
John-Ivar Eidsmo
ABOUT
A self-initiated AI series turning some of the world’s most recognizable cities into miniature worlds. Each city was rebuilt as a hyper-realistic diorama, compressing its architecture, landscape and cultural details into a single tiny environment. Monaco became a cliffside model of casinos, yachts and the Grand Prix circuit; other pieces explored cities including Athens, Milan, Tangier, Rabat, Bucharest and Ankara. The project started as an experiment in whether generative AI could do more than create individual impressive images — whether it could sustain a recognizable visual concept across an entire series. The result became a repeatable format built around one simple idea: the world’s most fascinating places, made miniature.
SELECTED FRAMES — 6 OF 8
PROCESS
The challenge was never simply generating a miniature city. It was deciding what makes a city recognizable when you only have one image to work with. I started each piece by researching the place and reducing it to a handful of visual cues: landmark architecture, landscape, materials, vegetation, infrastructure and small details of everyday life. Those elements were then recomposed onto a compact, isometric base rather than trying to reproduce the real geography literally. The image generation was highly iterative. Too many landmarks made the result feel like a collage; too few made the city generic. I refined composition, scale, lighting and detail until each miniature could be understood almost immediately, while still rewarding a closer look. Once the still image worked, I treated it as the master frame for animation. Selected scenes were brought into AI video tools to introduce camera movement, traffic, water, people and environmental motion, before being edited into short-form videos. Over time, the process became a small visual system of its own: the same perspective, physical logic and miniature aesthetic, but a completely different world inside every frame.


