E-commerce Product Still Life: Skincare Bottle Photography Prompt Example

Using the text-to-image capability of qwen-image-3.0-pro, this demonstrates a set of "e-commerce main image"-grade product photography prompt writing: organize the prompt in five layers—subject description, material and surface finish, lighting structure, lens and depth of field, background and props—translating the controllable variables in a photography studio into language the model can read. The example subject is a frosted glass skincare bottle, because it simultaneously tests refractive transparency, soft gradient backgrounds, and highlight control—the combination most prone to failure in product images. The prompt explicitly specifies an 85mm macro feel, a large softbox key light from the upper left, a reflector fill on the right, and a contact shadow on the ground, preventing the model from producing a floating look or sourceless ambient light. On the base side, aspect_ratio is fixed at 4:3, resolution at 2K, output_format as png, with a constraining negative_prompt (remove watermarks, remove extra text, remove distorted labels), and seed fixed for reproducibility under the same parameters. The schema supports prompt_extend; this example disables it so the prompt takes effect verbatim, making it easier for developers to attribute results by comparing prompt and output image. Applicable scenarios: e-commerce main images and detail page headers, brand still-life posters, concept validation for materials and packaging. Note that the model's official documentation describes the prompt context as approximately 800 tokens; in actual use, it is recommended to streamline descriptions according to the layers above to avoid stacking adjectives.