Capturing PyVista's user stories #2133
Replies: 19 comments 36 replies
-
|
what is your domain/background? I have a background in mathematics and computer science. Last year was more heavily 3D focused, with a major project looking at gravity inversions and the analysis of, so 3D capability was crucial there. Basically, whatever I needed to do, PyVista provided - quick visualisations for overviews, for conversations, for reports. The ability to take in data from elsewhere and analyse, with a bit of scripting Unlike some desktop software application, PyVista's pythoneseque api let it be plugged into pipelines, whether manual or automated on AWS. Also doesn't cost 10 grand a year for a licence where the program stops operating when you are not connected to the internet, or a vpn, etc. So more capable, more efficient, more flexible, for my use case. Which also leaves more money for science - [in this case, computational geophysics] Filters like slicing, thresholding, connected body analysis were very handy. The data analysis capabilities are the crucial part, to enable building other sorts of models. You can also quickly add some geospatial location to something like mtpy's model outputs in model space for overviews. A growing number of broad scale inversion models are being made available in Australia, so PyVista being able to handle this sort of data on the desktop saves time. When I was building a higher resolution geology model of South Australia - I could look at it. what is your familiarity with Python and the Python Data Science Stack? Several years. what is your experience level with 3D visualization? Used it occasionally in the past. More critical now. have you used other 3D visualization software in Python or otherwise (open source or otherwise)? Yes, Blender, Paraview, some things I don't remember the names of over the years, domain specific geoscience software. what was your experience between these software? Significant learning curves to work out. what are your primary visualization tools? At the moment? Python and R, with QGIS and Geoscience Analyst for domain specific geoscience things. PyVista <=> geoh5py is very handy. The data is the important thing and the analysis of such, visualisation is a byproduct of this to see that it makes sense. So tools that are designed for clicking and making bespoke models slowly over a long period of time aren't so useful. where would you like to see PyVista go? Distributed possibilities? Something like Holoviews/Napari? Allow higher resolution work at scale. |
Beta Was this translation helpful? Give feedback.
-
|
This is awesome, thanks @RichardScottOZ!! I LOVE seeing the geological models and density models plotted in 3D with PyVista! I really appreciate your summary here and I'm thrilled that PyVista has had such a positive impact on your work, allowing you to not only focus on the science at hand, but afford the science at hand.
Would you elaborate on what you mean by this? |
Beta Was this translation helpful? Give feedback.
-
|
200m resolution 3D model of all of Australia - or Brazil, handle that sort of size - things one machine can't do, if that makes sense? Billion scale stuff and up. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, without all the amazing pyvista work, this would have been harder and taken considerably longer. |
Beta Was this translation helpful? Give feedback.
-
My work is in earthquake engineering. I use the finite element method and vibration analysis methods such as FFT as numerical analysis methods.
I am a big fan of Numpy, Scipy, and Matplotlib, and I use GetFEM for finite element analysis in Python (Fenics may be more major now).
GetFEM (and other OSS finite element software) outputs 3D data in VTK format, and while Matplotlib is useful as visualization software, it is not sufficient for 3D visualization and cannot read VTK. I used Paraview and Mayavi2 for a while, but I was not satisfied with them. Paraview is not Pythonic and cannot handle a large amount of processing because it requires mouse interaction. Mayavi2 was Pythonic, but the commands were difficult to understand (meaning they didn't work for me).
Now I'm a big fan of PyVista.
My only complaint about PyVista(VTK) is that the meshing capabilities of Delaunay2D and Delaunay3D are not good enough. Meshing is a very important factor in the finite element method. I know that tetgen is available in PyVista, but its license is difficult to use. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for sharing, @tkoyama010! I had no idea you work in the realm of earthquakes and seismology. I'd love to see some more of your visualizations specifically in this domain.
We hear you! This has been a frequent complaint and is really tough to address directly in PyVista itself. It'd be great if we could find another OS Python library that focuses on this and has a good algorithm for Delaunay triangulation and build interoperability between PyVista and that library. Have you found other software that has satisfactory Delaunay algorithms? Doesn't scipy have something? |
Beta Was this translation helpful? Give feedback.
