Realism-to-Flat Style Spectrum: Illustration Dimensionality-Reduction Experiment with a Single Reference Image

Starting from the same realistic portrait reference image, the reference-image editing capability of qwen-image-3.0 is used to push the picture step by step toward flat illustration: the subject's pose, composition, and clothing structure are preserved, while only the light-and-shadow layers, number of tonal levels, and brushstroke texture are reduced, yielding an edited final piece in an intermediate "semi-realistic—flat" state. The schema allows submitting 1–3 reference images at a time; this work uses only a single one, making it easy for developers to compare under the minimal controllable variable of "reference image unchanged, only style intensity varied." It is suitable as a style-intensity calibration, a preliminary sample for illustration-oriented workflows, and an evaluation of the model's degree of style transfer while preserving the key details of the original image.

Style style-transfer · Illustration · Editing

Model Qwen Image 3.0

Reference inputs

image_urls

Try it yourself

Prompt

Try this model

An edited illustration generated based on a single photo reference. Strictly preserve the subject's pose, silhouette, facial structure, clothing folds, and composition from the reference image; progressively flatten the rendering: compress the value range into five to six clean tonal levels, replace photographic gradients with hard-edged or slightly feathered color blocks, simplify skin shadows into two-tone planes plus a single highlight shape, turn hair strands into solid ribbon-like masses supplemented with a few decisive brushstrokes, and reconstruct fabric textures as flat color blocks with sparse line accents. Retain restrained semi-realistic volume in the face and hands, so that the result stops in the middle of the realism-to-flat spectrum rather than becoming fully posterized. Palette: low-saturation warm ochre, gray terracotta, sage green, and off-white, all drawn from the reference image and reduced to a limited set of color groups. Clean 4:3 editorial composition, no outline-only cartoon feel, and no props or background details added except for a soft flat background.

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