Qwen-Image-2.0 Portrait Prompt Reference: Two Approaches — Photorealistic and Artistic

Using Qwen-Image-2.0's text-to-image capability, this is a side-by-side prompt demonstration focused on portraits: two sets of prompts with different stylistic directions are given under the same model, to observe how its output differs between photorealistic portraits and artistic portraits. The photorealistic approach emphasizes natural light, skin texture, and lens language (such as 85mm with shallow depth of field); the artistic approach emphasizes painterly light and color and a stylized look. The piece uses a single 2:3 vertical composition and serves as a reference baseline for developers writing portrait prompts, and is also suited to prompt tuning for generation scenarios such as avatars, character posters, and character design. In the schema, prompt is required; this piece additionally sets negative_prompt and a fixed seed to improve reproducibility; prompt_extend stays enabled by default, making it easy to compare before and after automatic optimization. Qwen-Image-2.0 is provided by Alibaba. Public materials indicate that it supports complex instructions up to 1,000 tokens and output resolutions up to 2K, and performs notably well on Chinese text rendering; the exact capability boundaries are subject to the official documentation.