The Official ImageGen, 4o and Dall-E Megathread

They’re enabling aspects of gpt5 i heard on the downlow.
agent mode going wild too.

oh and got this printed up today for the farmers markets

practically highway robbery selling prompts for 7.50 that literally write themselves according to my prompt writing contexts

prompt

ChatGPT - Elven ear image description

Oh they are ears…

The image is completely blank with a white background. (Captioned by AI)

Tried to make them flap but it’s difficult lol

@polepole how did you do the sears catalogue mockups? That’s brilliant. Surely not “four identical quadruplets wearing identical though differently colored shirts”?

Also…Sora is fun! With mixed results. I’ve been trying all sorts of prompts from beginning with pictures, to feeding it philosophy, to business stuff, to just storytelling.

I DO wish we could feed them here.

Which post, can you share the link or post number?

EDIT:

Here there is no prompt similar to “four identical quadruplets”.

I think you asked following post:

The Official 4o and Dall-E image Megathread - #817 by polepole

For that I used different prompts based on size or positions.

Especially for full-body in multiply grids some words are important for no cropped image:

A single full-body portrait lonely in focal point too far distance a [FIGURE]. Her/His feet/shoes are eye-catching at first glance. [WEARING] The character is centered in the frame, fully visible from head to toe [SKIN DETAILS] Standing too far away. The man/woman is perfectly centered in the image. Only one character is shown, fully visible and facing directly forward. [POSITION] The entire scene is framed to emphasize the character’s full body from a distant perspective. Shrink the full-body of the character to fit in the image.

Examples:

Prompt-1

Layout
• Main panel centered horizontally
• The same figure exactly same pose in main panel and in two left sidebar and in two right sidebar:
Ultra hyper detailed photo-realistic style tall image, a single full-body portrait lonely in focal point too far distance a smiling young Yoruba woman with an athletic build and rich brown skin. Her bare feet are eye-catching at first glance, she wears no footwear, just a fitted plain color Yoruba iro (wrapper) and buba (blouse) made with elegant fabric, a matching inner wrapper or skirt, and a large, elaborately wrapped plain color Yoruba gele (head wrap). She holds a matching plain color Yoruba shawl or scarf draped over one arm. Her face is striking and beautiful, with soft, even features, and she holds a neutral gaze. Standing too far away. The woman is perfectly centered in the image. Only one character is shown, fully visible and facing directly forward. The style is ultra-realistic, focusing on cultural symbols, skin texture, and heritage details. The character is centered in the frame, fully visible from head to toe. The background is a softly blurred turquoise sea and pale sky horizon under bright overcast beach lighting. The entire scene is framed to emphasize the character’s full body from a distant perspective. Shrink the full-body of the character to fit in the image.
• Width: 614 px (60% of canvas)
• Left and right sidebars
• Each: 205 px wide
• Each sidebar contains 2 vertically stacked thumbnail frames
• Vertical spacing between thumbnails: 12 px
• All gutters and margins: pure white (#ffffff)
Main Panel (Center)
• Wearing a fitted, plain white Yoruba iro (wrapper), white buba (blouse), and a white inner wrapper/skirt
• Large, elaborate plain white Yoruba gele (head wrap)
• Holding a plain white Yoruba shawl or scarf draped over one arm
• Background: Softly blurred beach scene
• Turquoise sea and pale sky horizon
• Horizon line exactly at y = 575 px
• Lighting: Bright overcast daylight, soft shadows
• Color balance: Neutral white balance, consistent across all panels
• Fabric texture: Crisp, visible 8K weave
Left Sidebar Thumbnails
• Shrink the full-body of the character to fit in the image.
• Identical pose, crop, lighting, and background as the main panel
• Only attire and gele color changes:

