vedo, formerly known as vtkplotter

kinnala / scikit-fem

Simple finite element assemblers

https://scikit-fem.readthedocs.io

BSD 3-Clause "New" or "Revised" License

512 stars 80 forks source link

vedo, formerly known as vtkplotter #361

Closed gdmcbain closed 3 years ago

gdmcbain commented 4 years ago

As noted in #44, matplotlib distorts three-dimensional views uncontrollably. Some very nice plots were added to #357 using vtkplotter. It might be a worthwhile addition to skfem.visuals #286. (I haven't looked at the licence or portability.)

kinnala commented 4 years ago

License seems to be MIT. I don't oppose this, the plots look great. I'll try to get it working on my Python setup and try it out.

gdmcbain commented 4 years ago

A quick experiment went well.

On a laptop running Microsoft Windows 10

pip install vtkplotter

Then after having run ex32.py from #342, run

from vtkplotter import Plotter

vp = Plotter()
vp.load('ex32.vtk')
vp.show()

This pops up a 'Visualization Toolkit Win32OpenGL' showing a green triangulated ellipsoid which can be rotated, translated, and zoomed with the mouse.

Hit 'h' in this window to display help back in the terminal.
Hit '4' to cycle through scalars to colour by; here twice for pressure.
Hit 's' to save screenshot.png.

screenshot

This does nicely show the p ~xy pattern expected from creeping convection in a horizontally heated ellipsoid.

I assume this could be scripted, once I read the docs.

gdmcbain commented 4 years ago

To generate a PNG rather than a window,

from vtkplotter import Plotter, screenshot

vp = Plotter(offscreen=True)
...
screenshot('ex32.png')

See marcomusy/vtkplotter#55.

That saves the PNG without throwing up a window; however, it doesn't colour by pressure. The answer to that might be in marcomusy/vtkplotter#131.

gdmcbain commented 4 years ago

Here we go:

from vtkplotter import Plotter, screenshot

Plotter(offscreen=True).load("ex32.vtk").pointColors("pressure").show()
screenshot("ex32.png")

gdmcbain commented 4 years ago

I'd deferred this as it wasn't working on my Ubuntu 19.10 desktop (something about nvidia nouveau); as mentioned the previous work had been done under Microsoft Windows 10.

Yesterday I updated the desktop to Ubuntu 20.04 LTS and it's working fine. Here's a recipe.

conda 4.8.3 # from Miniconda3-latest-Linux-x86_64.sh
python=3.7 in a conda create # gives Python 3.7.7
conda install vtk
pip install vtkplotter

kinnala commented 4 years ago

I've been encountering issues trying to run this in a docker container, I'm only developing inside containers nowadays.

I think I have to next try it out in a Ubuntu 20.04 virtual machine.