AI 照片重绘并排对比
生成一张方形对比图，将原始的休闲 UGC 风格照片与更清晰的 GPT Image 2.5 风格重绘图进行对比。
gpt-image-2 · 产品营销 · 摄影 · 文本 / 排版 · 极简主义 · 社交媒体帖子 · 网红 / 模特
中文提示词
目标：制作一张方形社交媒体对比图，展示照片重绘质量测试，包含两个并排的图像面板和醒目的标签。 画布：1:1 方形图像，黑色背景，带有微妙的深灰色网格图案。顶部保留黑色页眉区域，底部保留黑色页脚区域。 布局：两个大型垂直照片面板填充中间区域，中间由一条垂直线分隔。左侧面板上方居中显示巨大的加粗白色大写文本 {argument name="left header text" default="ORIGINAL"}。右侧面板上方居中显示巨大的加粗白色大写文本 {argument name="right header text" default="GPT IMAGE 2.5"}。在底部中心位置，放置白色小写品牌文本 {argument name="footer brand text" default="maxfusion"}，其中“f”采用略微风格化的斜体外观。 面板内容：两个面板展示相同的场景概念：一位年轻女性在明亮的客厅里，手持一个微型黑色麦克风靠近嘴边，站在一个超大纯白色长方形盒子后方或手持该盒子。总计包含 2 个人物，每个面板 1 个；总计包含 2 个大型白色盒子，每个面板 1 个；总计包含 2 个小型手持黑色麦克风，每个面板 1 个；总计包含 3 个可见文本标签：“ORIGINAL”、“GPT IMAGE 2.5”和“maxfusion”。 左侧面板细节：低分辨率、略显模糊的休闲 UGC 手机照片外观。一位深棕色头发扎起的年轻女性，戴着超大圆形黑框眼镜，身穿白色短袖 T 恤和浅灰色运动裤，站在室内。她微微抬头看向右侧，带着俏皮的微笑，同时将小麦克风靠近嘴唇。一个巨大的白色长方形盒子占据了前景，底部被裁剪，几乎横跨整个宽度。背景包括中性色调的公寓内部，有吸顶灯/玻璃吊灯、浅色墙壁、门廊、模糊的装饰和柔和的室内光线。 右侧面板细节：同一瞬间更清晰、更精致的 AI 生成重绘图。女性形象更清晰、更逼真，留着 {argument name="hair color" default="dark brown"} 的头发，扎成马尾并系着黑色蝴蝶结，戴着超大黑框眼镜，身穿白色 T 恤和宽松灰色运动裤。她微笑着看向右侧，同时手持麦克风。超大白色长方形盒子横在她的躯干前，透视角度清晰，边缘锐利，带有浅灰色阴影。背景是明亮的现代客厅，配有沙发、地毯、木地板、咖啡桌、窗帘、窗户光线、盆栽植物、墙面艺术画、右侧的电视、编织篮和头顶上方清晰的玻璃球吊灯。 视觉风格：照片级逼真的对比拼贴画，简洁的编辑布局，黑色网格科技演示美学，高对比度白色排版，真实的室内光线，除指定的页脚文本外，不含额外的徽标或水印。
原始提示词
Goal: Create a square social-media comparison graphic showing a photo recreation quality test, with two side-by-side image panels and bold labels. Canvas: 1:1 square image, black background with a subtle dark gray grid pattern. Leave a top black header band and a bottom black footer band. Layout: Two large vertical photo panels fill the middle, separated by a straight vertical split. The left panel is labeled {argument name="left header text" default="ORIGINAL"} in huge bold white all-caps text centered above it. The right panel is labeled {argument name="right header text" default="GPT IMAGE 2.5"} in huge bold white all-caps text centered above it. At the bottom center, place the white lowercase brand text {argument name="footer brand text" default="maxfusion"}, with a slightly stylized italic-looking “f”. Panel content: The two panels show the same scene concept: a young woman in a bright living room holding a tiny black microphone near her mouth while standing behind or holding an oversized plain white rectangular box. Count exactly 2 human figures total, one in each panel; count exactly 2 large white boxes total, one in each panel; count exactly 2 small black handheld microphones total, one in each panel; count exactly 3 visible text labels total: “ORIGINAL”, “GPT IMAGE 2.5”, and “maxfusion”. Left panel details: Low-resolution, slightly blurry casual UGC phone-photo look. A young woman with dark brown hair pulled back, oversized round black glasses, a white short-sleeve T-shirt, and light gray sweatpants stands indoors. She looks slightly upward and to the right with a playful smile while holding the small microphone near her lips. A very large white rectangular box dominates the foreground, cropped at the bottom and extending across almost the full width. Background includes a neutral apartment interior with a ceiling light/glass pendant, pale walls, a doorway, faint decor, and soft indoor lighting. Right panel details: Sharper, more polished AI-generated recreation of the same moment. The woman is clearer and more realistic, with {argument name="hair color" default="dark brown"} hair tied back in a ponytail with a black bow, oversized black glasses, a white T-shirt, and loose gray sweatpants. She smiles and looks to the right while holding the microphone. The oversized white rectangular box is held across her torso and angled in perspective, with crisp edges and light gray shadows. Background is a bright modern living room with a sofa, rug, wooden floor, coffee table, curtains, window light, potted plants, framed wall art, a TV on the right, a woven basket, and a clear glass globe pendant lamp overhead. Visual style: Photorealistic comparison collage, clean editorial layout, black grid tech-demo aesthetic, high contrast white typography, realistic indoor lighting, no extra logos or watermarks beyond the specified footer text.