3D Manim Animation of CNN (Convolution)

it's cool.

but I really don't understand why RGB is the goto encoding, it's easy to get unwanted artifacts and longer training times. YCbCr is better imo, for some cases you can only use the Y channel for edge detection which is much faster and more compact. any ML engineers can shed some light on the topic? perhaps I'm missing something.

The spinning objects made me nauseous.

It could probably be presented still and at 10-20x faster and still communicate the essential idea.

You could/should post this as a usual HN submission. Plus the link would work. Posted like this it goes into AskHN, which it isn't. I notice you have 0 comments and have only submitted your own stuff - bear in mind :

"Please don't use HN primarily for promotion. It's ok to post your own stuff occasionally, but the primary use of the site should be for curiosity." https://news.ycombinator.com/newsguidelines.html