I’ve always loved holograms, those virtual objects intermingled with the real world in so many of my favorite films, from Star Wars to Zenon: Girl of the 21st Century. But I wasn’t looking forward to learning Unity game development and Windows system APIs before I could make my first animation for the hololens platform.
So while I still have to dig into Windows-Guts and Direct3D to make interactive applications, I was excited to find a straight forward way to generate hololens-compatible animations with just a few lines of python.
Using the python embedded in the open source Blender project, we can read files, make database queries, do whatever data science we want to do, and then import meshes or generate cubes, apply colors and materials, define keyframes, and export an FBX file that can be displayed in Hololens or on web pages. Hololens lets you open multiple animations and place them around the room, so while I can’t define any gaze-and-click behaviors, I can still move and scale and rotate my animation in the mixed-real world.
The first step of course, is to head to blender.org and download the latest. I’m going to be really verbose and try to help out people on both Windows and Mac, because I had to figure it out on both myself. Follow all the default options, and on Mac move blender.app to /Applications like you normally do. From here, you can check out a million youtube videos on what Blender is for (That’s how I learned it!) but next we’re going to add ‘blender’ to our path so we can execute it from the command line anywhere in our system and not even have to learn how to use the GUI.
For MacOS, you can run this in your terminal, or add to your bash.rc:
For Windows, open up powershell (right click, ‘run as administrator’, which you’ll need to do for any commands that save to disk) and run:
[Environment]::SetEnvironmentVariable("Path", $env:Path + ";C:Program FilesBlender FoundationBlender", [EnvironmentVariableTarget]::Machine)
Now anytime you open your terminal/powershell, you can type ‘blender’ to launch blender. (On MacOS, you may need to restart your shell for it to take effect, or type ‘source bash profile’ to force bash to reload your saved preferences, including path) (Another option for windows is to search for ‘edit system variables’ from the start menu, click ‘environment variables’ and use the GUI to add C:Program FilesBlender FoundationBlender to your Path)
But going through all that trouble just to open Blender isn’t the point. The point is now we can save python files and pass them to Blender as a command line argument. Before we do tho, there’s one last step: we have to re-save the start-up scene to be a blank slate, otherwise our animations will all have a cube sitting in the middle of the scene (Blender’s default start up scene).
So open blender, and hit Select -> (De)select All, then Object → Delete, confirm. (or just ‘a’, ‘x’, ‘Enter’ if you want to use keyboard shortcuts). Next click File → Save Startup File. We’re finally ready to generate some animations with python!
Save the following code as helloworld.py, navigate to the folder you saved it in terminal, and type:
blender -b -P helloworld.py
And you should see Blender printing out a load of information about what’s happening, then it should create an fbx file and exit.
Check out the inline comments for a little bit of understanding of what the code is doing, and I’ll be writing more example code in future blogs.
And of course, the answers to all your questions are in the docs!
To make it easy to deploy these to Hololens, I just drop them in the onedrive belonging to the account I created with the hololens. There’s a python API for uploading direct to onedrive, too, so I’ll probably explore that in the future.
Optional: import interesting python modules like numpy and pandas and all the rest.
Blender comes packaged with its own python executable hooked into all of Blender’s guts, so to install any modules that aren’t built-in to Python3 we need to run get-pip.py with Blender’s python. First, download get-pip.py, I’ll put in in my desktop. Then, navigate to blender’s python executable.
command on windows:
cd "C:Program FilesBlender FoundationBlender2.78pythonbin"
command on MacOS (assuming Blender was moved to Applications):
(Note that 2.78 is the blender version, not the python version) Run ‘ls’ and take note of the python exectuable’s name. On windows it was just python.exe, on mac it was python3.5m. Once you’ve cd’d into python/bin:
Install Pip on MacOS:
For Windows I had to include to the full path of get-pip.py. Also this requires administrator privileges. So run powershell “As Administrator”
./python "C:UsersColten JacksonDesktopget-pip.py"
If that exits successfully, you can run ‘pip install pandas’ and whatever python modules you want to use from within Blender. On Mac, pip added itself to my path and I could use it right away, but on Windows I had to reference the pip.exe to run it from the python directory, so it ends up looking like this (again as Admin so pip can save files to disk, your permissions may vary)
Let me know what questions you have, tho of course I would appreciate it if you googled it first !
More examples and explanations here:
PS all my gifs are in black and white because I don’t know how to do color correction, so I just get rid of the color 🙂
Also, shout out to my primary sources that taught me all this: