Dear Experts,
I found my sphere become not such smooth after upgrade into paraview 5.9
i created my own sphere and calculated its normal vector, which works good in paraview 5.8, but after upgrade to paraview 5.9 it become useless, anything changed in paraview 5.9?
my sphere is added into multiblock dataset as a sub block and there other sub block like line, which dont have any normal vector, so my normal vector is partial normal vector from multiblock view, not sure anything with this?
Can you post the code so that we can try to replicate this?
Incidentally, there is an issue with GenerateSurfaceNormals but it affects 5.8.1 and 5.9 so it seems unrelated.
https://gitlab.kitware.com/paraview/paraview/-/issues/20651
Thanks,
Dan
This turned out to be user error. There is a parameter to the GenerateSurfaceError that causes those edges to be preserverd.
Hi Dan,
Thank you for your reply!
i created my own sphere follow the instruction from below:
http://www.songho.ca/opengl/gl_sphere.html
This works good in paraview 5.8.1 but after i upgrade to paraview 5.9 seems these normals failed to work as intended:
every point’s normal vector should from sphere center to point, not the normal of those triangle plane,right?
Hi Dan,
What do you mean “this is user error”, and where can i found the parameter
Can you give me more detail or further explanations?
Thank you!
Hi Sharon,
The link to the bug report I sent (now closed) has more information about that. Cory pointed me to the Feature Angle parameter that causes the sphere edges to be visible in the examples I gave. The Splitting parameter needs also to be on which it is by default.
HI Sharon,
The exact code would be needed to reproduce the behaviour you are seeing.
The normals are specified with one point and they represent the vector from the origin (center of the sphere probably) to that point. They do represent the normals of the triangle planes.
Is this a programable filter? Can I have that so that I can just plug it in paraview and try it? It would be the easiest to track any changes from 5.8 to 5.9.
@sharon It is not clear what you are doing here. Are you producing this dataset in a plugin or in an external piece of code and save it as a VTK file or a compatible file format? If so, could you share this file so we can check what is going on here?
The fact that the sphere mesh is part of a multiblock might be an interseting lead here as some changes occured between 5.8 and 5.9.
Hi Dan,
Thank you for you reply and help!
I am on a trip and sorry for replying late.
My code is coupled with other part, it is not easy to split them as a whole that can run separately.
However, after my local test, it seems because of multiblock reason:
sphere.obj (237.2 KB) line.obj (155 Bytes)
Hi Joachim,
Please see my reply to Dan, you are right, it seems after group sphere and other data together as a multiblock dataset, then its normal vector become useless…which is not such behavior in paraview 5.8…
Sorry for my late reply because i am on a trip…
Hi Sharon,
Indeed this seems to be a bug. I filed it here:
Thank you so much, Dan!
