View Issue Details [ Jump to Notes ] | [ Print ] | ||||||||
ID | Project | Category | View Status | Date Submitted | Last Update | ||||
0002954 | ParaView | (No Category) | public | 2006-03-10 15:52 | 2011-01-13 17:00 | ||||
Reporter | Ken Moreland | ||||||||
Assigned To | Ken Moreland | ||||||||
Priority | normal | Severity | major | Reproducibility | always | ||||
Status | closed | Resolution | open | ||||||
Platform | OS | OS Version | |||||||
Product Version | |||||||||
Target Version | Fixed in Version | ||||||||
Summary | 0002954: SNL: PVR: Ordered compositing ruins normals at cuts. | ||||||||
Description | As part of its operation, ordered compositing cuts polygons. If normals are defined, the shading sometimes gets dorked pretty baddly at the place where the geometry is cut. Most likely, this is due to poor interpolation of the normals when the polygons are cut. | ||||||||
Tags | No tags attached. | ||||||||
Project | |||||||||
Topic Name | |||||||||
Type | |||||||||
Attached Files | badboxclip.pvs [^] (26,752 bytes) 1969-12-31 19:00 BoxClip.xml [^] (1,069 bytes) 1969-12-31 19:00 BoxClip.pvsm [^] (1,081 bytes) 1969-12-31 19:00 | ||||||||
Relationships | |
Relationships |
Notes | |
(0004361) Brian Wylie (reporter) 2006-06-26 16:31 |
Ken, please fix. :) (Brian Wylie) |
(0004870) Ken Moreland (manager) 2006-09-06 17:52 |
I have narrowed down the problem to vtkBoxClipDataSet. This is the class used internally by D3 to clip by the k-d tree (which is used internally by the ordered compositing). There appears to be an issue specifically when clipping a set of polygons that are not triangles. These are the type of polygons created by the tube filter and the cylinder source. The issue does not happen when the data is triangulated, nor does it happen on triangulated polygons that are not clipped. Although the issue appears to be related to triangulation, that is not sufficient. Clipping a plane does not seem to cause the problem. The actual problem may not be the triangulation itself but something that is tickled in the case where triangulation occurs. I have attached a state file that tickles this problem with vtkBoxClipDataSet. I also attached package files to load the necessary Box Clip filter that the state file uses. |
(0005043) Ken Moreland (manager) 2006-09-26 15:55 |
The problem should now be fixed. The issue was that vtkBoxClipDataSet was lookup up indices in the original cells using indices relative to the triangulated primitive, which of course, is wrong. To make the problem even more complicated, the interpolated values were sometimes interpolated themselves. The resolution was to perform interpolation using values already in the output data. The changed code is vtkBoxClipDataSet.h (1.9->1.10) and vtkBoxClipDataSet.cxx (1.20->1.21). I also added a regression test: BoxClipTriangulateAndInterpolate. |
Notes |
Issue History | |||
Date Modified | Username | Field | Change |
2009-12-09 14:51 | Berk Geveci | Project | @3@ => ParaView |
2011-01-13 17:00 | Source_changeset_attached | => VTK master a2bd8391 | |
2011-01-13 17:00 | Source_changeset_attached | => VTK master 020ef709 | |
2011-06-16 13:09 | Zack Galbreath | Category | => (No Category) |
Issue History |
Copyright © 2000 - 2018 MantisBT Team |