-
Good point. Using third party package power is a way to solve problem. I would like to investigate including scipy. |
Beta Was this translation helpful? Give feedback.
-
|
FEniCSx tutorial uses gmsh, which is probably the easiest solution for the current situation. |
Beta Was this translation helpful? Give feedback.
-
|
For some cases requiring constrained 2D Delaunay triangulation, I have been using |
Beta Was this translation helpful? Give feedback.
-
|
I'm using PyVista for my PhD research in Space Plasma Physics. I'm coming from about a decade experience working- I'm not knew to using and evaluating python packages. I was pulled in by the ability to calculate streamlines from vector fields on adaptive grids. I knew VTK could do this, but the "excessively object-oriented" (my words!) style of the VTK python library turned me off. Calculating streamlines is the kind of thing where I could write my own integrator... but that would just take time away from getting my science done. This library has been a huge help not only because the API is clean, but because it's heavily vetted code that I am 100% comfortable relying on for my research. I don't use much of the 3D visualization, because I have visualization tools already developed in matplotlib. |
Beta Was this translation helpful? Give feedback.
-
🤯 I don't even know what that is! Sounds super cool though!!!
This is SO amazing to hear! Your point about "getting [your] science done" is 100% my motivation for building PyVista - to make 3D visualization and analysis approachable to domain-scientists so they can focus on the research questions at hand. I am thrilled to hear this is panning out for you (even though you're using other parts of PyVista and not necessarily the 3D viz parts)
Are most of your visualization 2D? I'd love to see some links (papers, blogs, etc.) or screenshots of your work if you can share! |
Beta Was this translation helpful? Give feedback.
-
|
Yes, I would like to see some space plasma too. |
Beta Was this translation helpful? Give feedback.
-
|
PyVista is used extensively within FEMORPH for data management and manipulation in conjunction with several Python C extensions. It's used for data visualization as well in a GUI developed using the
PyVista is used within the PyAnsys project within several open (and closed but will likely be open) source projects for data visualization. The public gallery examples:
Expect several additional releases in the short term, most requiring PyVista.
|
Beta Was this translation helpful? Give feedback.
-
|
It is so awesome to see all the hard work you've put into creating PyVista and building an ecosystem of seriously impressive FEM software between FEMORPH and PyAnsys, @akaszynski! |
Beta Was this translation helpful? Give feedback.
-
|
This is also a wonderful testament of PyVista's code quality, documentation, and community that you could build a thriving set of tools using PyVista's visualization capabilities in a commercial environment |
Beta Was this translation helpful? Give feedback.
-
|
@banesullivan Thanks for the initiative, this is Awesome! what is your domain/background? I have a background in geology/hydrogeology. My workflows are 95 % open source. And PyVista has been recently included as a visualization/3D information extraction tool. So, I used PyVista to read 3D meshes done in a third-party program. These meshes were geological zones that I would later use for assigning model properties. I have also used PyVista to build a Finite-Grid from topography. PyVista was a life saviour as I could learn it fast enough to do some plotting and exporting the information for the numerical grid fast enough to convince a client to stay. What is your familiarity with Python and the Python Data Science Stack? Familiar with numpy/scipy/pandas etc and also some tensorflow Also, use some field-specific libraries such as Flopy/pcraster what is your experience level with 3D visualization? Low experience. Some Paraview. have you used other 3D visualization software in Python or otherwise (open source or otherwise)? Yes, Paraview. What was your experience between these software? I have only done basic visualization in Paraview. To be honest, PyVista helped a lot to plot just what I wanted. what are your primary visualization tools? Matplotlib. R's ggplot2. QGIS is also handy in my field. where would you like to see PyVista go? So, for this project, I needed to interpolate the 3D meshes into a finite grid. In the end. I was not successful in using the interpolate functions or other tools to directly do it in PyVista. After running out of time. I used the voxelize tool to generate sample points and do a simple interpolation in numpy. I would love to use PyVista to directly interpolate/intersect 3D elements into a finite grid (Or maybe irregular). This would simplify enormously my workflows and also be able to convince other colleagues to drop expensive software in favour of open-source solutions. I would also love to contribute to that. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks so much for sharing, @vcantarella!! It is so awesome to see a fellow hydrologist here (though I'm more of geophysicist than a hydrologist 😉 )! This sounds like some really awesome work you're doing and I'm so happy to hear that you could learn PyVista easily to start making visualizations that directly impact your work and clients. I'm bummed to hear that PyVista's interpolation features didn't cut it for you. PyVista (via VTK) has a few re-sampling and interpolation methods that I've found useful with some of my geo/hydo workflows in the past, though they don't compare to proper geostatistical methods. However, they should hold their own against some of numpy's interpolations. Perhaps we can do a better job of demonstrating these interpolation features, because they will handle irregular meshes quite well - all while outlining the limitations. |
Beta Was this translation helpful? Give feedback.
