Scripted visualizations in blender

The goal of this project is to script 3D scientific visualizations using blender.

To achieve this, we aim to bring together blender's powerful visualization toolkit with Anaconda's scientific computing and package management capabilities. For example, the code in ./apps/concepts/fourier_signal_composition.py was used to generate the following visualization.

Getting Started

See the detailed setup instructions at the end of the file.

Contributing

This project is still a work in progress. Contributions and feedback are welcome!

Authors

Praneeth Namburi - praneethnamburi

License

Files in the directories apps and illustrations are copyright praneethnamburi.

The general purpose blender scripting code (bpn, pntools) is under the MIT license.

Acknowledgments

All the wonderful people that make open source software
Inspiration - 3blue1brown's videos and pedagogical clarity

Setup instructions

These are detailed instructions that worked for me on a windows 10 laptop.

blender+Anaconda+VSCode

Download blender (get the zip file, NOT a binary installer), or just follow this link: https://builder.blender.org/download/
Unzip to C:\blender\2.93.0 (which has a folder called 2.93)
Open the python console within blender, and check the python version
- e.g. 3.9.2
Delete the python folder and all its contents (C:\blender\2.93.0\2.93\python)
Install Anaconda (NOT miniconda), and open anaconda prompt with admin privileges
- Make sure you have "C:\Users\Praneeth\anaconda3\condabin" in the system path
- On windows, check the Path variable in the 'System Variables' box when editing environment variables
Clone this repository, and the dependency to your workspace (make sure git is installed and added to your system path)
- Navigate to your workspace in the command prompt (e.g. C:\dev)
- git clone https://github.com/praneethnamburi/blender-ScriptViz.git
- git clone https://github.com/praneethnamburi/pn-utilities.git
- cd blender-ScriptViz
Create an anaconda environment using the following commands:
- Recommended method:
  - conda create -n blender293 python=3.9.2 numpy scipy pandas jupyter ipython matplotlib blinker scikit-learn
  - conda activate blender293
  - conda install -c conda-forge pybullet multiprocess pysimplegui
  - pip install decord imageio imageio-ffmpeg ffmpeg-python pytube ahrs urdfpy pint soundfile celluloid
- Alternate method: Create an anaconda environment using the _requirements.yml file (simpler, but doesn't always work)
  - conda env update -f requirements.yml
  - Make sure to wait until it finishes running. It might appear stuck when installing pip packages. You might see a temporary text file created by conda in your current directory, for example "condaenv.p5qt3m3s.requirements.txt". Conda has finished doing its job when this file is deleted.
  - conda activate blender293
Install VSCode and activate the environment from within VSCode's command line
Call blender from the command line. The idea is to pass an extra argument while launching blender to set the path the python we want to use.
- C:\blender\2.93.0\2.93\blender.exe --env-system-python "C:\Users\Praneeth\.conda\envs\blender293"
- C:\blender\2.93.0\2.93\blender.exe --env-system-python "C:\Users\Praneeth\anaconda3\envs\blender293"
- Remember to use double quotes if there is a space in the path!
- If this works, you're good to go! Rest of the steps make are meant to make your life easier in the long run.
- You should be able to install additional packages using conda and import them in the blender console.
Recommended: Add the path to this repository to your python path. For example, create a 'paths.pth' file, open it in notepad, and type the following lines into it:
- C:\dev\blender-ScritpViz
- C:\dev\pn-utilities
- Save this as C:\Users\Praneeth\anaconda3\envs\blender293\Lib\site-packages\paths.pth

Troubleshooting:

temp.yml, and check which packages failed to install. If it is pybullet, then it is probably because it needs Visual C++ 14. You can install Visual Studio Community edition (if you have the space, or perhaps the redistributable VC++ also works, I haven't tested it.)">

- A useful tip is to check if you're able to find the correct python, pip, conda and blender commands from your command prompt. Most of the issues I encountered had something to do with the correct paths.
- Use 'where blender' in the windows command prompt inside VSCode
- Result: C:\\blender\\2.93.0\\blender.exe
- where python
- C:\\Users\\Praneeth\\.conda\\envs\\blender293\\python.exe
- where conda
- C:\\ProgramData\\Anaconda3\\condabin\\conda.bat
- Check VSCode settings - 
  Add these settings in VSCode (to your workspace) - Modify this example
     -  "settings": {
           "terminal.integrated.env.windows": {
              "PATH": "C:\\blender\\2.83.0;C:\\Users\\Praneeth\\.conda\\envs\\blender2830;C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\Library\\mingw-w64\\bin;C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\Library\\usr\\bin;C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\Library\\bin;C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\Scripts;C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\bin;C:\\ProgramData\\Anaconda3\\condabin;*OTHER THINGS IN YOUR PATH*,
           },
           "python.pythonPath": "C:\\Users\\Praneeth\\.conda\\envs\\blender2830\\python.exe",
        },
- If conda env create -f _requirements.yml fails, activate the environment, and use conda env export > temp.yml, and check which packages failed to install. If it is pybullet, then it is probably because it needs Visual C++ 14. You can install Visual Studio Community edition (if you have the space, or perhaps the redistributable VC++ also works, I haven't tested it.)

Current workflow

Start VSCode
Activatec conda environment from the terminal
- conda activate blender293
Start blender
- C:\blender\2.93.0\2.93\blender.exe --env-system-python "C:\Users\Praneeth\.conda\envs\blender293"

Folder structure

_auth: Authentication

This folder contains authentication keys for interfacing with applications. Don't commit this when working with multiple people.

_dev: Developer notes

Notes for development and learning during the course of the project. _requirements_topLevel.txt is meant to help with python's package management. It is a good idea to add this to source control when developing with multiple people, but this will eventually disappear from distribution.

_temp: Temporary folder

Local cache for storing intermediate data generated by the software.

apps

Applications that use the main package bpn, and supporting package pntools. See License.

bpn

This folder contains the core scripts for using core module has wrappers around blender objects, divided into three types:

Thing - class that initializes all wrappers
Object, Mesh, Collection, GreasePencil - wrappers around bpy.data.(*)
MeshObject, GreasePencilObject - Initialize object + data
- Each of these classes have analogs in the new module (mesh, pencil)
- The user should only need to interact with core classes through functions in the new module utils.get is the dispatcher that automatically creates objects from the appropriate classes in the core module.

names are very important in bpn. We use names to insulate bpn from bpy. That means, bpn tries very hard not to keep a copy of things from bpy. Instead, it tries to get the appropriate information from bpy when needed. names determine the interaction between bpy and bpn.

*args are for the 'new' function to create a new blender data instance *kwargs are for initializing that instance inside bpy.data.(type).(instance)

Objects and lights need to pass one argument through *args. I did not set it to have the flexibility of initializing empty objects with Nonetype. Classes inherited from Object also send *args up to Thing class (e.g. MeshObject, and GreasePencilObject) Rest of them ONLY send kwargs for initialization.

Modules vef, trf and env currently do not depend on any other files within bpn. env requires blender and therefore, will stay within bpn, but the other two can become their own packages that bpn uses. Perhaps move them to pntools?