtu).

 Issue History
Date Modified Username Field Change
2016-06-13 13:02 Ken Sheldon New Issue
2016-06-13 13:02 Ken Sheldon File Added: Subset.zip
2016-07-12 11:31 Utkarsh Ayachit Tag Attached: Hackathon-July-2016
2016-07-13 11:35 Andy Bauer Note Added: 0036362
2016-07-13 16:01 T.J. Corona File Added: Sample_0_cleaned.vtu
2016-07-13 16:04 T.J. Corona Note Added: 0036413
2016-07-13 16:11 T.J. Corona Assigned To => T.J. Corona
2016-07-13 16:11 T.J. Corona Status backlog => customer review
2016-07-13 16:11 T.J. Corona Resolution open => fixed
2016-07-14 08:48 Ken Sheldon Relationship added related to 0016746
2016-07-14 14:51 Ken Martin Note Added: 0036472
2016-07-18 14:44 Ken Martin Note Added: 0036526
2016-07-22 19:19 Alan Scott Note Added: 0036549
2016-08-11 13:52 Utkarsh Ayachit Note Added: 0036691
2016-08-11 13:52 Utkarsh Ayachit Status customer review => closed
2016-08-11 13:52 Utkarsh Ayachit Fixed in Version => git-master


Copyright © 2000 - 2018 MantisBT Team