I have added some code to attempt to parse chapter titles inside 'mdat' atoms. This is certainly reverse engineering and not based on any documentation (because I wasn't able to find anything relevant), and primarily based on a single sample file. If you find another sample file where this doesn't work as expected, please send it to me for analysis. Of course, if you find documentation on how chapters are actually stored inside 'mdat' atoms, that would be appreciated too.
Patch here:https://github.com/JamesHeinrich/getID3 ... 38ddc7e374
Chapters from 'mdat' atoms are now returned directly inside ['quicktime']['comments']['chapters'] (which is generally also copied to ['tags']['quicktime']['chapters'] and ['comments']['chapters']) as an array of strings.