I'd like it if more people got involved and posted the results of their experiments
the development of GANS and AI art is interesting - it's at the wild west stage at the moment - the technology is in its early stages, and opposing streams are emerging - there's the "everyone can be an artist - let's share and learn - free to all!" crowd and then there's "let's mint NFTs", "set up a limited run print shop", "this time net year we'll be millionaires" crew who want to enslave the AI to do their bidding and make them rich...and then there's Stan, who will be the worlds first successful multimedia AI artist...
imagine...
72 HD Screens controlled by 18 GPUs ( 1 GPU per block of 4 HD screens + master controller server to control layout )
Stan fed by tubes, slow drip of crystal meth, wired awake reading aloud from Gravity's Rainbow until the very last page, 18 GPUs processing his words and generating images ( and sound - these sort of setups have a decent sound system and dedicated audio server )
I look at posting pictures as examples that I ( and maybe others ) can learn from - what prompts are people using? It helps to understand what material the GANs have been trained on - I've learnt "hyperreal" as a prompt this week thanks to
@WashYourHands
also - not obvious at the moment, but this is on the same level of the introduction of sampling in music - at some point there will be a copyright claim from some dead artist's greedy estate over the use of "in the style of" prompts, there will probably be an AI assisted forgery on sale at an auction, and if the technology improves then there will be agencies who can destroy a life with a simple "photographs of x indulging in unspeakable vice" prompt
next steps - greater processing power will enable video, film, stories - a director could type in a prompt "car being chased by 60 foot mechanical crocodile in the style of Kubrick" or whatever and the sequence will be rendered
---
and - art is still 50 years ahead of writing -
I've been interested in neural networks, AI, procedural generation, etc., for years, and I can say that the text generation AIs are really shoddy compared to what is happening with the visual side at the moment