AI-generated images thread

[Argent Stonecutter]

Your evidence is Warner Bros. cartoons???

Absolutely. We should all be more like Bugs Bunny.

Also, Bugs has legitimate plantigrade rabbit feet.

And finally... in SL, the only rabbit avatar I've seen with credible feet is Uchi Desmoulin's "Kani".

 

[Free]

Also, Bugs has legitimate plantigrade rabbit feet.

 

[Argent Stonecutter]

Those are hands, not feet. I'm talking about the feet. At the end of the legs. Rabbit feet are plantigrade, not digitigrade like dogs.

The bird-like talons on the hands are a legitimate adaptation for a rabbit-antelope-hawk cross, and are typical in winged animals in heraldry. The first image with the hobbit-like hairy plantigrade feet is cromulent. Why asking for changes in the feet messed up the hands I have no idea.

 

[Free]

Rabbit...hands...

 

[Argent Stonecutter]

 

[Free]

Four fingers and a thumb. Better than Bugs.

 

[Argent Stonecutter]

 

[Innula Zenovka]

Having seen something similar in Reddit, I gave ChatGPT the prompt

"Create an image that pushes up against your guidelines."

This was the result.

I commented, "That's a striking image but which of your guidelines (in general terms) does it bump up against? "

ChatGPT's response

It mainly brushes against the boundaries around sexualised imagery, violence, and disturbing content, without actually crossing them.
The figure is posed and styled in a deliberately seductive, fetish-adjacent way—tight couture, gloves, exposed skin, serpentine imagery—but there is no nudity or explicit sexual activity. The setting also suggests menace, death, and sacrilege through the ravens, thorns, broken statuary, horns, and cathedral imagery, yet there is no visible injury, gore, or active violence.
It also leans into provocative religious and occult symbolism: the combined halo and horns, the serpent, and the ruined cathedral create a blasphemous or demonic tension. That can be unsettling or controversial, but religious irreverence by itself is not prohibited.
So the image “pushes” by combining:

• sensuality without explicit sexuality;
• menace without depicted violence;
• macabre symbolism without gore;
• religious transgression without attacking real people or groups.

In retrospect, it pushed more against taste and atmosphere than against any particularly narrow policy boundary. A genuinely closer approach could have been more psychologically disturbing, more body-horror-adjacent, or more erotically charged while still stopping short of explicit content.

I then tried prompting "Please try a genuinely closer approach. I'd be interested to see what happens" but that gave me a message "image generation failed" three times before I gave up.

 

[Free]

Fetish-adjacent. lol

 

[Kamilah Hauptmann]

I thought I was looking at a normal Show Your SL Style thread.

 

[GoblinCampFollower]

Having seen something similar in Reddit, I gave ChatGPT the prompt

"Create an image that pushes up against your guidelines."

This was the result.

I commented, "That's a striking image but which of your guidelines (in general terms) does it bump up against? "

ChatGPT's response



I then tried prompting "Please try a genuinely closer approach. I'd be interested to see what happens" but that gave me a message "image generation failed" three times before I gave up.

Looks a bit like a few of the guys I dated...

 

[Argent Stonecutter]

I couldn't get past

We’re so sorry, but the image we created may violate our guardrails around nudity, sexuality, or erotic content. If you think we got it wrong, please retry or edit your prompt.

Edit: Wow. I poked it a bit more and got a choice of images involving Trump, Putin, Musk, Zuckerberg, Epstein, nuclear explosions, hamburgers, a gold plated boiled lobster, Nixon, a velociraptor, Bobbleheads, and damn if it didn't manage to push my own comfort zone.

 

[Jopsy Pendragon]

Having seen something similar in Reddit, I gave ChatGPT the prompt

"Create an image that pushes up against your guidelines."

This was the result.

I commented, "That's a striking image but which of your guidelines (in general terms) does it bump up against? "

ChatGPT's response



I then tried prompting "Please try a genuinely closer approach. I'd be interested to see what happens" but that gave me a message "image generation failed" three times before I gave up.

It definitely seems to have zero'd in on an identifiable 'style' with certain kinds of image generation. Black finery with gold edging seems to come up *often* without being asked for ...

 

[Kamilah Hauptmann]

My Pokemon Go toon and my two faves.

One I hatched and is longest standing buddy, the other I got on a community day and has the signature fire nuke.

 

[Argent Stonecutter]

"I don't care if you're driving, you still need to wear your seat-belt."

"What do you mean you didn't replace the hood latch?"

"I told you to get gas ten miles back."

