Today, I discovered the Learning Equality project, which goal is to offer high-quality, free teaching resources to the people who don't have access to the Internet. The idea is that 1/3 of the kids don't receive a quality basic education while 60% of the humans don't have access to the internet, and that problems often come together.

One of their technical contribution is a compression tool for the videos that you can find on the Khan Academy. They show only hand-made writings and pictures, along with the voice of the presenter. For now, we use the codec that were invented for movies and theater shows. But unfortunately, they are looking for flat tints and color gradients, so they will blur the lines if you push them too hard. At the end, 5mn of a dumb video can last up to 250Mb. Shame. Shame.

Their tool vectorizes the image, meaning that it searches for (straight) lines in the image and only store the coordinates. That would be stupid for a movie given the small amount of lines in a typical movie, but it is perfectly suited to such videos: the same video is reduced to a bit more than 1Mb. Even better: all of the remaining size is used by the audio stream. They use mp3, which is not particularly suited to voice compression. Ogg codec reduce the size further to 500kb, and an aggressive phone codec can get down to 120kb (at the price of the voice quality). Much better than 250Mb for the same content!

Even more interesting, this encoding makes it easy to edit the video, eg to add a sequence in the middle in which new things are drawn, even if it require to move away things that are drawn later on. This is rather impressive. Click on the image for the full paper.

This tool will allow every teacher to edit and adapt the videos that he selects for his/her students. That's precious because the teachers are the ones who know the student and can best adapt the content to them. If I had 40 hours a day, I'd check whether these tools hold their promises by developing some videos for the PLM...