Wan2.7-Image Portrait Dual-Style Demo: Realism and Artistic

This piece uses the same prompt structure to demonstrate in a comparative way two directions of wan2.7-image-standard on the portrait subject: on one end is a highly realistic portrait emphasizing skin texture, natural light, and environmental storytelling; on the other end is an artistic character illustration emphasizing brushstrokes, color blocks, and decorative qualities. Both prompts share the layered writing approach of "subject — pose — lighting — lens — texture," making it easy for developers to migrate the structure directly to scenarios such as marketing visuals, serialized character imagery, and concept exploration. In terms of composition, only the text parameters of the author's pipeline are used: the prompt is written into the author's input via a top-level field, and base retains only non-media fields such as aspect ratio, resolution, and seed; the reference image path is not used (image_urls is optional, and once uploaded, the reference image determines the aspect ratio), so aspect_ratio and resolution both take effect under text-to-image semantics. The chosen 3:4 vertical composition fits the portrait subject better, and the 2K tier is used to observe the restoration of detail levels. According to research on the model family, wan2.7-image-standard supports Chinese and English prompts, a maximum 2K resolution for the standard text-to-image version, and covers both generation and editing within the same API; portrait customization, Hex palette guidance, and long-text rendering are capabilities of this series, and this entry only demonstrates the shortest path of text-to-image. It is suitable as a prompt template reference: the realistic version can serve as a baseline for product figures and brand imagery, while the artistic version can serve as a starting point for a stylistically unified series of images.