  1. Saffron yellow iro and buba, saffron yellow gele and shawl (top)
  2. Emerald green iro and buba, emerald green gele and shawl (middle)
    Right Sidebar Thumbnails
    • Shrink the full-body of the character to fit in the image.
    • Same structure, identical to left
    • Attire and gele colors:
  3. Coral red iro and buba, coral red gele and shawl (top)
  4. Violet purple iro and buba, violet purple gele and shawl (middle)
    Camera and Render Settings
    • Camera distance: z = 2.2 meters
    • Alignment: Frontal, eye-level
    • Lens: 85 mm full-frame equivalent
    • Depth of field: Soft background blur, subject sharp
    • Consistent lighting and exposure across all 5 frames
    • Accurate color rendering for fabric
    • No tonal shifts or lighting imbalance between thumbnails
    Negative Prompt
    • Do not crop sleeves, wrappers, head, or feet; full-body must be visible
    • No pose changes, no sleeve, or wrapper length variations
    • No logos, text, or watermarks
    • No duplicate or incorrect attire colors
    • No background variations
    Seed: 42

Prompt-2

Canvas & gutters
• Matte-black background flush to every outer edge; absolutely no horizontal or vertical white bars except where specified.
• One 8 px vertical white gutter separates main panel (left) from sidebar (right).
• In the sidebar, exactly 3 portrait-oriented thumbnail frames, each separated by an 8 px white spacer.
• No other outlines, keylines, or baseline strips anywhere on the canvas.

Layout — right-weighted composite
• Main panel ≈70 % of canvas width (viewer’s left).
• Sidebar occupies the remaining ≈30 % (viewer’s right).
• Thumbnails are equal-height, sized so that a head-to-hips view plus at least 40 px of free black space below the wrists fits wholly inside each frame.

Main panel (unchanged spec)
• Full-torso Swedish woman, early-30s, very fair freckled skin, vibrant natural red hair in neat low braid/bun, oval face, calm soft smile, light blue/gray eyes.
• Long-sleeve linen Henley-style tunic (untucked, buttoned placket) + off-white tailored trousers.
• Front torso only shows symmetrical Swedish kurbits embroidery (rich red, forest green, gold ochre, navy) in a straight vertical column, collar to hem.
• Plain sleeves. Pose: relaxed, hands at sides / in pockets, facing forward.
• Soft studio lighting; matte black backdrop.

Sidebar: shirt-colour variants (face, pose, lighting identical)
1️- sky-blue
2- buttercream-yellow
3- forest-green
• Each thumbnail is a scaled-down head-to-hips shot; both wrists fully visible with an extra safety margin of black backdrop below them (no bottom cropping).
• Only the shirt’s base colour changes; embroidery palette, trousers, background remain identical.

Technical notes
• Photorealistic 8 K clarity; crisp linen weave and embroidery threads.
• Embroidery is flat, front-only; none on sleeves, shoulders, or side seams.
• No motif distortion on fabric curves.
• Uniform lighting across all four panels; precise 8 px gutters and spacers; zero cropping or truncation anywhere.

Negative prompt
text, logos, watermarks, sleeve/arm motifs, asymmetrical embroidery, duplicated limbs, surreal backgrounds, extra graphics, branding, wrinkles obscuring embroidery, cropped wrists or missing hands in thumbnails, bottom- or top-cropped sidebar frames, horizontal or vertical bars beyond the defined gutters/spacers.

Render main panel first with seed 42. Then render the three colour variants reusing seed 42 (identical settings), changing only the shirt-colour token. Composite in a 1024×1536 canvas per the layout above.




Image(Square, Pin Art Earth)

The image shows a completely white, blank canvas with no visible objects or features. (Captioned by AI)

The image is completely blank and white. (Captioned by AI)

5 Likes

The image is completely blank with a plain white background. (Captioned by AI)

6 Likes





I LOVE these…especially the owls!

My Fairy Queen looks insane on aluminum. Just got the sample from my man moments ago.

That being said, ive looked at the prices on aluminum printing on the internet and my pricing wrecks those.

If you folks want something like this done for your art or mine, lmk!

… with dall-e-3 API …

Getting full-body portrait

40 mini. Greed.

Yes, @polepole, those are the prompts I was after.

And y’all check this out!

https://sora.chatgpt.com/g/gen_01k1w9dq2gfpar84dz7j41z0t2

OMG! :heart_hands: This was built from an image prompt after like 15 generations. Each new attempt got slightly closer to the desired effect, and all I prompted was,

“Pan around the armor as though it were 3d,” in addition to the image prompt which you can see somewhere up above in the beginning of the video.