-
What is your domain/background?I'm a condensed-matter physicist. I'm probably a bit of an odd one out here, because I rarely get to use PyVista and I use very few of its functionalities. Working on PyVista is pretty much a hobby for me, which might also partly explain why I'm hesitant to make larger design decisions, and why I'm more likely to be seen fixing grammar issues in PRs rather than implementing new features. What is your familiarity with Python and the Python Data Science Stack?I switched from MATLAB to Python in 2015, and I haven't looked back since. I use Python for all my research post-processing scripts, most of which are one-off data crunching and plotting scripts for a given problem. I occasionally use SciPy although I don't usually need it, but I heavily rely on NumPy and I'm pretty good at it. And all my 2d plots are Matplotlib. I've contributed some PRs to SciPy stack libraries, but most of them were docstring fixes. I did find and fix a bug in NumPy C code though which I'm pretty proud of 😄 What is your experience level with 3D visualization?Fairly limited. For professional work I mostly have to plot crystal structures and spin configurations, so I basically only work with point clouds with sphere or arrow glyphs. (Incidentally, my first PyVista contribution was related to customisation of arrow glyphs.) There are also one-off use cases for 3D visualisation, like when I couldn't resolve an apparent discrepancy between a cube and a hexagonal slice. Turns out that if you stand a cube on one of its vertices you get a hexagonal horizontal cross-section: Have you used other 3D visualization software in Python or otherwise (open source or otherwise)? What was your experience between these software?I first used ParaView (following recommendations from coworkers) to visualise spin configurations exported in .vtk format by one of the programs in the research group. I loathe manual labour, so I always look for scripting opportunities. So I ended up finding mayavi. It was great for being able to put together complete visualisation scripts which could be run with no manual intervention. But at the same time I found its use cumbersome, its documentation lacking (at least unintuitive to me personally). I would often end up putting together scenes manually, and using the "record" feature to get an autogenerated script I could work up from. It was inconvenient even when I was using it, but this became even more obvious when I finally found PyVista. No offense to Enthought and the mayavi contributors; it was great help in moving away from a click-click based workflow to something runnable. And most of my difficulties probably stemmed from unfamiliarity with VTK itself. What are your primary visualization tools?I switched to PyVista for my 3D visualisation tasks when someone on Stack Overflow recommended it under an answer of mine (it was either a mayavi answer, or a "don't use Matplotlib for this 3D scene" answer). The library looked interesting, but (apparently) I thought the API needed "some getting used to". Well, I got used to it real quick. And the documentation was overwhelmingly excellent: extensive API reference, lots and lots of examples. All I'd ever want. I loved the pythonic, unbloated API, and I've been loving it since. And when I was missing that aforementioned arrow glyph feature, I could
It was really empowering both as an end user of the library and as an open source enthusiast. I've also learned a lot about 3D visualisation through contributing to PyVista. Interestingly, I'm still largely oblivious to the fine details of VTK and its pipelines, which I also consider to be a sign of PyVista's success at being what it is. Allowing to harness VTK's sheer power and wealth of features by exposing an API that's accessible to end users who come from all sorts of technical backgrounds. Where would you like to see PyVista go?As I mentioned earlier, I have both very few needs personally, and insufficient perspective on what the diverse userbase of PyVista might really need, so I would rather not try to tell. As a contributor I only want PyVista's core strength to prevail: offering a straightforward, end user friendly and pythonic API with thorough and accurate documentation. |
Beta Was this translation helpful? Give feedback.
