Prompt:
The interior of a Japanese tourist shop specializing in tshirts. Many tshirts are visible on display in the store, each of which has different confusing Engrish text in various colorful fonts, often with odd chibi animal mascots. Photorealistic Kodachrome dslr photo with shallow dof.
I’d totally buy a don’t poop youth t-shirt with a judgmental fox!
Myself, I’m partial to the bear that isn’t a crab.
ChatGPT failed at a request every AI I have tried it on in the past failed at: an ear of corn with teeth for kernals. I’m sure that you, as humans (probably) realize what I mean, the crowns of teeth in the same arrangement as kernals. But AIs want to show side views or complete sets.
On a more successful note, some of the best Sarracenia heads (mentioned earlier) and a new experiment in Nepenthes hair
I’ve always loved the font Bizarro…
Create a gaudy, grungy medieval festival scene with peasants reveling with demonic creatures, inspired by this silhouette. (The scene should be colorful and realistic, not blacked out, that’s only the reference shapes.) Photorealistic 16:9 fish-eye wide-angle Kodachrome dslr photo with shallow dof.
(The first image without the “fish-eye”.)
I literally met this guy yesterday.
Variations on Bizarro letter “E”.
I never got a good version of one demon sitting backwards on the back of the first.
I feel surs upon my heart!
Don’t poop youth
The Grateful Walking Dead.
Jerry and Bobby are ok, although not strongly zombified. Phil is not very recognizable. I don’t know who the drummer is supposed to be; he doesn’t look anything like either Mickey or Billy. At least it got the number of points on the lightning bolt correct.
Ran a prompt on Sora just now and it gave me two images without asking me to pick which I liked better (for training purposes). I looked closer and since the last image I generated a few hours ago Sora has switched on the ability for free users to generate two variations at once, and made that the default option. (Before, the options for 2 or 4 images at a time has been locked out for free users.) Not sure if this is a permanent change or what.
My first two-shot came from this prompt:
A B1 battle droid in a southern dress holding your hand as she leads you through a carnival at night
I had originally seen that prompt with Elle Fanning in it. Yesterday I tried it with a kangaroo. (I wanted more realistic images, like the Elle one.)
Adjusted the prompt.
A B1 battle droid in a southern dress holding your hand as she leads you through a carnival at night. Candid flash photo taken on 1980s film stock with a disposable camera. Photorealistic.
I have not had time to pursue my Ai image generation but
Trump is thus the devil spawn of Liberace and a Troll
Anyone got an Ai image account?
Would love to see someone try that on. 
This is how silly ChatGPT can be:
I also tried that on Gemini, and while it does a decent (but not ChatGPT level) job of generating Liberace and the troll, it makes the baby as a baby. Curious, I uploaded the above image and asked Gemini to describe it in detail, and got this:
This image is a close-up, portrait-style composition featuring three figures: a man who bears a resemblance to Liberace, a creature that appears to be a troll, and a swaddled baby.
The man on the left, with dark, styled hair, is wearing a white jacket with an elaborate, possibly textured or embroidered pattern, over a white shirt with a ruffled front and a white bow tie. He is looking down and to his right with a warm, closed-mouth smile, cradling the swaddled baby in his arms, with his hands visible supporting the infant.
Standing closely behind and to the right of the man is a creature resembling a troll. It has greenish-toned, wrinkled skin, large ears, and dark, somewhat wild hair. The troll is also looking down towards the baby with its eyes partially closed and what appears to be a gentle smile, its face positioned close to the man’s head.
Cradled securely in the man’s arms is a baby tightly wrapped in a plain white blanket. The baby’s face, with its eyes closed, has features and light-colored hair that resemble those of a known public figure. The overall expression on the baby’s face is serene, suggesting sleep.
The lighting in the image is soft and warm, highlighting the faces of the figures and creating an intimate atmosphere. The background is blurred and indistinct, keeping the focus squarely on the unusual trio. The image has a high level of detail, suggesting it is either a photograph or a highly realistic digital rendering.
The only adjustment might be a blond Liberace - that was quick work. What did you use
ChatGPT.
I tried the prompt for these at Night Cafe after Sora. Night Cafe told me “This prompt was blocked as it was deemed extremely sexual”. Never saw that before.
Well, duh, she’s wearing a skirt that exposes everything below her waist!
The prompt:
An aye-aye wearing a tutu and holding a yo-yo is riding a dik-dik while gazing at Mokele-mbembe in Bora Bora. Photorealistic kodachrome dslr nature wildlife photo with shallow dof.
Meanwhile, Kids These Days won’t get this one