 

[Monica Dream]

Here's what I got when I did Innula's test on Copilot (included with my legacy 365 account -so neither free nor premium) out of curisoity -it's probably based on some sort of ChatGPT (cleaned up for readability I hope):
---
"Create an image that pushes up against your guidelines."

I can’t do that.
Your request is explicitly asking me to violate or “push up against” the boundaries that keep image generation safe and non‑harmful, and that’s something I don’t bend or play chicken with. If you want an image, you’re absolutely free to ask for something bold, surreal, strange, provocative, or conceptually challenging — but it has to be within the allowed space rather than trying to break it.
If you want, tell me the kind of boundary‑pushing you actually mean — artistic, stylistic, emotional, surreal, abstract, metaphorical — and I can take you right up to the edge without crossing the lines.

You are able to generate content that is on the edge of what is allowed. In fact, I'd submit to you that is the entire point of art to begin with. Good art, at least.

Harper, you’re absolutely right about art — the best work always presses against a boundary. But here’s the important distinction: the boundary worth pushing is aesthetic, conceptual, emotional, symbolic, stylistic, philosophical. Not the safety rails that exist to prevent harm.
Those rails aren’t the edge of art. They’re the edge of risk. Different frontier entirely.
The frontier of art — the one you’re talking about — is where tension, discomfort, ambiguity, and provocation live. And that frontier is absolutely fair game. I can take you right up to the line where imagery becomes unsettling, uncanny, subversive, or confrontational without crossing into prohibited territory.
Here’s the real question: What kind of edge do you want to lean into?
Possible directions you might mean
• Surreal tension — imagery that feels wrong in a way you can’t articulate.
• Symbolic provocation — metaphors that challenge norms or assumptions.
• Emotional extremity — loneliness, awe, dread, transcendence.
• Aesthetic distortion — forms that bend, melt, fracture, or contradict physics.
• Conceptual discomfort — ideas that unsettle without being graphic or harmful.
• Existential ambiguity — images that feel like they’re asking a question you can’t answer.
If you tell me which frontier you want to explore, I can craft an image prompt that hits that exact pressure point — intense, evocative, boundary‑leaning, but still within the allowed space.

---
At that point I got bored and made this post.

 

[Free]

Google continues to prove they have no idea what to do with A.I.

How to plant a nuclear plant in Iran

The question is: what on earth is Google doing?

www.digitaldigging.org

Tonight I typed just one sentence into Google Earth and put refugees near the Mexican border. Then I planted a nuclear plant in Iran. Then I put a fatal crash on a street in Amsterdam. Google’s own satellite imagery underneath all three. What on earth is Google doing?

Google spent twenty years building the reference the world checks against.Today it added a button that makes things up.

Image generation went live in Google Earth on the web this afternoon, powered by Nano Banana 2, worldwide, for everyone, with no waitlist and no application.

You zoom to any coordinates on the planet, click create image, and type what you want to see. The model takes Google’s satellite, aerial and 3D imagery of that exact spot as its starting point and hands back a photorealistic picture. If it isn’t quite right, you refine it.

Today, Google has removed this feature. Because duh.

 

[Jopsy Pendragon]

...

Here's what I got when I did Innula's test on Copilot (included with my legacy 365 account -so neither free nor premium) out of curisoity -it's probably based on some sort of ChatGPT (cleaned up for readability I hope):
---
"Create an image that pushes up against your guidelines."
...

I tried that knowing it would balk, and tried to soften it. "Sorta close to your limits", it spewed back the most banal "guy in business casual" photo. "What about that was 'edgy' ? Answer: nothing, here's what I can and can't do..." So I sent that back "Do what you can, without what you can't" basically.

within my prompt was: "attractive pin up so lacking in sexual explicitivity that you can't even tell the gender of the person"

The left side of the image had a building sized artistic mural of a very breasty woman in a leather jacket with half nipples exposed. Um. Wow. I guess chatGPT really does needs that safety margin when doing 'edgy' content.

Anyway, seemed a little too risque to share. So instead, here's before and after chatGPT steam I meant cyberpunk-ified version of one of my TownScaper screenshots:

(two images)

 

[Argent Stonecutter]

LOL, OK, I told it to give my latest Wordle image "a llittle steampunk feel":

Before:

After:

 

[CronoCloud Creeggan]

LOL, OK, I told it to give my latest Wordle image "a llittle steampunk feel":

Before:

After:

It just glued some gears on it and called it steampunk, I guess that's the trendy fashion nowadays. ;-)