[xorcist]

A great photograph can take just a few minutes to get right, it's still great art.

There used to be a debate whether photography really was an art form, but that was long ago. Pretty much everyone now agrees that photography is a tool, and a great artist can produce great art with it. There is nothing about text-to-image learning models trained on huge corpuses that makes it fundamentally different from that.

Art is constantly evolving. That's what makes contemporary art relevant. The judges, who one may expect to have seen a lot of highs and lows, seem to find the tool used here less controversial than some of the commentators.

[FollowingTheDao]

> A great photograph can take just a few minutes to get right

As a professionally selling photographer whose work appeared in magazines I can say with great authority that you are wrong.

> There used to be a debate whether photography really was an art form

Yes, and we are just beginning that debate with AI.

In an artists medium there are constraints. For the painter, the paint, for the photographer, the subject. With AI there are no constraints, so it is not art.

[visarga]

In Stable Diffusion constraints are

- the imagination of the prompter, or ability to find good inspiration somewhere

- the concepts learned by the model, or the training set used in creating the model

But you're right, it might not be art. It's something else. It's one-time-use-art because most of the images only get one glance and are forever deleted. It's an imagination-bicycle for the mind. Art does not follow your thoughts like the models do. It's a cognitive enhancement.

[FollowingTheDao]

> the imagination of the prompter, or ability to find good inspiration from somewhere

Imagination is unconstrained. You can literally imagine anything.

> the skills learned by the model, or the training set used in creating the model

This is not a constraint of the artist, it is a constraint of the AI if is even a constraint.