The cameras on the S20 appear to be approximately near the edge, however they might be a little bit farther away from the edge than on an iPhone. Maybe that’s all I’m seeing. If you have the “ultra” model it looks like there is an additional camera called the DepthVision camera which is located more toward the center of the phone. It might only be used to measure distance, not to produce an image. That camera would definitely produce a high angle if it’s used, but I’m not sure if it is.
It’s not that I’m against finger motion per se, it’s that I’ve become wary of any motion that only appears to activate intermittently. When we’ve filmed players, the helper motion seems to be where most of the mistakes occur. If the technique is working then I don’t worry about it. But if there are accuracy or other issues, and I also see helper motions, then I think ok, better off getting rid of them to reduce the number of variables and complexity.
Your motions look excellent in terms of being the right ballpark. There are simply a few lingering mechanical issues that you can potentially streamline or standardize (e.g. finger vs no fingers, anchor location, consistency of attack, etc.). The accuracy issues aren’t so much because motions are incorrect, as they are simply occasionally playing the wrong note or string. i.e. “Mistakes”, in the classic sense of the term.
I notice many of your clips are super fast, at speeds where it’s hard to make accuracy improvements because everything is a blur. We advise fast motion as a starting point, but only a starting point. This is just to make sure players don’t inadvertently choose an awkward and inefficient motion which can’t be sped up, or can’t be sped up without effort and tension. You’re beyond that point, so there’s no need to go full blast other than as an occasional test.
Of course if you slow down too much the motions are no longer the same and you’ll just get all the notes right with the wrong technique. So slowing down only a small amount to a zone where the hit rate is high but not perfect will give you a more manageable number of errors to try and fix. I would just make the mental note to use this slightly slower speed whenever possible and not slip into the all-out shred/fun zone. You don’t need hours per day to improve accuracy, just a greater percentage of time spent trying to get rid of errors in the “improvement zone”, where mental effort is greater.