Sunburst E-commerce Product Photo Scene Reconstruction Final Image

Starting from an ordinary product photo, this demonstrates how the gpt-image-2.6 Sunburst (image-to-image) pipeline is used for e-commerce product photo scene replacement. The source image only provides the product shape and approximate viewpoint; the prompt is responsible for rebuilding lighting, background, and staging props, and the output is a finished-grade image at a uniform size. In terms of composition, the work includes one product source image (as image_urls[0]) and one editing instruction: after enabling high render quality, high-complexity planning, 2K resolution, and a 4:3 aspect ratio in base, the model redraws the background and lighting while preserving the product's outline and proportions. The Sunburst version focuses on detail and editing consistency, making it suitable for repeatedly iterating different scenes on the same product (studio white background, home countertop, outdoor flat surface, etc.). Use cases: scene-based drafts for e-commerce hero images and detail pages, batch variants for advertising assets, and visual direction comparison before listing. Note that the model does not guarantee accurate reproduction of precise logos, label text, and materials; if the final image is to be used as a finished-grade product image, the official recommendation is to composite the confirmed edited result with the original image rather than relying solely on the prompt.