OpenAI announced Dall-e 2. openai.com/dall-e-2
They do not intend to commercialize it. The waitlist is for testing.
The system allows you to input any text and it turns it into images. Currently only available for use by openai employees and people granted access to the demo (handful of twitter users have access, are taking requests), you can't just go and use it yet, there is a waitlist. But the results are unlike anything I've seen coming from AI
Examples, all entirely generated with AI with input text:
twitter.com/sama/status/1511732024028983298?s=20&t=U_z3wj4Iev8ZX28BY--nsg
thread including some interesting art:
How long until we have AI creating genuinely good electronic music? Would you listen to AI created electronic music if it was enjoyable?
⠀⣞⢽⢪⢣⢣⢣⢫⡺⡵⣝⡮⣗⢷⢽⢽⢽⣮⡷⡽⣜⣜⢮⢺⣜⢷⢽⢝⡽⣝
⠸⡸⠜⠕⠕⠁⢁⢇⢏⢽⢺⣪⡳⡝⣎⣏⢯⢞⡿⣟⣷⣳⢯⡷⣽⢽⢯⣳⣫⠇
⠀⠀⢀⢀⢄⢬⢪⡪⡎⣆⡈⠚⠜⠕⠇⠗⠝⢕⢯⢫⣞⣯⣿⣻⡽⣏⢗⣗⠏⠀
⠀⠪⡪⡪⣪⢪⢺⢸⢢⢓⢆⢤⢀⠀⠀⠀⠀⠈⢊⢞⡾⣿⡯⣏⢮⠷⠁⠀⠀
⠀⠀⠀⠈⠊⠆⡃⠕⢕⢇⢇⢇⢇⢇⢏⢎⢎⢆⢄⠀⢑⣽⣿⢝⠲⠉⠀⠀⠀⠀
⠀⠀⠀⠀⠀⡿⠂⠠⠀⡇⢇⠕⢈⣀⠀⠁⠡⠣⡣⡫⣂⣿⠯⢪⠰⠂⠀⠀⠀⠀
⠀⠀⠀⠀⡦⡙⡂⢀⢤⢣⠣⡈⣾⡃⠠⠄⠀⡄⢱⣌⣶⢏⢊⠂⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⢝⡲⣜⡮⡏⢎⢌⢂⠙⠢⠐⢀⢘⢵⣽⣿⡿⠁⠁⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠨⣺⡺⡕⡕⡱⡑⡆⡕⡅⡕⡜⡼⢽⡻⠏⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⣼⣳⣫⣾⣵⣗⡵⡱⡡⢣⢑⢕⢜⢕⡝⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⣴⣿⣾⣿⣿⣿⡿⡽⡑⢌⠪⡢⡣⣣⡟⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⡟⡾⣿⢿⢿⢵⣽⣾⣼⣘⢸⢸⣞⡟⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⠀⠁⠇⠡⠩⡫⢿⣝⡻⡮⣒⢽⠋⠀
yes i would listen to AI music if it was good
S***s about to get so dope when the skills of artists and ai technologists intersect soon
i hope s*** like this never takes off for the sake of literally every creative
this s*** is so scary like how the hell is it that good
actually its not even that spectacular also the text to picture isnt as op stated, its not quite as specific. However, it might be useful for companies/people, but not really that much.
Also you have to feed it with sample pictures
i hope s*** like this never takes off for the sake of literally every creative
If anything this will push creativity into something absolutely unheard of
this deadass might automate illustrators away if it is as powerful as it is, all it takes is a couple more iterations.
When we get the first AI beef
!https://m.youtube.com/watch?v=GxoaYX5pi7Ys*** has me rolling every single time
Wall-E really predicted the future man.
imagine being able to input a prompt into an ai to make your own procedurally generated virtual world :scary: along with tech like augmented or virtual reality
humans are F***ED
actually its not even that spectacular also the text to picture isnt as op stated, its not quite as specific. However, it might be useful for companies/people, but not really that much.
Also you have to feed it with sample pictures
You don't feed sample pictures, all you feed it is the text. The system has all the data on their end so you don't supply any samples. Also seems like the more specific you make your request, the more it follows it
that second one HARD AS F***
concept artists are shaking rn
luckily the earth will be uninhabitable and all will perish before the AI uprising can take hold