GPT Image 2.5 Sunburst 提示词 - 玻璃虚影后的抓拍角色扮演者
- 画面比例
- auto
- 画质
- 1K
提示词
explicitly an adult character, high-quality live-action COS character photography with a candid, secretly captured snapshot aesthetic. Preserve the character’s signature hairstyle, hair color, recognizable costume silhouette, iconic color palette, and accessories.
The character is naturally going about her day in an ordinary public environment, with no formal posing. The camera feels as if it is discreetly observing and photographing her from nearby, partially hidden behind foreground obstacles. Parts of the frame may be obstructed by door frames, pillars, shelves, glass panels, table edges, or other environmental elements.
Place the character slightly off-center. The composition should feel somewhat rushed and imperfect rather than carefully balanced, allowing subtle cropping, slight camera tilt, and asymmetry. Automatically choose a focal length between 35mm and 200mm depending on the scene: use a casual smartphone snapshot feeling at close range, and telephoto compression for distant candid shots.
Shallow depth of field, with clearly blurred foreground elements while keeping the character’s face recognizable.
The character may naturally be looking down at her phone, adjusting her hair, drinking a beverage, waiting for transportation, looking into a shop window, walking, or briefly stopping. Automatically choose the most natural action based on her personality. In a small number of shots, she suddenly glances sideways or turns back toward the camera, creating the fleeting moment when she realizes someone is photographing her.
Realistic ambient lighting, authentic skin texture and fabric detail, natural human proportions, subtle smartphone sensor noise, slight motion blur, reflections on glass, or minor auto-exposure imperfections.
Spontaneous candid photography, telephoto candid shot, foreground obstruction, imperfect framing, realistic secretly captured snapshot aesthetic, while clearly remaining a fictional and staged photography scenario.
使用这条图片提示词
- 01
从提示词开始
复制提示词,或点击生成按钮在生成器中打开。
- 02
加入你的想法
调整主体、场景或细节,按需添加参考素材。
- 03
生成前确认
检查模型与输出设置,每次生成的结果可能有所不同。







































