search

js1019 / PlanetaryModels

Build 3D planetary models on deformable meshes
GNU Lesser General Public License v3.0
19 stars 9 forks source link

Dead link: Tutorial video link for visualizing modes using ParaView #5

Closed j3soon closed 5 years ago

j3soon commented 5 years ago

In visual/README.md, there's a recommended tutorial for readers to follow, however, the link is unavailable... Is there any alternative link?

js1019 commented 5 years ago

In visual/README.md, there's a recommended tutorial for readers to follow, however, the link is unavailable... Is there any alternative link?

@j3soon Sorry about that. It was a neat video. I updated README. Please let me know if you have questions about the visualization.

j3soon commented 5 years ago

@js1019 Thanks for the update! However I'm still struggling to visualize it like the expected outcome in NormalModes:

I've tried using the models both in PlanetaryModels/visual/demos/ and the output of NormalModes. (converted to .vtk through the MatLab scripts) I've tried Glyphs and Streamlines with/without custom source, but both failed.

I would appreciate if there's some further tips or maybe even re-upload the video?

js1019 commented 5 years ago

@js1019 Thanks for the update! However I'm still struggling to visualize it like the expected outcome in NormalModes:

I've tried using the models both in PlanetaryModels/visual/demos/ and the output of NormalModes. (converted to .vtk through the MatLab scripts) I've tried Glyphs and Streamlines with/without custom source, but both failed.

I would appreciate if there's some further tips or maybe even re-upload the video?

@j3soon Are there any error messages? I simply use Glyph. If you can see clear patterns, that should be enough. Sadly, I can't find a similar video.

j3soon commented 5 years ago

@js1019 There are no error messages. I can see clear patterns but are not exactly the same as the expected outcome. They're quite similar, though.

image

js1019 commented 5 years ago

@j3soon It looks correct. The sizes of the arrows are quite large. You may play with the scaling factor to resize the arrows and the number of points needed.

j3soon commented 5 years ago

okay! Thanks a lot!

j3soon commented 5 years ago

@js1019 After fixing the size of the arrows and played with the option: Maximum Number Of Sample Points. I still have 3 questions about visualizing:

  1. The arrows still seems sparse comparing to the visualization example, regardless the high value and different settings of Maximum Number Of Sample Points.

  2. The visual I created seems to belong to spheroidal mode, how can I calculate toroidal modes?

  3. There is 1 extra sphere inside the toroidal modes, 2 extra spheres inside the spheroidal modes. Can you elaborate more on this?

js1019 commented 5 years ago

@js1019 After fixing the size of the arrows and played with the option: Maximum Number Of Sample Points. I still have 3 questions about visualizing:

  1. The arrows still seems sparse comparing to the visualization example, regardless the high value and different settings of Maximum Number Of Sample Points.

Please check the total number of vertices of your current model. Once you obtain results from a large-scale problem, you will see the differences.

  1. The visual I created seems to belong to spheroidal mode, how can I calculate toroidal modes?

Please check and visualize other computed modes. You will see toroidal modes as well.

  1. There is 1 extra sphere inside the toroidal modes, 2 extra spheres inside the spheroidal modes. Can you elaborate more on this?

Here is a sample script to create multiple interfaces: https://github.com/js1019/PlanetaryModels/blob/master/Demos/GravityETC/plot_interfaces.m

Please let me know if you have more questions.

j3soon commented 5 years ago

Thanks, I'll try them out!

j3soon commented 5 years ago

@js1019 I still have trouble creating multiple interfaces, the script you linked above seems to generate only a single sphere.

After some searching, I can only find the code snippet below that seems to create multiple interfaces (but actually doesn't):

https://github.com/js1019/PlanetaryModels/blob/a9b567e4e44bffe1daa129abfa84f0bfe364b909/Demos/PREM/PREM_mesh.m#L38-L48

I think I might overlook some details, can you explain more about creating multiple interfaces?

js1019 commented 5 years ago

I am not sure if I understand your questions correctly. To visualize multi-faces, you can simply add several single interfaces together.

j3soon commented 5 years ago

Oh, I think I overthought the meaning of the multi-faces... So the inner spheres are only used for indicating the size of the outer / inner core? (for better visualization?)

So, to decide whether showing both cores or only the outer core:

Is my interpretation above correct?

js1019 commented 5 years ago

Oh, I think I overthought the meaning of the multi-faces... So the inner spheres are only used for indicating the size of the outer / inner core? (for better visualization?)

This is correct.

So, to decide whether showing both cores or only the outer core:

  • For toroidal modes, only the outer core are shown, since all normal modes should lie outside the outer core (in the Earth's mantle).
  • For spheroidal modes, both cores are shown since the normal modes might originate within the inner core.

Is my interpretation above correct?

This is correct for simple models (spherically symmetric, or close to spherically symmetric). For a general model, it is not also the case. The patterns could be quite complicated.

j3soon commented 5 years ago

Got it! I have another question: In the figure, each visualized normal modes has its subscription n and l, which is the overtone number and the spherical-harmonic number respectively. The output of the calculated normal modes are labeled as the nth normal mode. Are there any way to map the nth mode to n and l in the figure?

js1019 commented 5 years ago

Got it! I have another question: In the figure, each visualized normal modes has its subscription n and l, which is the overtone number and the spherical-harmonic number respectively. The output of the calculated normal modes are labeled as the nth normal mode. Are there any way to map the nth mode to n and l in the figure?

There are rules to determine n and l but it will be better to use a radial solution. The spherical harmonic decomposition works well for spherically symmetrical problems. However, I don't think it is required for computer scientists. In addition, for the general cases, there are no clear n and l.

j3soon commented 5 years ago

okay! Thank you very much!