Running on Mac OSX

[ryangrush]

I would like to get this working on my mac but it looks to have trouble on a non-Ubuntu machine. Forgive my ignorance but is there a good recipe to get this working. Maybe install the dependencies via Homebrew or MacPorts?

[dhilowitz]

Here's what I did:

sudo port install pngcrush
sudo port install opencv +python26
sudo port install py26-wxpython py26-numpy py26-matplotlib
sudo port install py26-scipy py26-ipython
sudo port select --set ipython ipython26
sudo port install py26-pil

Hope this helps!

--David

[lilith]

This should be the rough equivalent with Homebrew (I'm on x64 Mountain Lion, so YMMV).

Note: gfortran is needed to build all of scipy, and freetype is needed to install matplotlib

brew install python
brew install gfortran pkg-config freetype
pip install numpy ipython scipy pil
pip install matplotlib
brew install --python wxmac --devel
brew install opencv
brew install pngcrush

The OpenCV Python module will not work until you edit your PYTHONPATH like so:
export PYTHONPATH="/usr/local/lib/python2.7/site-packages:$PYTHONPATH"

[lilith]

Here's a good article on getting the basics installed: http://www.lowindata.com/2013/installing-scientific-python-on-mac-os-x/

[tjramage]

Hi guys,

I'm running OS X 10.8.5 - trying to get this working also. The example screenshots seem to work fine - I can run the python script and it generates a packed PNG sprite and the relevant JS timeline. However, as soon as I point it to my own folder with some test PNGs, I keep getting this error:

ValueError: could not broadcast input array from shape (562,980,4) into shape (562,980,3)

Can anyone help or at least point me in the right direction? Thanks!

[vanyossi]

@tjramage I had the same problem. To me it triggers when one of the images has an alpha channel, or palette information. I had to strip all that before using the images. also images must be the same exact size wxh

With image imagemagick I did
mogrify input.png -strip -define png:format=png24 PNG24:input.png

[tjramage]

GENIUS! @vanyossi - that's fantastic, works like a charm now! Thanks :)

[umaar]

@tjramage did you solve the black screen issue?

[beefcheeks]

Hey everyone,

It looks like it's been awhile since this issue has been updated, and I had some trouble getting the anim_encoder to work on my MacOS Sierra machine. I did figure out the steps, but it took me awhile. Thought it might be helpful to have these steps available for other folks.

EDIT: fixed the steps below to include more detail. I've also updated the instructions to use my fork instead of merging commits yourself.

brew install python
# Add brew path to bash_profile if not already present
echo "export PATH=/usr/local/bin:$PATH" >> ~/.bash_profile
source ~/.bash_profile
# virtualenv steps below recommended but not required
pip install virtualenv virtualenvwrapper
echo "source /usr/local/bin/virtualenvwrapper.sh" >> ~/.bash_profile
source ~/.bash_profile
mkvirtualenv anim_encoder
# Install dependencies
pip install numpy
brew install gcc
pip install scipy
# Install PIL
pip install http://effbot.org/media/downloads/Imaging-1.1.7.tar.gz
brew tap homebrew/science && brew install —HEAD opencv3
brew install libpng libjpeg pygtk pngcrush
# Commands directly below required for virtualenv users only (parts of the paths below may be different for you)
cp /usr/local/Cellar/opencv3/HEAD-2038434_4/lib/python2.7/site-packages/cv2.so ~/.virtualenvs/anim_encoder/lib/python2.7/site-packages/.
workon anim_encoder
# Clone this repo and enter the directory
git clone https://github.com/baronomasia/anim_encoder
cd anim_encoder

EDIT: To switch between virtualenv python environments using the wrapper, use workon <env name>
EDIT: Removed merging instructions, use my fork instead.
With this fork, you can designate a specific area of the screen to capture by editing the screencapture line in capture.py below:

        os.system("screencapture -x -R0,0,800,450 capture/screenshot_" + str(int(round(time.time()*1000))) + ".png")

where R0,0 is the horizontal and vertical offset, and 800,450 is the area captured.