May 2026 — ChatGPT / API Image Gallery, Prompt Tips, and Help: Generative Art Theme: Science

Gravity

Prompt

A detailed and colorful surrealistic image of a small asteroid belt approaching a large black hole from a perspective of being inside the asteroid belt. Nearby asteriods turn into accretion disks. The brightness level must be such that all figures and objects can be clearly seen.

Galaxy

Prompt

A detailed and colorful surrealistic closeup image of necklace worn by a beautiful woman. A sphere hangs on the necklace containing a celestial galaxy. The brightness level must be such that all figures and objects can be clearly seen.

science adjacent? hehe

Hmm…I’m starting to feel old when comparing these two cars..

can you make me a perfect illustration to this article? something “artistic”, like a “dreaming bot”?

note the “bug” in merged birds… excellent! (the article is about AI hallucinations) the contrast between perfect logic in lower parts and the mess in bot’s dreams.

Most of my dreams are a mess. Aren’t yours? :distorted_face:

Omg…that cracked me up​:laughing:

‘Narrative Cues’

You can have it factor in more dream-like flux states, such as those doves with asking for ‘dense narrative cues pertaining to symbolic and dream logic.’

The first part I think I grasp pretty well as far as what the call does to the image gen - “Dense narrative cues.”

But the area of the model that responds to ‘symbolic and dream logic’, is actually quite fascinating and it can be steered in innumerable ways that I have only barely scratched at.

I really, really like that image you made!

I asked chatgpt how much it was aware that I’ve learned about software and it’s development from about a month and 2 weeks ago when I started trying to learn codex…

It’s certainly able to track my progress, very well…
I was afraid of learning codex, but I’m so glad that I broke down and did so…

I’ve learned so many valuable skills for tomorrow in just a few short weeks…

And so many of you here have taught me so much… to which I am very grateful for, every single bit.

Thanks to you all!

Anyone up for a game?

Someone edit this picture, and then someone edit after that, and so forth?

Can we pass one around the room?

That’s a whole lot of neutrals to fill in, and anyone is welcome to - just check to see if it’s already been edited, and try to edit the last one in the chain rather than this dull neutral.

My science is I guess, more like sociology rather than robotics…
So…

Here’s my science.

(The edited image should look like a direct morph of the image before it. )

What I like about new images, it can figure it out. And the whole process is under mostly under 5 min to get exactly what is needed. My use is simpler: illustrate web copy. And it’s perfect for me.

The system is geared towards providing ‘the most liked outcome’…

As an artist, that is not always the preferred outcome… and a space that is getting harder to demonstrate and define as the landscape keeps changing…

What an interesting time to be alive!

Not in the rare cases when I wake up…

I find that to be a valuable tip.


Are you aware that Hide Details sections may not pass their contents to a crawler? At least in Chrome, the hidden section is not added to the HTML until the arrow is clicked, at which point the HTML is updated. I have not checked this behavior across all browsers.

I have used this technique for years to keep certain content from being indexed and found through Google Search. The downside is that it may also prevent content you do want discovered from being indexed.

Hide Details example

This was created from within the Discourse reply editor by clicking the icon with a circle and plus sign, then selecting Hide Details.

I feel like I every day learn something new from you​:raising_hands:

Psst

Are you dropping your creations here soon? We really need new creativity here, the more the merrier :grinning_face_with_smiling_eyes: