Home › Design Assets › Yacht Poster, Collage Written as a Formula
Yacht Poster, Collage Written as a Formula · GPT Image 2 generated image

Yacht Poster, Collage Written as a Formula

Four real yachts run through the same algorithm — wherever the edge field spikes, the collage texture follows.

Prompt notes

✍️ What's really in charge isn't what the yachts look like, it's the silhouette-preserving field Φ — wherever Div spikes, that's where the halftone dots burst. Swap one word in SYS1/SYS2 and torn-paper collage trades places with risograph halftone; the seed is locked to the subject's hash, so swapping subjects needs no parameter tuning.

Full prompt

📋 Prompt YAML Sectioned Style1444 characters
Apply the same treatment to four real-world yachts with genuinely unconventional designs, laid out across a 16:9 two-by-two grid. INPUT: $SUBJECT, optional [$MOVE1, $MOVE2, $SYS1, $SYS2] — fall back to an auto-seed when none are given. Stage one, triggered only when movements or systems are missing: IF not provided: MOVES = random_choice(Poster_Movements, 2, seed=hash($SUBJECT)) SYS = random_choice(Graphic_Systems, 2, seed=hash($SUBJECT)+1) IDENTITY = NAME_FROM(MOVES, SYS) Derive a compositional field Φ whose job is holding onto the subject's outline and spatial layout. Φ = preserve_silhouette( edge_map($SUBJECT) ) // how closely it tracks the source F = ∇Φ // the type grid is laid along these flow lines Curl = ∇×Φ // rotation sets the collage tear edges Div = ∇·F // density becomes halftone / screenprint layers Left-half poster render: movement 1 supplies the palette, movement 2 the texture and print logic. Graphic systems steer the execution — say SYS1 = "Torn Paper Collage" shifts F along tear vectors, while SYS2 = "Risograph Halftone" scatters stochastic dots once |Div| crosses the threshold. Typography tracks the F streamlines, its hierarchy set by Div's magnitude. Close with genuine print-shop artifacts — misregistration, paper grain, a hint of age-yellowing. Constraint: keep the outer contour locked to $SUBJECT at no less than 92% IoU. Final render: 8K resolution, landscape orientation.

Related Cases

Xerox Poster of a Palm Torn Open to the StarsXerox Poster of a Palm Torn Open to the StarsA Stylish Tourist Gets Lost in a Torn Paper Travel CollageA Stylish Tourist Gets Lost in a Torn Paper Travel CollageAnemone Poster Grown in a PhotocopierAnemone Poster Grown in a PhotocopierTop Half Real, Bottom Half RepaintedTop Half Real, Bottom Half RepaintedFour-Grid Ad Mockups for Golden Week TrafficFour-Grid Ad Mockups for Golden Week TrafficReal Brick Wall on the Left, Painted Poster on the RightReal Brick Wall on the Left, Painted Poster on the Right

More GPT Image 2 cases

All GPT Image 2 cases →
Starry-Grey Hair, Purple Eyes, Embroidery Kept Low-KeyStarry-Grey Hair, Purple Eyes, Embroidery Kept Low-KeySame Moment, Photo vs. Watercolor — Which Reads TrueSame Moment, Photo vs. Watercolor — Which Reads TrueThree Panels, Three Moods, One Wink and a PoutThree Panels, Three Moods, One Wink and a PoutDusty Rose on Palace Steps, Embroidered in Matte GoldDusty Rose on Palace Steps, Embroidered in Matte GoldStone Wall Backdrop, a Red Saree That Never MissesStone Wall Backdrop, a Red Saree That Never MissesCamper Van Night: A Pink-Quilted Milky Way MomentCamper Van Night: A Pink-Quilted Milky Way Moment

Related guides

How to Write Wallpaper Prompts: Lock the Ratio First, Then Write Where the Icons and Text GoThe Complete Nano Banana Pro Prompt Guide: Structured Writing, Text Rendering, and Multi-Image ConsistencyHow to Write Negative Prompts: A Common Checklist and the Reasoning Behind It

Tools & Further Reading

IMAGE TOOLExtract a Color Palette from Any ImageIMAGE TOOLOG Image & Social Cover Generator OnlineAI Video PromptsGPT Image 2 AI Video Prompts & Outputs
TOP