-
Background in mathematics (phd in probabiity). Now I'm working as a statistician and programmer (mainly with R).
I'm rather still new to Python. We use it at my work now, and my boss asked me to learn (to do Flask applications for example).
I play for a long time with the R package rgl, a wrapper of OpenGL. I also played with the Haskell version of OpenGL, PovRAY, Asymptote and JavaScript (three.js for interactive 3D and paper.js for 2D). Now I believe I mostly use PyVista, just for fun, no applications. With rgl I don't know how to get shiny objects such as those you get in PyVista with a high specular value. Also, PyVista is faster.
I find it is cool as it is currently. There are some missing features though, e.g. the convex hull. And it would be nice to have an export to webgl (that's feasible with rgl). But maybe this is already implemented? |
Beta Was this translation helpful? Give feedback.
-
|
For the Delaunay tessellation and the convex hull, there is the Qhull library, available in C or C++. However it fails sometimes for some cases which don't fail with the C++ library CGAL (harder to use). |
Beta Was this translation helpful? Give feedback.
-
|
Slightly tangential here, but import numpy as np
import pyvista as pv
# generate random points in a ball
points = 2*np.random.rand(1000, 3) - 1
points = points[np.linalg.norm(points, axis=-1) < 0.5]
poly = pv.wrap(points) # point cloud
triang = poly.delaunay_3d().extract_geometry() # convex hull with Delaunay triangulation as faces
plotter = pv.Plotter()
plotter.add_mesh(poly, color='navy', render_points_as_spheres=True)
plotter.add_mesh(triang, color='red', render_points_as_spheres=True, show_edges=True, opacity=0.2)
plotter.add_mesh(triang.points, color='yellow', render_points_as_spheres=True)
plotter.show()Or is this not what you want? Is it bad for pathological cases? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, Delaunay computes the convex hull, but it produces some tetrahedra, there are some triangles inside the convex hull with Delaunay. Moreover it does not allow to get non-triangular faces. |
Beta Was this translation helpful? Give feedback.
-
|
What is your domain/background? My background is computer science, image processing and computer graphics. I like 3D stuff 😁. What is your familiarity with Python and the Python Data Science Stack? Although I was originally familiar with C and C++, I had to transition to Python for my latest job and I was surprised how fast it is to get going. Now I can say I am procifient enough to be productive while developing my projects. And that's what I like with Python: right to the point 👍 ! What is your experience level with 3D visualization? Roughly 7 years of scientific visualization of data from various fields like CFD, astrophysics, molecular biology and neurophysiology. Have you used other 3D visualization software in Python or otherwise (open source or otherwise)? For work, I played with a few python packages: I also used ParaView pretty extensively for topological data analysis. Finally, I have fun with Blender as a hobby :) What are your primary visualization tools? Now, I'm all in with PyVista! Where would you like to see PyVista go? More testing, I want 100% code coverage 🤩 |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the package list, too! |
Beta Was this translation helpful? Give feedback.
-
+100 for increasing code test coverage. And I also want to increasing the coverage of type hints. |
Beta Was this translation helpful? Give feedback.
-
what is your domain/background?
what is your familiarity with Python and the Python Data Science Stack?
have you used other 3D visualization software in Python or otherwise (open source or otherwise)?
what was your experience between these software?
how do you use PyVista?
where would you like to see PyVista go?
Thanks to the many PyVista developers (@akaszynski @banesullivan @adeak @MatthewFlamm @tkoyama010) who respond promptly to my inquiries, review my pull requests, and make this an awesome tool and community! |
Beta Was this translation helpful? Give feedback.
