Как это сделано
Pocket Worlds
Об этом повторе
Pocket Worlds asks for four tabletop-model scenes that look like handcrafted dioramas photographed with a tilt-shift lens. The shared art direction is long and concrete: polymer clay, balsa wood, flock grass and resin water on a small wooden base; a soft diffused softbox light with warm low key; a thin plane of focus with soft bokeh around it; a slightly high 45-degree angle; and an amber, rust and sage palette. Each scene also needs tiny concrete props and one oversized real object as a scale cue. All frames are 3:2, generated independently with GPT Image 2 and no source image.
The four scenes are a noodle stall at night with a giant porcelain spoon, an autumn bookshop with matchbox books and a giant brass coin, a harbor pier with a red boat and a giant seashell, and a snowy train stop with a giant pinecone. Sizes such as a 3mm chef and a 2mm cat are given in the brief, which pushes the miniature scale.
The agent called the generator once for four images, each returned at 1536x1024. It says each frame got its own full prompt built from the shared direction. Its summary describes the scale objects and the lighting, and says the amber, rust and sage palette and shallow focus hold across the series. No edit pass was run, and the agent did not report checking whether the giant objects came out as requested, so confirm on the page.
Tip: asking for one oversized real-world object per scene is a simple way to make a miniature read as a miniature, and listing the same material and lens language in every frame keeps four different scenes in one set.
Запрос
Create a 4-frame miniature-art showcase called Pocket Worlds - four distinct handcrafted-miniature diorama scenes unified by ONE look: macro tilt-shift model photography that makes each scene read as a tiny tabletop model. Shared art direction across all four: built from polymer clay, balsa wood, flock grass and epoxy-resin water on a small wooden base; soft diffused softbox light with a warm low key and crisp tiny shadows; razor-thin plane of focus on the hero with foreground and background melting to soft bokeh; slight high 45-degree top-down angle; cohesive warm amber / rust / sage palette; photoreal model-photography finish. Pack each scene with tiny concrete props and put ONE oversized real-world object in each as a scale tell. Use gpt-image-2, 3:2 for every frame, generate each frame independently (no source_image; they share the look through this description). Frame 1 Ramen Alley at night: a thumb-sized noodle stall with tiny glowing paper lanterns, a 3mm chef, hand-rolled cotton steam; a giant porcelain spoon leans in for scale. Frame 2 Autumn Bookshop: a tiny corner bookshop with stacked matchbox books, a moss-flock maple shedding paper leaves, a 2mm cat on the step; a giant brass coin for scale. Frame 3 Little Harbor: a tiny fishing pier with a palm-sized red boat on glassy epoxy-resin water, miniature crates and coiled rope; a giant seashell for scale. Frame 4 Snowy Train Stop: a tiny platform under cotton-wool snow, a thumb-sized red train, 3mm lamp posts; a giant pinecone for scale. Generate all four.
Что сделал агент
I'll create four independent miniature diorama scenes in the Pocket Worlds series. Each frame gets its own full 5-axis prompt, unified by the shared macro tilt-shift look, warm amber/rust/sage palette, and photoreal model-photography finish.
Создано 4 изображения · GPT Image 2
All four Pocket Worlds frames are ready, each rendered in warm macro tilt-shift model photography with the shared amber/rust/sage palette and razor-thin depth of field.
Frame 1 — Ramen Alley at Night: A glowing thumb-sized noodle stall with paper lanterns, a 3mm chef, and cotton steam; the giant porcelain spoon leans in to shatter the scale illusion.
Frame 2 — Autumn Bookshop: A cozy corner shop with matchbox books, a moss-flock maple shedding paper leaves, a 2mm cat on the step, and a brass coin resting nearby as the scale tell.
Frame 3 — Little Harbor: A palm-sized red boat on glassy epoxy-resin water, miniature crates and rope on the pier, with a giant seashell anchoring the scale.
Frame 4 — Snowy Train Stop: A red train under cotton-wool snow, glowing 3mm lamp posts, and a giant pinecone dusted with flakes beside the tracks.
The softbox low-key light, creamy bokeh, and 45-degree top-down angle hold the series together. Want me to refine any frame's focus or swap a scale object?
Результаты



