AI generated art

Imagine a realtime knob-and-slider interface for dialing in prompt-based musical parameters on an optimized model that’s been trained on billions of hours of sound and music… running locally at low latency… geez louise.

1 Like

A cultural misunderstanding in the prompt but I was impressed enough with the results that I didn’t care :slight_smile:

Cooking up some classy-ass branding for a design agency called Egg & Chips

3 Likes

A warrior bird in 1983

image
image
image

2 Likes

A warrior bird in 1968

image
image
image

2 Likes

Thanks.
That’s a particularly great episode of a generally great podcast. Holly & Mat are some of the few people who are optimistic but not utopian when talking about all this stuff, really appreciate their work and output.

I find this new technology equal parts exciting and terrifying. I think the best course of action is to try to enjoy it and do good things with it because it’s not going to go away.

A computer drawn by Leonardo Da Vinci

image
image
image

3 Likes



5 Likes


3 Likes


1 Like





2 Likes

Some abstract experiments from this morning







10 Likes




3 Likes

1 Like




6 Likes

Plaid &
Visual work by Emma Catnip

2 Likes

Although the discussion is around if AI art, there is a new breed coming out of AI Artists, embracing the technology as it progresses, jumping to the new tools as soon as they are out. There will be coder/programmer/admin who will build the machine and algorithm and the artist using them.

I assume there was also the same being said about Synth and electronic music instruments once we started seeing them. It was probably not music or even art back then…

All that to say that: Dude! You’re really pushing the envelop as far as AI Artist, your stuff is getting really awesome!

2 Likes

Thanks mate! I find myself creating more album covers than albums…

3 Likes

How do you generate such beautiful translucent spheres. I had no luck so far with stable diffusion.

This is using the Midjourney implementation of Stable Diffusion and I found the best luck with ‘Crystal Ball’ - that seemed to give it the context it needed. I find trying to use words like ‘ball’ or ‘globe’ as descriptors can sometimes introduce unwanted context with a lot of stuff like this - so having something very specific to target was convenient :slight_smile: :crystal_ball:

Also qualities like ‘backlit’ and ‘shining edges’, things like that can help get some extra good stuff with something like this.

Edit: Interestingly using straight Stable Diffusion (the one at huggingface), it does a great job with the sand but seems to get totally stumped by the crystal. Maybe that’s something Midjourney is handling much better.

Edit edit: Maybe with a bit of work you could get there, these are better - I’d love an upscale of the last one


more

1 Like

Although these images are stunning (really like your input in this thread natehorn, the image in the crystal balls should be flipped upside down shouldn’t they?

1 Like