It's actually fascinating watching #Aigen create an image frame by frame.
It's like reading tealeaves.
Say you asked for a cat, the first frame is just a random diffusion noise.
With each frame, the AI tries to shoehorn a cat into the noise.
By about frame 15 it starts to look like a cat.
The helpful diagram is describing the first stage, training.
Or as Picasso said;
"Good artists copy
"Great artists steal"