25 个产品对比 & 审美分析 Prompt 精选：GPT Image 2 最被低估的实用玩法

以眼部特写图片为基础，生成3:4的四屏构图超写实眼部特写，四屏按春夏秋冬上下排序。

第一屏：眼眸中带着绽粉樱色的美瞳，睫毛缀满迷你春花，脸颊散落樱瓣与黄蕊小花，粉蝶萦绕眉眼，浅金发丝轻垂，下方簇簇樱花怒放，画面中央"SPRING"白色艺术字点缀，风格细腻唯美，光影柔和，色彩粉嫩治愈，下面用书法体写着春；

第二屏：眼眸中带着着清荷色的美瞳，睫毛饰以粉莲与绿荷，脸颊挂着晶莹水珠，粉瓣、绿荷点缀其间，蜻蜓轻绕，浅金发丝若隐若现，画面中央"Summer"白色艺术字凸显，光影通透流光感，色彩清透凉爽，下面用书法体写着夏；

第三屏：眼眸中带着金黄红相间的美瞳，睫毛饰以橙红枫叶，脸颊散落金红秋叶，橙蝶翩跹眉眼间，浅金发丝隐约可见，画面中央"AUTUMN"白色艺术字醒目，光影暖金流光，色彩浓郁温暖，下面用书法笔写着秋；

第四屏：眼眸中带着雪花蓝色的美瞳，睫毛覆满冰晶雪片，脸颊散落白色雪花与红色腊梅，银白蝴蝶翩跹眉眼，浅金发丝朦胧似雪，画面中央"WINTER"白色艺术字亮眼，光影冷冽蓝白流光，色彩清透纯净，下面用书法体写着冬。

整体呈现梦幻眼眸四季交替的唯美梦幻治愈画面，微调各屏的光影强度，让画面氛围感更浓郁。

Create a personal color analysis graphic using this portrait. Point out which season colour suits the subject best. Show side-by-side clothing color comparisons to highlight which colors suit the subject best. List out what texture/accessories/hairstyle suit the subject best. Make it visual-first, with short labels only and no paragraphs.

请根据我上传的人像照片，做一套个人形象分析图卡，包含发型、妆容、色彩和珠宝。要求：保留五官脸型肤色，不要过度修图，所有变化在同一张脸真实展示，风格干净高级。发型：长短、卷直、刘海，对比最适合/普通/不建议（显脸小、显老）妆容：眉眼鼻唇分析，标签（自然、提气色、柔和）色彩：不同颜色上身，对比推荐/普通/不适合（显白、显老）珠宝：珍珠、翡翠、红蓝宝、钻石、黄金，对比推荐/普通/不建议 整体：视觉为主，文字简短，4:5比例

Create a clean vertical fashion infographic from the uploaded portrait. Preserve the same face identity, hairstyle, body shape, and overall outfit style. Place the full-body character in the center in a relaxed T-pose, facing forward. Surround the character with realistic photo-style outfit breakdown elements, connected by thin arrows.

Include separate cutout sections for: head details, cardigan, sailor top, inner blouse, plaid skirt, bag, socks, loafers, color palette, styling notes, and fabric texture. Add 3–4 head close-ups from different angles at the top. Use short English handwritten-style labels and concise bullet points.

Visual style: soft pastel cream and blush-pink background, clean fashion board layout, elegant magazine-style composition, sweet preppy aesthetic, realistic fabric textures, delicate borders, small bow and heart doodles, airy and polished design. Keep the text short, readable, and fully in English. Original vertical aspect ratio.

Negative Prompt:

Chinese text, long text, messy layout, old parchment background, yellow aged paper, blurry details, distorted face, changed identity, extra limbs, bad hands, duplicated body, unrealistic fabric, cartoon style, anime style, 3D render, watermark, logo, unreadable typography, overcrowded design, harsh colors, low resolution.

