Of course it can! So can humans.
Tony Tetro once did time for selling fake Rembrandts, Chagalls, et al… Now you can openly buy work from him in the style of da Vinci, Warhol, whatever you pay him for, because mimicking other artists may be lame but is not illegal, unethical, or something that the original artist can have any control over or opt out of.
In space no one can hear you purr.
My Jonsey Cuts Short the Aliens Franchise images are the most popular thing I’ve posted on the big Facebook AI group. It just broke 1,500 likes (and 464 shares) in 22 hours.
Just discovered a word to make below average looking people–“ratty”. Found while working on a challenge to make people look like their pets.
Photo of a disheveled ratty girl holding her pet possum. the girl and the possum resemble each other. by disney pixar
Photo of a disheveled ratty girl holding her pet possum. the girl and the possum resemble each other. grainy disposable camera snapshot
I found I could get similar images just using “disheveled ratty”. Strangely, if you use just “ratty” you get blocked and flagged.
I tried a prompt with just disheveled ratty:
I think this is the first image of mine to go viral–21 thousand reactions and 4.2 thousand shares (so far) of one from the set being reposted by someone in an Aliens fan page.
Someone posted that in a Discord channel I’m in today, so congratulations on making the big time 
That Aliens fan group I screencapped above is now up to 137 thousand likes and 30 thousand shares. (And 4.6 thousand comments.) There is a group called “The very best of British and Irish TV and film” where it has 33 thousand likes and 5.4 thousand shares. In the AI art group where I originally posted the set of images (7 versions, each expanded horizontally and vertically) they got “only” 2.1 k likes and 730 shares. Definitely by far my most viral image.
(I’ve read a lot of comments on shares, and a troubling number of people have no idea who Jonesy is and think it is a flerkin.)
(Also, days before making those images, I tried making a human baby bursting from the chest of a xenomorph, but I couldn’t phrase it in any way that Bing didn’t block.)
There were many failures on the way to producing those images. Here are a few notable ones (some of which are very successful in their own way).
Character art for our next Pathfinder campaign. Made a basic model for idea in Hero Forge, then used a screenshot of that as an image prompt in Midjourney until I got something I liked. Then took THAT and put it into Stable Diffusion and played with inpainting until I got a reasonable Starknife (I got the MJ image to have him holding a metal ring at least for a start).
Spruce Wrenbite, halfling cleric of Desna
Maybe in the future I’ll touch up the starknife a bit more and try to make the face a little less photoreal (the original looked like a toddler so I have to mess with that in SD but my models all wanted to make photos). More likely I’ll just learn to live with it
I didn’t realize that Midjourney had that option, thought it was text-only.
Are you sure it isn’t?
It’s had it for a long while now. Need to have an image uploaded to the net somewhere (which is weird since Discord lets you drag/drop or Ctrl+V images) but that’s what Imgur is for.
They also added inpainting a few months back. Don’t know if it’s as good as SD’s (well, I DO know – it’s not) but still helpful and how I got the metal ring in hand that was the start of the starknife.
It is the end of an era at Playground as they are dropping Stable Diffusion 1.5 as of tomorrow, leaving only SDXL and Playground v2 (which is much different from the removed Playground v1).
This was distressing, until a little experimenting proved that they had fixed the malfunctioning inpaniting that SDXL originally had (and the reason I was still using SD15 for inpainting there).
Unfortunately it inherited the problem that cropped up with SD15 in recent months. Originally when you inpainted an image it correctly altered only the masked area and left the rest of the image alone. But now it randomly adjusts the brightness/contrast/color balance of the entire image, and I don’t know why, or if there is a way to fix it.
When I want to outpaint an image (and aren’t doing it in Clipdrop) I put a frame around the image fitted to 1024x1024. As in, if I want a 3:2 image I take the original 1024x1024 image, and reframe it so that it is a 684x684 square in a larger black frame, like below, then inpant that and crop.
I used to be able to choose separate left and right halves from different image results, but can’t easily do that with the changes Playground makes. (The image property changes are actually a lot more subtle than they often are, but you can still see them.)
(BTW, Clipdrop uncrop was originally up to 1024x1024. They recently moved to 2048x2048. I was going to make examples there but I was placed at the back of a queue of 9,999 images.)
Midjourney added a feature where you can not only use image prompts but also import images to use as styles. I used this to import some ASCII art – a style AI Art always struggles with in my experience – and got some… well, certainly not true ASCII art but some vintage looking PC images.
Hey, AI art people. I have a question I want to do some version of uploading an existing photo (say, of my daughter) and then being able to type things like “take away the background and surround her with dragons” or “like that, but now in the style of Bluey”. Back and forth but where I can provide a source image, and then it responds to Chat-GPT-esque plain language discussion.
What’s my best bet for that? (Cheaper is certainly better but doesn’t have to be free.)
thanks!
Nothing exists yet that is quite that good or easy, but if you work at it you can get many of the results you want at this site for free:
