HiAPI
Showcases
Realistic and Artistic Dual-Style Comparison of the Same Portrait

Realistic and Artistic Dual-Style Comparison of the Same Portrait

Using a realistic portrait as the sole reference image, this demonstrates the image-to-image pipeline of qwen-image-3.0-pro for "preserve identity, change only style" editing: the same face is repainted into an oil-paint-textured artistic portrait driven by the prompt, while keeping the facial feature structure, hairstyle silhouette, and composition position unchanged. The value of this work lies in the comparative prompt writing—the positive prompt focuses on style vocabulary (brushstrokes, color temperature, lighting), while the negative prompt suppresses style drift and identity changes; this reusable structure offers direct reference value for developers debugging portrait stylization workflows. It is suitable for avatar/portrait stylization, brand portrait asset variants, and as a baseline template for writing prompts for stylization tasks. The reference image itself is also generated by the pipeline, so the entire chain can be fully reproduced.

Style Portrait & Photography · style-transfer · Photo editing

Model Qwen Image 3.0

Reference inputs

  • image_urls

    image_urls

Prompt

Try this model
Use image-to-image style reshaping on a single portrait. Strictly preserve the subject's facial structure, hairstyle silhouette, and body pose from the reference image; keep the composition and background layout unchanged. Repaint the surface texture in an oil painting portrait style: visible directional brushstrokes, soft skin transitions, warm amber key light from the upper left, cool blue-gray fill light from the right, and a dark mottled background with fine canvas texture. Beneath the paint layer, the skin still retains realistic pores and hair strand details, and the eyes remain clearly discernible. Medium close-up, shallow depth of field, 3:4 vertical composition, soft earth tones, with no text or graphic elements.

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

Tags