9:16 竖向构图，单人女性艺术肖像，年轻东方女生，五官清秀，脸部线条柔和，皮肤自然通透，保留真实肌理，气质安静高级，带一点疏离感和故事感。
摄影棚风格与自然光融合，柔和侧光，面部有细腻高光，阴影轻柔，整体光线通透不刺眼，带轻微黑雾滤镜效果，微朦胧、微泛光、空气感强。
背景极简干净，奶油灰、米白、浅卡其或雾感暖灰色墙面，留有大面积负空间，整体画面简洁、有呼吸感。
模特坐在地面或低台上，一条腿自然弯曲，一条腿放松伸展，身体轻微前倾或侧倾，肩膀不对称，头部轻轻倾斜，动作自然松弛，不刻意摆拍。
表情平静克制，眼神柔和，略微疏离，带一点若有所思的情绪，嘴唇自然微张或轻闭，状态慵懒、安静、细腻。
发型为自然蓬松的长发，微凌乱碎发，发丝轻柔，有空气感和层次感，像刚整理过但保留自然随性感。
妆容为高级淡妆，韩系清透底妆，皮肤柔雾光泽，鼻梁与面颊有自然高光，眉形干净，眼妆淡雅但有神，睫毛纤长，唇色为低饱和玫瑰豆沙色或奶茶裸粉色。
服装为简约高级风：米白色紧身罗纹针织背心，外搭宽松白衬衫或柔软针织开衫，下装为高腰半裙或简约短裤，布料柔软贴合身形但不过分暴露，呈现自然身体线条与文艺感。
画面强调细腻质感、柔和色调、轻法式与韩系杂志感结合，真实摄影感，电影级肤色，细节丰富，层次分明，构图克制，高级审美，时尚 编辑人像，柔和的电影感人像，细腻的质感，超高细节，逼真，优雅，精致，高端时尚摄影，含蓄的性感，简洁的构图。

A side-by-side comparison graphic on a black background demonstrating a camera-angle change in the same restaurant scene. At the top, large white sans-serif text reads: "Show me the POV from someone standing behind the bar looking out over this crowded restaurant. Change NOTHING in the scene other than the pov". Below, place 2 stacked rectangular photos centered vertically: the top image labeled "Source" in large white text on the left, and the bottom image labeled "Output" in large white text on the left. The top photo shows a warmly lit, upscale, crowded restaurant interior seen from the dining room side, facing a tall back bar filled with many illuminated liquor bottles on wall-to-wall shelves, with bartenders and guests in front, amber lighting, globe pendant lights, wood ceiling, beige columns, and tightly packed seated diners in the foreground. The bottom photo shows the exact same restaurant, same crowd density, same warm lighting, same decor, same bar shelving, same globe pendant lights, and same overall composition elements, but now from the point of view of someone standing behind the bar and looking outward across the crowded restaurant; the foreground includes the bar counter with glassware, metal bar tools, bottles, and a point-of-sale screen visible at the lower left, while guests and staff fill the middle ground and the dining room extends into the background. Preserve the sense that only the camera position changed between the 2 images, with no other scene alterations.

Create a dramatic Japanese YouTube thumbnail in a futuristic neon cyberpunk style, 16:9 landscape. Use a dark tech-city background with faint skyscrapers, digital grid lines, glowing particles, and high-contrast blue, pink, and gold lighting. In the exact center, place a young woman from the waist up with long straight pastel blue hair, wearing a plain white short-sleeve T-shirt and a light pink skirt, posing thoughtfully with one hand near her chin and the other arm folded; anonymize her face with a soft rectangular blur. Across the very top, add huge distressed bold white Japanese headline text reading 主導権が揺れた, and directly below it add large bold yellow text reading {argument name="subheadline text" default="Nano Bananaから"}. On the left side, create a glowing blue hexagonal-framed panel titled Nano Banana with a smaller subtitle 画像生成. Inside that panel, include exactly 4 image tiles in a 2x2 grid: 1) a fantasy floating island landscape at sunset, 2) a sunlit forest path with tall trees, 3) a neon futuristic city street at night, 4) an outer-space planet scene with stars and a spacecraft. Beneath the left panel, add a blue glowing ribbon label reading かつては優位だった. On the right side, create a glowing magenta hexagonal-framed panel titled {argument name="right panel title" default="GPT Image 2"} with a smaller subtitle 実務で使える出力へ. Inside it, include exactly 4 example thumbnail cards in a 2x2 grid, each featuring the same blue-haired woman with a blurred face and bold Japanese text. The 4 card labels above the tiles are: サムネイル画像, 記事のアイキャッチ画像, LPのセクション画像, SNS投稿画像. The large text inside the 4 cards should read respectively: 1) AIで変わるクリエイティブの未来, 2) AI時代のクリエイティブ戦略 成功する企業の条件, 3) AIで加速するビジネス成長, 4) 未来をつくるのは AI×あなたのアイデア. Between the left and right panels, place a bright glowing gold arrow pointing from left to right with spark-like particle trails, indicating transition or superiority shift. Along the bottom, add a very large black banner with a glowing gold border and massive bold gold text reading {argument name="bottom banner text" default="GPT Image 2へ"}. Overall composition should feel like a comparison graphic showing a shift from older image generation to more practical commercial output, with aggressive thumbnail typography, strong glow effects, metallic texture on major text, and polished social-media marketing visuals.

