WebM Screenscast Stitcher

Manage a screencast project and merge it's parts into a WebM file.

A project can contain:

WebM (VP9) video files
SVG slides (rendered for a given duration)
SVG overlays (rendered for a given duration on top of a single WebM)

The project is defined through a yaml file.

The WebM and SVG files are managed through external tools:

ffmpeg
imagemagick (convert)

Warning: this is alpha quality software!
Use it at your own peril and only on copies of your files.

Features

Stitching WebM files together.
Adding SVG slides.
Applying SVG overlays.
Adding text overlays.
The project file can be managed in a git repository.
Optional caching of the temporary files.

Limits

No sound support (for now)
No validation is done, that what you feed the program is correct (dimensions, file types, ...; for now)
All resources (WebM, SVG, PNG) must have the same dimensions.
All the text overlays have the same formatting and position

Usage

Process the project

screencast-stitcher.py [-o output.webm] screencast.yaml

Create an ouptput.webm file according to the structure in screencast.yaml

Options:

-o: the output file name (default is output.webm)

Creating a project file

Create the scheleton for the project file:

screencast-stitcher.py -generate [-o screencast.yaml] first.webm

The file will contain:

the width and height of the WebM file
the entries for the first WebM and for the cover and closing slides.

size:
  width: 1400
  height: 1400
tracks:
  - file: cover.svg
    duration: 4
  - file: first.webm
  - file: closing.svg
    duration: 4

Creating a slide template

Create an empty SVG file, of the size of the webm file:

screencast-stitcher.py -svg [-o template.svg] first.webm

Adding overlays

as SVG with a transparent background
as a one line text string

Creating an overlay template

Create an SVG file, of the size of the WebM file and with the screenshot at a specific time

screencast-stitcher.py -svg-frame 0:02 [-o frame.svg] first.webm

Caching the tracks

Merging the WebM files is very fast, but generating the WebM from the SVGs is not.

Enabling the cache will store in the cache directory the generated yaml files.

screencast-stitcher.py -cache screencast.yaml

Currently, you can only have one project per cache directory.

TODO: add an option in the project file for defining the cache directory

Converting to VP9

Convert a VP8 encoded WebM to use VP9:

screencast-stitcher.py -to-vp9 [-o output.webm] track.webm

Get informations about a WebM

Shows the encoding, width and height of a specific WebM file

screencast-stitcher.py -size track.webm

TODO: probably, change it to info and show the size + the codec of each of its tracks

Dependencies

External tools:

ffmpeg (for most of the hard work)
imagemagick (for convert)
mkvtoolnix (for mkvmerge)

Python libraries:

python3-yaml

Ideas and alternatives

https://www.npmjs.com/package/video-stitch

Todo

if the font name is not a path, first look in the local folder, than in ~/.fonts, /usr/share/fonts/... (or other def.
-svg-frame does not seem to work on WebMs with drawtext overlays
create demo / test files
fade out the text and slides?
- https://video.stackexchange.com/questions/23481/ffmpeg-drawtext-fade-out
check the dimensions of all svg and WebM files!
check that all WebM have vp9
check the total times when extracting frames or adding overlays
can mkvmerge be replaced by ffmpeg?
only overwrite a default filename if -force is set
check if the tools like -svg should not read from the yaml
add /document sound?
- espeak ?
- festival ?
  - Nick1 voice is not bad (http://www.cstr.ed.ac.uk/projects/festival/morevoices.html)... but it does not seem that one can download it.

check that the projects accepts file names with spaces
if not look into this:

  import shlex

  raw = 'mkvmerge -o output.webm -w /tmp/tmpgqjqu0jx.webm +01.webm +/tmp/tmp4uipiirx.webm'

  split = shlex.split(raw)
  print(split)

  and_back = ' '.join(shlex.quote(s) for s in split)
  print(and_back)

  print(and_back == raw)

aoloe / screencast-stitcher