Pinned Tweet
Hello twitter. Important survey attached; please complete:
53%yes
47%no
36 votes • Final resultsImagine being a random tech support scammer and getting an angry call from Greg Egan
Evan Banyash retweeted
Replying to @_sambk @images_ai
Taking my epic beautiful fashionable tattoed girl portrait professional photography trending on artstation octane render by wlop and running it through a wombo tier motion model to give it worse lip syncing than Vampire: The Masquerade Bloodlines
There we go, fixed!
Part four of catching up with the #septembAIr2022 challenge - prompt No. 4 was "Echo"
This piece is called "Echoes"
#septembAIr
I wish the artists would just say it: "I spent my life grinding to eke out a meager living in the most difficult industry and now machine learning will make it all for naught. Socialism/UBI isn't happening so that's not going to help me. It's over."
My field is literally next, programmers are going to be obliterated within the next couple of years. We can't face the problem if we won't admit it. Enough with this vague nonsense about "not real art" and "plagiarism". That's not what you're really talking about!
These text to image models are great but I'm getting really tired of the square pictures. You can go wider with Stable but you lose coherency since it was trained on squares. Best pictures in the dataset are undoubtedly wider, I'd like a model trained on 16:9 crops instead.
Pssst guys I made something epic with Stable Diffusion
youtube.com/watch?v=YwkJhwMP…
Got init images working with #StableDiffusion and had it paint me as "A portrait by Dean Cornwell" and boy do I wish I looked like any of these, this is gonna be uuuuuuhhhh difficult for a lot of people once it inevitably hits faceapp and all the others.
You can generate infinite variations of "yourself but more attractive" in the style of any artist, I assume because usually their work is biased toward attractive subjects by choice. I mean, it makes sense. What a time to be alive...
You literally "enslaved" an AI for the pictures on your website, and you didn't even so much as credit them either! This is a horrific abuse of machine learning model rights.




