A futuristic Japanese tech comparison poster in a dark cyberpunk control-room setting, wide 16:9 composition. Large distressed white Japanese headline text at the upper left reading "三つ巴", with a bold gold subtitle directly below reading "それぞれの武器". Across the center-left are 3 glowing holographic comparison panels arranged horizontally and connected by neon arrows: a blue panel labeled "Google", an amber-gold panel labeled "Claude", and a purple-magenta panel labeled "OpenAI". The Google panel contains 4 inner cards: 2 larger top cards labeled "Gemini" and "Antigravity", plus 2 smaller bottom cards showing analytics/dashboard-like visuals and a blue isometric cube graphic. The Claude panel contains 4 inner cards: 1 large top card labeled "Claude Code", plus 3 smaller bottom cards showing a network diagram, text/code list, and chart analytics. The OpenAI panel contains 5 inner cards: 2 larger top cards labeled "ChatGPT" and "Codex", plus 3 smaller bottom cards showing interface/code windows and a geometric wireframe cube. Add glowing bidirectional arrows between Google and Claude, and between Claude and OpenAI. At the bottom center, place a large neon-framed banner with gold text reading "Google / Claude / OpenAI". On the right side, include a young woman standing and pointing left toward the panels, with long straight split-dyed hair in pastel pink and cyan blue, a plain white t-shirt with black text reading "{argument name="shirt text" default="OKIHIRO AI Creative"}", and a soft pink pleated skirt. Her face is obscured by a smooth rectangular blur block. Use cinematic sci-fi lighting, glossy hologram UI details, high contrast, vivid blue-gold-purple accents, and a polished YouTube thumbnail aesthetic.

GPT Image 2.0 on ChatGPT vs Google Nano Banana 2 on Gemini. PROMPT: Create a hyper-realistic, cinematic portrait of me (use uploaded face) as a modern mafia boss. I'm sitting in a luxury black car, wearing a black suit and tinted aviator sunglasses, smoking a thick cigar. Cold, fearless expression. Background: moody sky + blurred city/street for noir feel. Cool tones, high contrast. Sharp details on face & smoke. Style: 8K, movie-poster quality, shallow depth of field 1:1

请帮我生成一张视觉品牌拆解图，稍后我需要用来生成讲解的视频。

一颗/一块/一枚【食物名称】，以博物学大师发现野外标本的方式解剖。
剖开、展开、固定——如同博物馆的珍贵藏品，
却以卡拉瓦乔为《国家地理》掌镜时的光线照亮。
每一个内部结构都以自身的材质真相发光。
截面锋利得近乎暴力。内部美丽得近乎神圣。
画面中呈现完整标本：
一半保持原状，展示【外表面描述：质感/颜色/纹理】；
另一半剖开至核心，【内部核心结构描述：最重要的1—2个内部视觉特征】清晰可见。
【补充1—2句该食物最具视觉张力的横截面细节描述】
背景：纯粹的黑丝绒。
【食物名称】悬浮其中，如同某件珍贵而危险的事物。
标注文字紧贴结构边缘，手写感衬线字体，绝不悬空飘浮。
画面包含以下标注，每处标注三行：第一行结构名称，第二行成分数据，第三行一句人话：
【结构01名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】

【结构02名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】
【结构03名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】
【结构04名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】
【结构05名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】

【结构06名称】
【成分／数据说明】
【这个结构在做什么，为什么重要】
省略其他如果有继续保持这个格式
主标题，左上角，暖象牙白大写字体：
【食物名称】·解剖

斜体副标题紧随其下：
【一句揭示这种食物本质的话，不超过15字】

整体气质：奥杜邦博物插画×卡拉瓦乔光影×有史以来最美的科学摄影。
4K精度，标本照明，极致内部细节。
没有任何临床感，一切都鲜活。
写实风格，非示意图，非卡通，非简化图解。
每一种材质都有真实的物理质感：
粗糙的、光滑的、湿润的、干燥的、致密的、疏松的。

