Extract keyframes and add them to the generated text

[shun-liang]

https://www.reddit.com/r/DataHoarder/comments/1g4342q/comment/ls4y0u6/

Any future of possibly grabbing frames at various time intervals to incorporate into the documentation with an argument in the command? "--framegrab 60" "--framegrab chapter" would be nice for the doc and help incorporate breaks in the text. Even if it spit out as separate files that could then be attached in obsidian or bookstack would be cool.

Taking frames will be awesome if it's done right. I have been thinking about the snapping "key frames" (yet to define what a key frame is), rather than just taking frames at a frequency or just the beginnings of the chapter.

There is a project https://github.com/hediet/slideo that matches slides (PDF pages) to video timestamps which I find very cool. That requires the user to have the PDF slides ready which isn't always the case though.

Need to investigate what is a key frame, the feasibility to extract keyframe, and how complex it would be.