Tokyo Giant Dessert Miniature World

Model N/AMiniature WorldFoodVFXCreativeDrone AerialTokyo9s
A silver fork plunges into a miniature Tokyo street and lifts up a strawberry cake, while a caramel waterfall pours across the road. The scale mismatch between city and dessert is pushed to the extreme — a lot packed into 9 seconds.

PROMPT · Video Prompt

Create with AI
Create a smooth 9-second surreal animation that merges Tokyo with a giant dessert: a huge silver fork plunges into a miniature Tokyo street, lifting up a slice of strawberry cream cake topped with fresh cream, strawberries, and a red glaze; the shot cuts to street level, where a massive slab of chocolate is embedded in the pavement, thick caramel pouring over it as pedestrians walk through the scene; a bustling avenue lined with skyscrapers is shown, their rooftops decorated with strawberries and fresh cream; finally, the camera pulls back to reveal that the entire city district actually sits atop a giant decorative cake on a white plate, and a huge fork slices off a piece of the "city cake" and lifts it up, before the image seamlessly transitions back into an aerial overview of Tokyo.

✍️ Editor’s Notes

This one is only 219 characters long, making it one of the shortest formal prompts in the whole batch, and it makes total sense that need_expanded is flagged true — it only sketches out four big images (a fork scooping up cake, a chocolate bar embedded in the road, strawberries decorating avenue rooftops, the reveal that the whole district is actually a cake), with absolutely no timeline, camera-movement approach, or transition technique given anywhere. That's really just a story premise, not an executable shot script. This kind of ‘concept-first, execution-left-blank’ writing works fine as a creative seed, but handed straight to a video model, it's very likely to split the 9 seconds evenly across all four images and rush every single one — it's missing a momentum-trigger rule like the one in ‘a man crossing through ten environments’ to guide the transitions, and it never specifies shot sizes or push/pull camera moves either. That's exactly where the value of expanding it comes in: turning these four images into an executable script with a proper time allocation and camera design.

Want tighter control? Try this expanded version:
9-second vertical surreal animation, one continuous unbroken take. 0-2s: Overhead view of a miniature Tokyo streetscape, a giant silver fork descends slowly from the upper right of the frame and plunges into the block, scooping up an entire strawberry cake covered in fresh cream, sliced strawberries, and a red glaze; the frosting glistens with a moist sheen in the morning light as the camera tilts up slightly to follow the fork. 2-4s: Hard cut to street level, eye height, a whole giant chocolate bar is embedded in the asphalt road surface, thick caramel slowly pours out and covers the curb, pedestrians walk around it as if it's nothing unusual, character proportions stay realistic. 4-6s: Camera rises to a 45-degree overhead angle, revealing a bustling avenue lined with skyscrapers, the rooftops of the glass-curtain-wall towers piled high with strawberry and fresh-cream decorations. 6-9s: Camera rapidly pulls straight back and up, revealing that the entire district is actually sitting on a white porcelain plate as a giant decorative cake; a giant fork enters from off-frame, cuts off a slice of the “city cake,” and lifts it away from the plate, then seamlessly cuts back to the real overhead view of Tokyo, forming a looping structure that echoes its own opening and closing. Colors stay bright and saturated throughout, the cream and frosting textures are rendered with fine, realistic detail, no text, watermarks, or live-action close-ups appear anywhere.
Create with AI

Related Works

׋›
↑