新中式极简东方美学 × 高端商业插画，主题一花一世界，
极简，克制，空灵，高级商业视觉，超现实东方意境，
画面干净通透，无灰雾、无脏色，

一朵巨大的荷花作为空间容器，从平静水面自然生长，轻微倾斜，构图优雅留白充足，

低饱和干净粉色，柔和胭脂调，花瓣半透明，轻盈通透，
哑光低对比，边缘柔化 + 轻微景深，

荷花内部为唯一视觉焦点：发光的3D微缩广州城市，包含：广州塔，珠江新城建筑群，猎德大桥，珠江水岸，少量岭南建筑，

城市超精细结构，真实材质，极高细节清晰度，城市高光是暖金色，城市阴影是冷青蓝，形成冷暖对比，

灯光通透有能量，局部高饱和但不泛滥，城市亮度明显高于荷花，

水面清澈极简平静，仅少量柔和涟漪，弱反射，

背景暖米白宣纸质感，无水墨、无笔触，大面积留白，
中心有极轻微光晕渐变，整体通透、不灰、不闷，

画面下方一艘极简小船，船上一位红衣渔女，极小比例，
静立仰望荷花，红色为唯一高纯度点缀，

整体光线通透、干净、有层次，无灰雾、无泛白，
高端CG商业插画，电影级真实光影，高动态范围，超精细，8K细节，ArtStation 级画质，强化分色，干净调色，青橙对比，暖高光冷暗部，仅城市灯光提亮饱和度，色调柔和通透，光影锐利明亮，无灰雾、无暗沉、无低饱和雾化。

请根据【主题：xxx】自动生成一张高审美的"轮廓宇宙 / 收藏版叙事海报"风格作品。不要将画面局限于固定器物或常见容器，不要优先默认瓶子、沙漏、玻璃罩、怀表之类的常规载体，而是由 AI 根据主题自行判断并选择一个最契合、最有象征意义、轮廓最强、最适合承载完整叙事世界的主轮廓载体。这个主轮廓可以是器物、建筑、门、塔、拱门、穹顶、楼梯井、长廊、雕像、侧脸、眼睛、手掌、头骨、羽翼、面具、镜面、王座、圆环、裂缝、光幕、阴影、几何结构、空间切面、舞台框景、抽象符号或其他更有创意与主题代表性的视觉轮廓，要求合理布局。优先选择最能放大主题气质、最能形成强烈视觉记忆点、最能体现史诗感、神秘感、诗意感或设计感的轮廓，而不是最安全、最普通、最常见的容器。

画面的核心不是简单把世界装进某个物体里，而是让完整的主题世界自然生长在这个主轮廓之中、之内、之上、之边界里或与其结构融为一体，形成一种"主题宇宙依附于一个象征性轮廓展开"的高级叙事效果。主轮廓必须清晰、优雅、有辨识度，并在整体构图中占据核心地位。轮廓内部或边界中需要自动生成与主题强绑定的完整叙事世界，内容应当丰富、饱满、层次清晰，包括最能代表主题的标志性场景、核心建筑或空间结构、象征符号与隐喻元素、角色关系或文明痕迹、远景中景近景的空间递进、具有命运感和情绪张力的氛围层次，以及门、台阶、桥梁、水面、烟雾、路径、光源、遗迹、机械结构、自然景观、抽象形态、生物或道具等叙事细节。所有元素必须统一、自然、有主次、有层级地融合，像一个完整世界真实孕育在这个轮廓结构之中，而不是简单拼贴、裁切填充、素材堆叠或模板化背景。

整体构图需要具有强烈的收藏版海报气质与高级设计感，大结构稳定，主轮廓强烈明确，内部世界具有纵深、秩序和呼吸感，细节丰富但不拥挤，内容丰满但不杂乱，可以适度加入小比例人物剪影、远处建筑、光柱、门洞、桥、阶梯、回廊、倒影、天光或远景结构来增强尺度感、故事感与史诗感。整体画面要安静、宏大、凝练、富有余味，不要平均铺满，不要廉价热闹，不要无重点堆砌。

风格融合收藏版电影海报构图、高级叙事型视觉设计、梦幻水彩质感与纸张印刷品气质，强调纸张颗粒感、边缘飞白、水彩刷痕、轻微晕染、空气透视、柔和雾化、局部体积光、光雾穿透、大面积留白与克制版式，让画面看起来像设计师完成的高端收藏版视觉作品，而不是普通 AI 跑图。整体气质要高级、诗意、宏大、神圣、怀旧、安静、具有传说感和叙事感。

色彩由 AI 根据主题自动判断并匹配最合适的高级配色方案，但必须保持统一、克制、耐看、低饱和、高级，不要杂乱高饱和，不要廉价霓虹感，不要塑料数码感。配色可以围绕黑金灰、冷蓝灰、雾白灰、褐红米白、暗铜、旧纸色、深海蓝、暮色紫、银灰等体系自由变化，但必须始终服务主题，并保持海报级审美与整体和谐。

