#StableDiffusion's image-to-image mode is more interesting to me than it's text-to-image (which ignores too much of your prompt). Here I used my Duckface painting as the input, and gave it the prompt “do what you like”. It's managed a coherent, if rather less funny, image. I interpret it as a rather dark take on drunkenness.
“A fox crossing a residential street. The fox has a human face. There are autumn leaves on the ground, terraced houses in the the background, and a slight mist.”
It's just ignoring most of my prompt (as well as really struggling with what foxes look like). I've tried many iterations and variations, they're all like this.
It's kind of funny how machine-learning AI works. If you feed Stable Diffusion XL "Pale woman,mirror,reflection" the result has no mirror or reflection. If you prompt it with "woman, mirror, reflection", there she is looking into a mirror. Of course, SDXL doesn't know how mirrors work, so the reflection doesn't match the original closely, but it's recognizably the same woman in the same dress. Interesting, SDXL gets depth-of-field right in this image.
Today I realised that if you have @elementary 's houseclening setup to delete files older than a month the Downloads folder, then it's not smart to use that folder as a temporary folder for important stuff.
In unrelated news, today I also realised that I lost most of my #StableDiffusion generations older than a month.
@elementary I love it. And I love that in 7 this extends to the screenshot folder, since mine is full of garbage. But between working and teaching, I hadn't have the time to upgrade. 😭
Just a quick note about my content and AI. If something I post is created using AI, I will say so in the post. If I do not hashtag it with #stablediffusion or some other AI software then it is an actual photo.
@technoshaman001 Because, I want to have a clear distinction between the two art forms. I post a lot of both. I think it is important for those that want to get to know my work and what I do to know there is a difference.
interesting messing around with #stablediffusion when you find a concept that the ai heavily favors. tried to generate some halloween broccoli. couldn't generate anything but regular broccoli.
or maybe the two concepts are just so different it has no connections between the two from training?
@wagesj45 I suspect if you describe give it more precise instructions, it going to give you something closer to what you expect. @stablehorde_generator draw for me a Broccoli monster style:sdxlhorror
Tried to get #stablediffusion to solve the p=np problem. Can some #mathematicians help explain these results? I wanted to be fair so I asked it in both orders.
Prompt: "pure red background if p=np, but a pure blue background if p!=np"
