GPT Image 2 · Dream Heist Key Visual

GPT Image 2 Dream Heist Key Visual: Folded Cities

A dream heist needs one frame that explains the trick before anyone speaks. Studio draws the folded city and the crew standing in it — one wide key visual, kept on-model by the world behind it.

Dream-heist key visual of Nahida on a mossy ledge under a Sumeru walkway folded back over her headDream-heist key visual of Homelander hovering in an upside-down Vought atrium, flags hanging downwardDream-heist key visual of Starlight in a mirrored ballroom, her reflection one step out of timeDream-heist key visual of Venti floating at a folded Mondstadt fountain, pavement climbing the right side
✦Dream heist
Draw a dream-heist key visual of Arthur in the rotating hotel corridor
Dream-heist key visual of Arthur braced on a red carpet that has become a wall in the rotating hotel corridor
Arthur · Inception — The page's own prompt, run in the Inception world: the carpet he walked on is now the wall under his shoes, and the elevator bank hangs sideways.
Dream-heist key visual of Ariadne on a Paris boulevard that folds into a vertical wall above her in the rain
Ariadne · Inception — The fold is the poster: cobbles stay cobbles where the boulevard turns vertical, and the rain meets itself along the hinge.
Dream-heist key visual of Cobb holding a spinning top on a rainy street with a mirrored city overhead
Dom Cobb · Inception — A mirrored downtown: the street above is the street below, and the cabs hang from the reflected asphalt.
Dream-heist key visual of Zhongli in a tilted Liyue vault with floating lanterns and a standing column of water
Zhongli · Genshin Impact — Liyue's vault tipped on its side — lanterns still hanging, a column of water standing where a basin should be.
Dream-heist key visual of Raiden Shogun walking a shrine corridor rolled ninety degrees, rain crossing the frame
Raiden Shogun · Genshin Impact — A tatami corridor rolled ninety degrees: the screens become her floor, a bronze lantern juts from the ceiling, rain runs across the frame.
Dream-heist key visual of Gojo at a Shibuya crossing whose far half rises as a vertical wall of signs
Satoru Gojo · Jujutsu Kaisen — Shibuya folded at the crossing — signage, crosswalk paint and cars running up the vertical half of the street.
Dream-heist key visual of Yuji running through a subway corridor that curves and spins around him
Yuji Itadori · Jujutsu Kaisen — A station corridor spinning on its own axis; the train is frozen on a vertical track and the paper has stopped falling.
Dream-heist key visual of Hori walking along a school landing turned into a vertical wall of lockers
Hori Kyouko · Horimiya — No magic trick needed: a school landing turns into a wall of lockers and she keeps walking, lantern in hand.
Dream-heist key visual of Miyamura sitting mid-air in a classroom with desks floating and the city on its side
Miyamura Izumi · Horimiya — Gravity switched off in an empty classroom — desks in orbit, the city outside lying on its side.

What Studio adds to GPT Image 2

GPT Image 2 draws the frame. The world keeps the next one in the same film: it holds the dreamer, the room, the palette and the rule that a floor can become a wall, so the second key visual belongs to the first. The image lands as a work with its own page, next to the characters it came from, instead of a file in your downloads.

  • A crew that stays on-model — every dreamer is a material with a cover image, and the next prompt uses that face as its reference.
  • Rooms you can point at — the Rotating Hotel stays that hotel — brass, carpet, a hallway that turns — so the landmark keeps the frame honest.
  • Frame to work in one run — the key visual publishes beside its world with a URL for a poster wall, a store page or a trailer card.

Questions before you draw

What do I get — one image or a set?

One wide key visual per run, saved as a work in the world. For a crew, run it again with the next name: the world keeps the palette and the rooms, so the frames sit together.

Do I have to write the GPT Image 2 prompt myself?

No. The button carries the prompt: frame size, the one impossible operation, the landmark, the accent colour, the quiet band at the top. You change the dreamer and the room, not the recipe.

How is this different from asking for surreal city art?

Surreal city art gives you a texture. This asks for one legible trick — a street folded at a hinge, a corridor rolled ninety degrees — and for the person who still has gravity, so the frame explains the heist instead of decorating it.

Can I use my own character instead of a film cast?

Yes. Add a character sheet and the prompt matches the face, hair and costume. Without one it uses that character's own cover in the world.

Why is the top of the frame empty?

So a title can sit there. No text is generated; the band stays calm and dark, and you set the type yourself.

Can I get a vertical poster or a square icon?

Ask for the shape you need: 16:10 for a key visual, 2:3 for a film poster, 3:4 for a store tile, 1:1 for an icon. The brief takes the size.

What if the fold comes out as a mess?

Name the hinge and the landmark. A fold with nothing to hold it — no awning, no crosswalk, no elevator bank — reads as a tilted camera. One reference per dreamer, one operation per frame.

What if I don't have a world yet?

Start from one sentence. Studio builds the cast and the rooms from it, then the first key visual; every later frame inherits them.

Draw the first frame of your dream heist.

One dreamer, one impossible room, one wide key visual — published next to the cast it belongs to.

×