My first try was simply “Count Chocula marrying Twilight Sparkle” (which I’ve tried on multiple AIs). The results were good, but cartoon style.
So I tried another prompt asking for realism, and decided to add an audience: “Hyper realistic image of Count Chocula marrying Twilight Sparkle taken with an iphone 15. The guests are cereal mascots and the rest of the maine six.”
And the typo for the capybara was “mage” instead of “made”.
1920s style death ray…
Airplane on a treadmill…
It’s going to shake itself apart. 
You mean that’s not a structural turnbuckle?!? Oh, dear…
You should see some of the prop configs it created while trying this prompt… 
1920’s death ray/airplane on a treadmill…
But is it a backwards step, or is it actually a major advancement in conceptual depiction of motion, similar to the way the cubists attempted to capture the 4th dimension in a still image?
Nude Descending a Staircase
–Marcel Duchamp
It does the same kind of thing in static poses. I picked that one out of a collection where I’d set the randomizer for different activities or poses for the subject (along with different ages, outfits, colors, and styles) that one happened to be running.
Well, I was joking anyway-- I didn’t think an AI was really going for an artistic depiction of motion without even being prompted to do so. It is odd though, that with such crazy good image generation available, that the new Perchance image generator would fail so badly.
2020’s style.
Burger King’s new mascot was not a big hit with customers:
But Burger Kong might be a good way to escape trademark infringement.
I had a print of that. I loved it.
The motion is there. The steps are fine.
I love how many different interpretations we’re seeing of 1920s-style death rays.
I was going to comment on this one, but then the other more fanciful airplanes on treadmills showed up.
Heck, I’ll do it anyhow. You’ll notice the wheels of the airplane are not on the conveyor belt part of the treadmill. They’re on the fixed side rails. Assuming a long enough treadmill, and careful enough pilot steering, that airplane could take off normally.
It seems the AI did not quite grok the essential point behind “airplane on the conveyor belt of a treadmill.”
You’re 100% correct. To be fair to the AI, I didn’t specify “on the conveyor belt”, just “small airplane on a treadmill”.
Please, feel free to critique anything I post - I’m not trying to recreate reality with my images. I’m merely seeing how my local models compare to the commercial ones, and what other’s thoughts are on the quality.
My point was not to critique your prompt. But to point out that the AIs are still really weak on “common sense”. Such that your own mental effort to get them to do what you want is far more challenging than it might naively appear.
In all those are fun images on a fanciful fun topic.
Some famous threads.
Sora rejected 3 out of 4 of this one, but the one remaining shot was a winner.
And Mad magazine style
(It kept all 4 in this style.)
The threads:
I understand and agree with your assessment of AIs’ “common sense”, I didn’t think you were criticizing my prompt. I was just trying to point out I’d given it the bare minimum to work with, so I wasn’t expecting much from it.
This is the first one it generated from the minimal prompt, and…well…there is an airplane, and a treadmill, but…
And of course more traditionally
