Preparing for Data visualisation in Blender

Coding Python Support
#1

It looks like data visualisation is (or starts to be) a hot topic. And since Blender is the digital content creation suite (visualisation), why not diving into data visualisation and (big amount of) external data.

To work with somewhat bigger (mined, scraped, etc) amount of data, I guess we could use SQL, SQL lite, Pandas, or other databases. But so far I haven’t seen implementations of databases in addons (or Blender).

How would you start, considering the ease of installation, performance, and Blender’s limitations on visualising (bigger amount of) external data.
( Is it possible to make importing pandas module hassle free for example? )

#2

This is one of my interests too. There was a post not to long ago about a plugin that did some of this, I did some updating on it, but it was initially presented as a complete, working program, but after investigating, I discovered that it was missing some of the functionality, so because I am presently taking a Udemy course on Deep Learning, I did not want to take the time to continue the work. Blender would be a great way of presenting data that has more variables, for example, you could plot on the x and y-axis, plot with colors and size of points, which, in my mind would mean that you could get a much better picture of the data, and, present it as a heat map, plus size of points, which means being able to visualize more than just three dimensions! That is my take on it anyway. Original post for that article I mentioned is here: https://blenderartists.org/t/a-script-to-import-a-csv-file-and-create-meshes-for-blender-2-5x-or-later/501410?u=shawn_irwin BTW the course I am taking covers pandas pretty well, along with other ways to display data, like seaborn. You can look these up on the web too, they have main pages with examples. The Udemy course I am taking is by Jose Portillo, It is Python for Data Science and Machine Learning.

#3

Blender comes with a base Python distribution and you can add external Python modules as much as you want. You can also use your own Python dist as long as you stick to the Python version Blender requires (3.7 as of now)

1 Like
#4

That’s a coincidence (or not): I guess I follow the same course: 2021 Python for Machine Learning & Data Science Masterclass. And same idea behind it: Blender has a bit more than a x and x axis.

1 Like
#5

Ah, as long as you stick with 3.7 for now. I just was worried about how users of my addon (long term plan) could install the libraries, but I bet I will know by that time.
I’ve seen some interesting modules which we use already frequently: scipy, numpy, but there are quite some more. I guess some are not suitable for Blender like matplotlib. Well, on my way to find those things out.

#6

Numpy is already in Blender. See Sverchok add-on for how they use external libraries (like Scipy and Scikit). You can implement module installers. You can also implement Cython stuff, and see the Animation Nodes add-on for how it is utilized there.

Blender’s Python is just regular Python, and there is not anything interesting or weird about it.

3 Likes
#7

I see, I thought I had Numpy because of another addon, but it’s already build in.
Yes, I check those addons how they do it. Scikit, definitely going to take a look at that.

#8

I see loads of packages on https://pypi.org/. Is that the same as modules?

#9

Yes, you can install py modules using pip.

For instance on Win I can just run the command below, and it will install the mdoule in Blender’s Python lib folder. Just make sure you also install the pip module prior.

C:\BLENDER\2.91\python\bin\python.exe -m pip install scipy

1 Like
#10

Awesome, there is a whole world of codes out there.