So, I was downloading some YouTube video when I thought about this… This will probably never turn into a feature but I need to write it down before it's lost forever 
It has to do with the ability to retain the metadata in a usable form between loads (comments, drawings, paths, etc.).
Currently, there is a saving option in Kinovea that let you combine the metadata and the video in a single file (or you can save the metadata in their own file somewhere else).
But if you want to upload it to YouTube or something, obviously you need to burn the drawings on the images, and you'll loose the comments.
This is because YouTube will change the format of your video and re-encode it, so even if you download it back, you only get raw images.
But how do you transmit metadata so they survive to format change and re-encoding, and stay editable when you open the video on the other side?
The idea would be to add a special image as the first frame of the video, containing a 2D bar code:
This kind of matrix code can contain several Kilobytes of data, and that might be enough to transport the drawings and path stuff, or maybe the comments.
When you redownload the video from YouTube, (using a firefox plug-in for example) and open it in the software, it would detect the bar code on the first frame, translate it back to the metadata, and display the comments.
If the website encoding doesn't completely kill this specially crafted image, it could work.
Or it could also be used to transport metadata in files that don't support it initially. (Currently the only option is to use the MKV container).
/end of thoughts dump