E-commerce Product Photography: gpt-image-2 Image Prompt Practice

An e-commerce-grade product image practice using gpt-image-2's text-to-image pipeline: build a commercial product photography scene of a skincare serum bottle entirely from text prompts, testing the model's performance on materials (frosted glass and metal cap), studio lighting, label typography, and compositional order. Prompt organization approach: first lock in the industry and subject, then describe materials and surface treatment, followed by the lighting setup (key light plus rim light plus softbox), background and surface materials, camera angle and focal-length feel, and finally constrain color palette, negative space, and image usage. The text portion separately declares the label copy and layout requirements, making it easy to check the text rendering results. base only carries size and the prompt automatically written from the top-level prompt; there is no reference image input, and the entire image is generated in a single text generation pass. Suitable for developers who need to quickly produce campaign-ready product key visuals, iterate across multiple style versions, or evaluate the model's text rendering and commercial quality capabilities.