GPT Image 2 对比 Midjourney v8.1
分享链接 · https://openimages.ajiang.me/p/2056373684839854124/


camera_lens · 85mm-portraitcamera_lens · shallow-depth-of-fieldcolor_palette · warm-tonescomposition · eye-levelcomposition · medium-shothuman_attributes · makeup
提示词(原文,逐字保留)
GPT Image 2 vs Midjourney v8.1
榻榻米晚间 / Tatami Evening Room
Prompt 👇👇👇
▸ GPT Image 2
A medium-shot authentic daily life photograph, natural candid moment, of an East Asian young woman standing in a low-ceilinged tatami room near dusk. 85mm lens, waist-up framing. The room behind her: worn tatami, a dark clay vase on the floor with a single dried branch, shoji screens glowing faintly with the last of the day's light.
East Asian young woman in her early 20s. Almond-shaped eyes with natural double eyelids, gaze soft and slightly lowered. Straight refined nose. Skin tone fair to light beige, warm with the last evening light. Skin subsurface scattering visible under directional light, specular micro-highlights on cheekbones and nose ridge, fine foundation powder grain perceptible.
She wears a dusty rose silk qipao with hibiscus embroidery in threads of faded burgundy and pale sage — the pattern worn, the rose pigment deepened slightly with age into something closer to old tea-stain pink. Mandarin collar, short sleeves. She stands facing the glowing shoji screens, arms at her sides, entirely still. The room requires no performance from her. Two or three stray hairs displaced by faint air movement, natural unplanned imperfection, not geometrically symmetrical.
The shoji screens glow amber-rose with the last light outside, backlit and softly diffused, casting the room into warm shadow. Her face receives only the reflected warmth — no direct light source visible. The dusty rose qipao and the amber shoji occupy the same temperature of color, the room folding her into itself. Subtle ISO 400 film grain in shadow areas, photographic noise texture not CG render smoothness. Aspect ratio 9:16. No watermark, no text overlay, not cartoon, not digitally painted, not illustration, not anime.
▸ MJ v8.1
East Asian young woman early 20s standing in low-ceilinged tatami room near dusk, waist-up 85mm shallow depth of field, worn tatami floor and glowing shoji screens behind her softly out of focus, almond eyes natural double eyelids gaze softly lowered, straight refined nose, fair light-beige skin warmed by dusk glow, black hair in loose low bun, wearing dusty rose aged silk qipao with faded hibiscus embroidery in dusty burgundy and pale sage mandarin collar short sleeves, arms at sides facing glowing shoji entirely still, expression absent and interior, amber-rose backlit shoji glow warm reflected light on face no direct source, Wabi-Sabi tatami room dusk atmosphere, dusty rose and amber palette within same temperature, Wong Kar-wai color, skin subsurface scattering specular micro-highlights, two or three stray hairs displaced by faint air movement, subtle ISO 400 film grain in shadow areas --ar 9:16 --style raw --s 100
中文译文
GPT Image 2 对比 Midjourney v8.1
榻榻米晚间 / 榻榻米暮色房间
提示词 👇👇👇
▸ GPT Image 2
一张中景的真实日常生活照片,自然抓拍的瞬间:一位东亚年轻女子在黄昏时分站在一间低天花板的榻榻米房间里。85mm 镜头,腰部以上构图。她身后的房间:磨损的榻榻米地面、地板上摆放着一只深色陶瓶、瓶中插着一枝干枯的树枝、障子门在白昼最后的光线中微微发亮。
20 岁出头的东亚年轻女子。杏仁状的眼睛、自然的双眼皮,目光柔和而微微低垂。笔挺秀气的鼻子。肤色白皙至浅米色,被傍晚最后的光线映得温暖。在定向光下可见皮肤次表面散射,颧骨与鼻梁上有镜面微高光,能感知到细腻的粉底颗粒感。
她身穿一件藕粉色真丝旗袍,绣有扶桑花图案,绣线为褪色的酒红与浅鼠尾草绿——图案已显陈旧,玫瑰色颜料随岁月略有加深,更接近老茶渍的粉。立领,短袖。她面向发光的障子门站立,双臂垂于身侧,完全静止。房间不需要她做任何表演。两三缕散发被微弱的气流吹动,自然而无计划的瑕疵,并非几何对称。
障子门被屋外最后的光线映成琥珀玫瑰色,背光柔和漫射,将房间投入温暖的阴影之中。她的面部只接收反射来的暖光——画面中看不到任何直接光源。藕粉色旗袍与琥珀色障子处于同一色温区间,房间将她融入自身。阴影区域带有细微的 ISO 400 胶片颗粒,呈现摄影噪点质感而非 CG 渲染的平滑。画幅比例 9:16。无水印,无文字叠加,非卡通,非数字绘画,非插画,非动漫。
▸ MJ v8.1
20 岁出头的东亚年轻女子,黄昏时分站在低天花板的榻榻米房间里,腰部以上构图 85mm 镜头浅景深,身后磨损的榻榻米地面与发光的障子门柔和虚化,杏仁眼自然的双眼皮目光柔和低垂,笔挺秀气的鼻子,白皙浅米色肌肤被暮色辉光映暖,黑色头发松松挽成低髻,身穿藕粉色陈年真丝旗袍,绣有褪色的扶桑花图案,绣线为灰酒红与浅鼠尾草绿,立领短袖,双臂垂于身侧面向发光的障子完全静止,神情空然而内敛,琥珀玫瑰色背光的障子泛光,脸部为温暖反射光无直接光源,侘寂榻榻米房间黄昏氛围,藕粉与琥珀色调处于同一色温区间,王家卫色调,皮肤次表面散射镜面微高光,两三缕散发被微弱气流吹动,阴影区域细微的 ISO 400 胶片颗粒 --ar 9:16 --style raw --s 100
来源与署名
原文由 @ToroJushiAi 发布在 X:查看原推文。 本页逐字保留原文并提供机器翻译的中文解读;版权归原作者所有。
同工具的更多提示词
- 修改图像:以已上传图像中的主体作为基础主体框架,保留面部细节和身体轮廓造型。忽略任何其他不相关的元素。保留主体的面部特征和姿势,但完全按照以下提示语进行视觉重…@ChillaiKalan__
- 受顶级Vogue、迪奥(Dior)、Sabyasachi、艾莉・萨博(Elie Saab)以及高端平面广告启发的奢华高级定制婚纱时尚大片,核心重点是突出一位惊…@obsidian_fr1
- 创建一个 16:9 比例的角色概念图集,主题为史诗电影级科幻奇幻风格的星际漫游者。 以干净的概念美术版面,展示同一角色的 4 个不同视角: 对称的全身正面视图…@0kncn
- 多角色场景过去总是把一切搞砸。 三个人围坐在一张桌子旁 —— 模型总是会弄丢某个人。位置错了。视线方向错了。直接消失。 新的节拍表解决了这个问题。 一条 GP…@Iancu_ai
- 我在 ChatGPT Image 2 中创建并测试了一套品牌人格拆解(BRAND PERSONALITY BREAKDOWN)。它能把任何品牌转化为一个有真实…@aatqa_hussain
- 动漫、卡通、插画、CGI、3D渲染、塑料皮肤、蜡像脸、过度磨皮皮肤、厚重美颜滤镜、假脸、换脸效果、扭曲的面部特征、被改变的身份、不自然的微笑、僵硬姿势、夸张表…@Shinning1010