-
|
The 1.2.3. Trifecta above I completely agree with. Are there examples online of your wing pressure simulations? |
Beta Was this translation helpful? Give feedback.
This comment has been hidden.
This comment has been hidden.
-
|
Perils of handheld typing! |
Beta Was this translation helpful? Give feedback.
-
|
what is your domain/background? I'm an Earthquake Engineer based in Wellington, New Zealand. My daily job requires seismic analysis of buildings, bridges and wharves. I'm also study ground response used seismic motions. what is your familiarity with Python and the Python Data Science Stack? what is your experience level with 3D visualization? have you used other 3D visualization software in Python or otherwise (open source or otherwise)? what was your experience between these software? what are your primary visualization tools? where would you like to see PyVista go? Thanks to the great developer team. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for sharing this information with us. I feel very close to you because I have the same background as you (analyzing buildings for earthquakes is my job too). I didn't know that OpenSees could be used as a Python interface. I'm very excited about this. I'm curious if this supports VTK output. |
Beta Was this translation helpful? Give feedback.
-
|
It is great that it supports VTK output. I have not done any visualization of building structures with PyVista myself, but there are already examples of trusses and I would like to try it in the future. |
Beta Was this translation helpful? Give feedback.
-
|
You may also like this discussion. |
Beta Was this translation helpful? Give feedback.
-
|
Great. It is really encouraging for me. When I look at the examples, I feel out of place. |
Beta Was this translation helpful? Give feedback.
-
|
what is your domain/background? |
Beta Was this translation helpful? Give feedback.
-
|
Thanks, @bujack! Regarding the Jupyter widgets: these are not out of scope but they are on the fringes of PyVista's scope. Personally, I want to see these widgets get better and better and hopefully, I can work to that end in the near future |
Beta Was this translation helpful? Give feedback.
-
|
Saw PyVista out in the wild at Using Ansys Scripting with pyMAPDL...
|
Beta Was this translation helpful? Give feedback.
-
|
I generated a 3D view of road maintenance GPR What you are seeing is a radar investigation of the road underground, first ~30 centimeters driven in the city of Kaiserslautern.
I am a Geophysicist, this work is part of our research projects RADSPOT and ZEBRA .
Only a few basic blender tutrials and the basic tools used in Geophysics
|
Beta Was this translation helpful? Give feedback.
-
|
Ah! This is fantastic! I love seeing more folks in the subsurface science using PyVista. Thank you for sharing!! |
Beta Was this translation helpful? Give feedback.
-
|
Gladly. |
Beta Was this translation helpful? Give feedback.
-
|
I don't think this would be too tough... feel free to reach out to me on the PyVista slack workspace: https://slack.pyvista.org/ |
Beta Was this translation helpful? Give feedback.
-
|
what is your domain/background? what is your familiarity with Python and the Python Data Science Stack? what is your experience level with 3D visualization? have you used other 3D visualization software in Python or otherwise (open source or otherwise)? what was your experience between these software? what are your primary visualization tools? where would you like to see PyVista go? I'd like to be able to use it within a broader software suite, but I need to enhance my PyQt5 skills further before venturing down that avenue. If PyVista has a published paper I will ensure I reference it in the next publication (discussions currently underway) Thank you! If you want to follow up on anything feel free to contact me. Like @GGDRriedel I also work with Geophysicists so when/if we have suitable GPR I'll look at PyVista |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for sharing :) The paper is at https://joss.theoj.org/papers/10.21105/joss.01450. |
Beta Was this translation helpful? Give feedback.
-
|
Sounds awesome! Thank you for posting!! |
Beta Was this translation helpful? Give feedback.
-
|
what is your domain/background? Example of using pyvista: I would like to thank all contributors to pyvista for this amazing software! Keep on with your great work!
|
Beta Was this translation helpful? Give feedback.
-
|
This looks awesome! Thank you :) |
Beta Was this translation helpful? Give feedback.
-
|
Such an excellent example! Really great work here! Thanks for sharing! |
Beta Was this translation helpful? Give feedback.
-
|
Very cool! |
Beta Was this translation helpful? Give feedback.
