It's very tough to say anything, it might need to look at the frames in more detail (slightly shifted frame, that is missing in timelapse) to create better abstractions, it's very tough to say anything in detail. Only with experimentation we can know, nonetheless, it is an interesting idea