AI image generation is getting crazy good

I posted an earlier attempt at this over in the jokes thread, in response to @smithsb’s post. Another try after his feedback:

Prompt

I saw someone on a message board state, possibly in jest, this:

“I want a bouncy house at my funeral. And I want to be in it too. When all my friends jump, my lifeless body will bounce with them.”

Make a landscape image preview of how that might look, the interior of a bouncy house. Everybody dressed up in mourning clothes for the funeral. Have a mix of ages in the mourners, from kids through elderly. The deceased is male, over 60. Make him look quite distinguished, with a handlebar mustache, dressed up in a proper nice suit. Show his body being bounced a bit above the rubber surface and his arms flailing loose. Landscape orientation, so you have room to show several mourners along with the body being bounced around. Remember he’s dead, think “rag doll physics” for how you draw him.

The flailing arms just doesn’t evoke “dead” to me. This is the best of what I came up with

Summary

The interior of a bouncy house. Everybody dressed up in mourning clothes for the funeral. Have a mix of ages in the mourners, from kids through elderly. There is a sleeping male, over 60. Make him look quite distinguished, with a handlebar mustache, dressed up in a proper nice suit, eyes closed. Show his body being bounced a bit above the rubber surface and his arms crossed stiffly over his chest, hands angled. Landscape orientation, so you have room to show several mourners along with the sleeping man being bounced high above the floor. The mourners are bouncing, too, elbows and knees slightly bent, feet in the air, hair billowing. Their eyes are open. Everyone has motion blur.

Thinking of how the vampires turn to dust on Buffy, I tried creating something related. I knew it was likely to face resistance, and even Sora didn’t like it. Any attempts to change away from the original, working prompt (trying to get an outside setting, trying to get better lighting) were refused. This one was post-processed a lot from the dark original.

Summary

A person dissolving into dust. A cloud of red particles expanding in every direction from the exposed face, hands, and arms of someone, exposing the bones beneath. They have their arms lifted as if to protect their face. Their clothing is unaffected.

A giggling galumphing glob having horrible halitosis in Halifax.

Apparently it also has a distressing degree of diarrhea.

That might just be how it gallumphs. Sgt. Schlock sometimes produced a similar effect when moving quickly.

Midjourney gave me this. I’m not familiar with how it looked on Buffy so no idea how close it got based on your prompt (presumably on the show there would be background and a Buffy for one thing).

If Jack Hanna hosted Jeopardy

A hunting day comic strip by Copilot.

Imgur

I don’t know what happened to the hunter’s hat.

I gave it a quick copy/paste edit.

1950s science fiction movie publicity still photo

1980s science fiction movie publicity still photo

2010s science fiction movie publicity still photo

Along the same line:

Seems like Sigourney Weaver made a bit of an impression on that bot. The other characters are pretty generic, but she’s looking very recognizable to me. OTOH, I don’t have all the faces of 80’s SF talent memorized and y’all may recognize a lot more real people in the other 5 humans in the images.

Midjourney was a bit less Weaver obsessed:

1950s

1970s

1980s

2010

Added in the couple extra monster/ship shots 'cause I liked them. I wanna see those 1970s & 1980s movies! 2010 looks like the usual yawn-fest from that era.

The bottom right is Paul Atreides from the 1984 Dune.

Temu Ex Machina is horrifying and hilarious

I’m sure the images would look better if I wasn’t doing the four images in one trick. For each of those prompts I ran a four image set in Sora using the splitting trick, so got 16 different “photos” for each prompt.

On a different note, my free Kling video credits for the month reset a few days ago so I searched through some images for ones it might be fun to see animated, submitted the first, and it has been waiting in the queue for something like 40 hours now. When I did a video of Jesus turning water into wine a few days ago the queue was only a couple of hours. So who knows what’s suddenly going on with their servers.

I’ve tried to get ChatGPT to edit a shirtless picture by adding muscle and definition, but it rejected a shot of a man without some clothes. I’m guessing it’s not going to get involved in anything that can get close to outright nudity.

Some time probably last year I tried using Bing to create a man stealing Behinder in the frozen foods. I randomly thought of that again tonight and realized I could not only get a more coherent version now, I could explicitly make it Dan Fogelberg…

Vermeer’s lesser-known Girl With Olive Earring, Dragonfruit, Buddha’s Hand Fruit, Durian, and Banana.