-
|
Hi there 🖐️, my name is Andreas, I'm the author of 🔍 FElupe - a GPL-licensed Python package on finite element analysis for continuum mechanics of solid bodies. I've realized all plotting related tasks in FElupe with PyVista.
What is your domain/background?PhD candidate in mechanical engineering with a focus on computational methods. What is your familiarity with Python and the Python Data Science Stack?I'm using Python for all kinds of data analysis since 2015. I also maintain my own packages at PyPI, see https://github.com/adtzlr. What is your experience level with 3D visualization?Absolute beginner. Have you used other 3D visualization software in Python or otherwise (open source or otherwise)?ParaView with XDMF-files using meshio :) What was your experience between these software?Great, but I found it frustrating to always open a separate GUI for the smallest / easiest stuff. ExampleHere's a little code snippet for FElupe + PyVista: import felupe as fem
mesh = fem.Cube(n=6)
region = fem.RegionHexahedron(mesh)
field = fem.FieldContainer([fem.Field(region, dim=3)])
boundaries, loadcase = fem.dof.uniaxial(field, clamped=True)
umat = material=fem.NeoHooke(mu=1)
solid = fem.SolidBodyNearlyIncompressible(umat, field, bulk=5000)
move = fem.math.linsteps([0, 1], num=5)
step = fem.Step(items=[solid], ramp={boundaries["move"]: move}, boundaries=boundaries)
job = fem.Job(steps=[step]).evaluate()
# create an instance of a PyVista plotter
plotter = field.plot("Principal Values of Logarithmic Strain")
plotter.show()
Beside the ax = mesh.imshow()
ax = field.imshow("Principal Values of Logarithmic Strain")
ax = solid.imshow("Principal Values of Cauchy Stress")I'm using only basic features of PyVista in FElupe and return the plotter objects for further modifications. Thanks for developing PyVista! 🚀 |
Beta Was this translation helpful? Give feedback.
-
|
It is wonderful. I too specialize in finite element structural analysis. Your library is excellent. I am currently using another solver called GetFEM, but I would like to try your solver as well. I am still in the experimental stage, but for pre-processing I am developing a PyVista accessors for Gmsh to generate 3D finite element mesh called pvgmsh. Please try it as a reference for your work. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you @banesullivan and @tkoyama010 for your kind words 😊. PyVista really provides one of the key technologies for FElupe to easily access the results! I'll definitely have a look on |
Beta Was this translation helpful? Give feedback.
-
|
Some weird rubber examples? |
Beta Was this translation helpful? Give feedback.
-
|
What do you mean with weird @RichardScottOZ 😄? |
Beta Was this translation helpful? Give feedback.
-
I have read your library's documentation and I believe you are correct. I have friends who are analyzing rubber's hyperelasticity, and I will recommend FElupe to them. The design of your document is also excellent, which stirs up inspiration. |
Beta Was this translation helpful? Give feedback.
-
Hi everyone!! 👋🏻
It has been so exciting lately to see how many people are using PyVista across so many different domains, each with varying levels of experience in Python, 3D visualization, data science, and any combination of these.
🌠 Hoping to capture PyVista's impact
I would so appreciate it if you all could take a moment to share with the community a bit about your experience finding and using PyVista.
I'm hoping we can use this discussion topic to gain insight on not only people's overall experience (for better or worse) using PyVista, but also understand where people are coming from, how PyVista impacts their work, and where the community would like PyVista to go.
Some questions I have in mind that would be awesome if you could answer or expand on are (please suggest other questions/topics as well!):
I'd love to capture our users' (your) stories using PyVista and understand how this toolset is impacting people's work.
Please free free to share as much or little as you feel comfortable!
🙏🏻 A thank you to our community
Thank you! PyVista's amazing community of users enabled this software to grow and we owe PyVista's success to you all. I would love to see the community around PyVista continue to steer this software and further build an impactful data visualization and analysis library - I hope that any insights gained here will help to that end!
Beta Was this translation helpful? Give feedback.
All reactions