OpenPose vs OpenCV

[DBonbon]

Dear python friend,
first thanks for the mocap project and even more for encouraging 'hunting' pythons... even though so far I'm on the hunted side.

I first go to the point, with a question, if you wish also read Item 2 - if and how I could contribute.

1. I have though one m meta-issue which I missing - that is why using OpenPose-JSON file and not OpenCV approach as you probably saw https://youtu.be/RY_eErKlilw

I can note that your script present a later python version is more updated and to my limited understanding better written (more precise).
The other script is rather easy to install and operate and mainly OpenCV incorporates OpenPose algorithms. I trust I miss here some elements, and you gave it a good consideration. That is can you help me to understand the advantage of your approach?

for my learning experience it helps to have conviction I explore the right path.

2. I read carefully your script. . As I have not python or coding BG, I literally study any dot:-j...).

I know some Blender animation and mainly focused on character animation. I tend to believe that mastering a script is a cine qoui non fro mocap. Even before any suits etc.
My end point is creating a nice animation, so I want to ensure I start from scratch rather then end point (suits/captures...)

Reading your script, it seems that with some not that major tweaks it can be really powerful too., that is, I'm interested to contribute, i.e. something I could try to code by following, copying your codes - like and add some more bones options for the mouth corners, guess some movement for the cheeks (knowing OpenPose has no cheek Keypoints), create option for bouncy Z curves ,etc'.

I'm not there yet... but I keep walking so I should arrive, one day;)

Cheers and happy coding,