Not able to visualize orientation with Quaternions in vtu files

ParaView Support
1

I am trying to visualize orientation of spherical particles. I have vtu files which has 4-component Quaternions in it, as can be seen in the screenshot. (or see the file: config_t90.vtu (1.3 KB) )

qtrnon
qtrnon816×205 15 KB

However, I am not able to properly visualize the orientation/rotations using it - it doesn’t change. I am using 3D Glyph for Representation, using Sphere as Glyph Type, Quaternion as Orientation Mode, and column “Quaternion” as Orientation vectors, etc.

tmp
tmp998×720 92.2 KB

Glyphs also looks really large in size. Setting radius and Scale Factor to 1.0 in Properties tab alone does not give true size. But setting raidus = 0.005 (true radius of particle in meter) in Display(GeometryRepresentation) tab makes it closer to true size. But still the overlap sometimes and is not perfect. What am I doing wrong here? Here is my state file for the same file : state_file.pvsm (296.0 KB)

2

Hi @pfahd

Orientation (with a quaternion or otherwise) on a sphere won’t give you anything interesting, it only make sense if you have a dinstinct shape.

Here is how it looks with a quaternion oriented cube:

image
image1584×1234 25.1 KB

Regarding the scaling, well it depends on your data. I’d assume you need a

  • Scaling=True
  • ScaleMode=Magnitude
  • ScaleArray=Radius
  • ScaleFactor=1
3

Hi @mwestphal , thank you for your reply. Is the image you generated using the data I shared? could you share the script please?

And about the sphere orientation, ParaView can actually do that. We just need to make low resolution Glyph. See an example below (here I used a vector to specify orientation)
test2

Another way is to color Glyph in different directions with different colors. See an example in my previous post. And about the scaling, when I use ScaleFactor=1, spheres are so big that they overlap each other. I am able to reproduce this still, using the same samples configuration files I shared. So, somehow my data needs a smaller value.

4

Here it is:
boxes.py (4.9 KB)

And with spheres:

spheres.py (4.9 KB)

5

Thanks a lot, it works perfectly!
I think the missing part was not using the filter ConvertToPointCoud. I have been doing the normal Glyph directly to the data. And again, using low resolution in Glyphs, I can see the rotation as expected!
WorkingQuat

6

Meanwhile, I had also got it partially working. Problem was, running the ParaView version I compiled locally ‘Orient’ button ( dg ) was not visible. Downloading the binary directly, fixed it, among many other weird things I had.

However, it was doing something weird. For the sake of understanding, could you please comment on what was happening there? i.e., applying Sphere Glyph, and changing/ introducing 3D Glyph only in Display (GeometryRepresentation) tab as: (that is by not doing ConvertToPointCould filter)

yfgyt
yfgyt426×116 7.57 KB

It produces something like below.. what is happening here? it looks like many small spheres combined. Otherwise, thank you again.
WorkingQuat3

7

Simple, do not use the “Glyph” filter, but use only the 3D glyphs representation.

8

oh! But shouldn’t I first apply filter ConvertToPointCloud then?
I tried selecting it directly under Display tab, and it shows nothing.
Like in the screenshot below:

para100
para100539×525 44.2 KB

What you see in pipeline is freshly loaded vtu files.
Just trying to understand how one can do it from the user-interface in proper way.

9

Yes, you should.