Note Work in progress.

anim_encoder creates small JavaScript + HTML animations from a series on PNG images. This is a fork of the original work to do a few things:

• Update the script from Python 2 to Python 3
• Improve the usability of the script (more options)
• Modernize the JavaScript player to use new browser features.

Original details are at https://www.sublimetext.com/~jps/animated_gifs_the_hard_way.html

Clone the repository and install the required dependencies with pip.

pip install -r requirements.txt

Alternatively, you may install it via your system's package manager.

Warning This is deprecated and will not be updated in the future.

sudo apt-get install pngcrush python-opencv python-numpy python-scipy

Prepare the animation that you want to encode and convert them into image sequences. For example, to encode image sequences in capture/:

python3 anim_encoder.py capture

This will generate capture_packed.png and capture_anim.js.

To specify a different output name, you can use -o or --output.

python3 anim_encoder.py -o output capture

In this case it will generate output_packed.png and output_anim.js.

If you have pngcrush or pngquant installed, you may be interested in optimizing the output image further. For animations that has fewer colors such as screen recordings, pngquant provides significant compression.

# -C uses pngquant, -c uses pngcrush
python3 anim_encoder.py -C -c capture

For more help, run python3 anim_encoder.py --help to see all options.

Images will be saved to capture, you simply need to run capture.py and then go about your task. Note you can just delete frames you don't want as you initially set up, should save you some time. Then to run the program just go

python capture.py

If you need to change any settings it should be pretty simple just jump over to config.py and edit the configuration options.

If you are not capturing the video via capture.py, then you need to use a lossless video codec (ffv1, huffyuv, lagarith, etc.). Many codecs such as h.264 produces quantization noise which completely messes up the diffing algorithm and produces a big final result.