Realistic to Flat: A Style Spectrum — Flare Image-to-Image Illustration Translation

Using a single realistic still-life photograph as the sole reference input, the gpt-image-2.5-flare image editing capability is used to restate the same composition across three levels of abstraction — "realistic — semi-flat — flat vector" — as illustrations. The creative idea is to verify the controllability of image-to-image style transfer: the reference image supplies composition, subject, and lighting relationships, while the prompt describes only material treatment, color-block segmentation, and line language, thereby observing the model's judgment about "what to keep and what to discard." The base uses a 16:9 landscape composition and a 2K resolution tier to avoid the mutually exclusive restriction between transparent background and the resolution enumeration; background is left at its default without being explicitly declared, switching to transparent only when asset compositing is needed. It is suitable for style exploration and rapid style-draft validation in scenarios such as editorial illustration, brand visual extension, and content imagery.

Style Illustration · style-transfer · Editing

Model GPT Image 2.5

Reference inputs

image_urls

Try it yourself

Prompt

Try this model

An editorial-style still-life illustration derived from a reference photograph, preserving the original composition, object placement, and light direction while translating the rendering into a mid-century modern flat illustration style: matte gouache-like flat color blocks, reduced internal detail, two-tone light and shadow replacing gradients, subtle paper texture, a restrained palette of warm terracotta, deep olive green, and off-white, crisp yet slightly imperfect edge shapes, major forms without outlines, and extremely thin accent lines used only where forms need to be distinguished. Retain the recognizable silhouettes and shadow logic of the source image; keep the composition simple, leaving quiet negative space in the upper right. Wide 16:9 composition, poster-like balance, no text, no extraneous objects.

Copy the full prompt and adapt it for your own generation.