@curious not really but ists allready quite good. A bit problematic to do low detail simplified stuff with it though

@Scott Seems like people are programmed by marketting to wont more features. I mean i have students building dimmers to their lights. Why well because a dimmer sounds like a great feature. THe only users who don't want a dimmer are the ones who had a dimmer in their home and realized that its a feature you very very rarely use. Besides how are you going to demonstrate a dimmer in a exhibition?

@Scott, you need a lawyer.

@joojaa yeah, I've been spending a lot of time with Stable Diffusion. Crazy good results sometimes if you're willing to look for the needle in the chaotic haystack. But images have sort of a fractal/cloud like composition. Works against simplicity.

And really hard to force a specific result. I constantly end up being lured in different direction than I initially wanted.

I've been stubbornly avoided adding artist names to prompts and of course then get chaotic results which need a lot of trial and error. A bit disheartening really that if I just write s simple prompt and add "art by x" I immediately get good results.

Feels like leeching.

@Wolff yeah you should teach it your own style. Or perhaps use 2 different artists?

@joojaa Yes, I believe that if artists should make use of AI they should definitely do more than just write prompts. Teach it their own style or do something clever or interesting with tit. Challenge the medium.

@joojaa Two artists can give nice results. Are you familiar with the [from:to:when] syntax?

(Uhm .. "do something clever or interesting with it" - not tit. Or they can if they want, but that wasn't what I meant.)

@Wolff you can draw and let the AI suplement

@joojaa yes and draw on the result and throw it back in the ai

@Wolff im learning how to write music

@joojaa cool! Sheet music or in a sequencer? Do you play an instrument already?

theatlantic.com/technology/archive/2022/10/…