@baldur I have no idea what Adobe is doing — best we can do there is ask Adobe directly.

On training: after training, one can then use a trained model to make other determinations. The other determinations could be what’s covered in this text. But again, let’s ask.