最终要求：第一眼有强烈的主题识别度和轮廓记忆点，第二眼有完整丰富的叙事世界，第三眼仍有细节和余味。轮廓选择必须具有创意和主题匹配度，尽量避免重复、保守、常见的容器套路，优先选择更有象征性、更有空间感、更有设计潜力的轮廓形式。不要普通背景拼接，不要生硬裁切，不要模板化奇幻素材，不要游戏宣传图感，不要过度卡通化，不要过度写实导致失去艺术感，不要形式大于内容。如果合适，可以自然加入低调克制的标题、编号、签名或落款，让它更像收藏版海报设计的一部分，但不要喧宾夺主。

参考这张图的透视和风格，绘制一张更加精细的超高清插画，表现超现实主义的日式未来都市，要能看清很小的细节，包括街道上的传统文化游行的人，小巷里的黑帮，烟花巷的舞女，疲惫的社畜，楼房的窗户里也有各式各样的人物，学习的学生，吵架的夫妻，玩游戏的宅男，以及更多的发挥细节。讽刺现实拥挤中的无聊，都市繁华下的孤独，无意义的人生中又有一种病态的美感。画面要有极高的审美价值 ，不能因为拼内容而损失美和协调感，比例是9:16

Research LIME Drug Design and make a detailed infographic about it

吕布游戏Boss设定，赤兔马方天画戟，暗黑进化形态双形态对比

1、生成不知火舞和貂蝉的游戏对战海报图
2、生成一张K-pop团体时尚专辑封面
3、请你生成 《斗破苍穹》 的关键人物关系图
4、帮我截一张上传图片的抖音首页的女网红图

社交App匹配成功界面，两个用户资料卡碰撞爱心特效

{"type":"comparison graphic","style":"anime cinematic demonstration image on a black presentation background","canvas":{"aspect_ratio":"4:3","background":"solid black"},"text_elements":[{"text":"{argument name=\"headline text\" default=\"Move the camera POV to be at ground level in the crowd.\"}","position":"top center","style":"large white sans-serif"},{"text":"Source","position":"left of upper image","style":"large white sans-serif"},{"text":"Output","position":"left of lower image","style":"large white sans-serif"}],"layout":{"sections":[{"title":"Source","position":"upper center","count":1,"labels":["overhead crowd scene"]},{"title":"Output","position":"lower center","count":1,"labels":["ground-level crowd POV scene"]}],"image_frames":2},"images":[{"role":"source image","composition":"busy top-down view of a densely packed historical street crowd, seen from above","scene":"a chaotic crowd gathered around a wagon and a horse-drawn carriage, people pressed shoulder to shoulder, many wearing caps and muted early-20th-century or old-European clothing, bundles and sacks visible, one brown horse at the right edge, wooden wagon wheel and cart structure partially visible","camera":"high overhead bird's-eye angle looking down into the crowd","lighting":"soft daylight","color_palette":"muted earthy browns, dusty blues, beige, olive, warm gray","rendering":"hand-painted anime film still, detailed crowd illustration, slightly soft shading"},{"role":"output image","composition":"the same crowded historical street reimagined from inside the mass of people at near-ground height","scene":"view from within the crowd beside a carriage wheel, bodies filling the foreground and midground, a person in dark maroon clothing bent forward at left, a crouched figure in green near the bottom center, a woman in a light blue dress at right-center turning back, tightly packed figures, horse and cart implied nearby, dramatic sense of compression and closeness","camera":"very low ground-level POV from inside the crowd, upward and forward through people, emphasizing complex occlusion and depth","lighting":"soft daylight with warm cinematic shadows","color_palette":"muted earthy browns, dusty blues, beige, olive, warm gray","rendering":"hand-painted anime film still, cinematic perspective shift, detailed character crowding, soft painterly shading"}],"overall_goal":"show a before-and-after camera angle transformation of the same anime crowd scene, with the output moving from an overhead view to a low immersive POV inside the crowd"}

ポラロイド写真の中に人が写っていて、その人がフレームから外に飛び出している画像。日本語が書いてある画像生成して

←下の画像
GPT Image-2で生成したやつ→

东方幻想风格女性，半身肖像，回眸侧脸，气质空灵优雅，柔和神性美感，细腻五官，微垂眼神，冷白细腻肌肤，淡雅橘粉妆容，金色高光点缀

长发飘动，发丝中融入彩色花朵与光粒（红、蓝、橙、紫），头发具有流动感与空气感

身穿半透明丝绸礼服与披肩，材质轻盈通透，布料随风飘动，表面带有鎏金纹理与闪耀颗粒。

整体光影为暖金色逆光，强边缘光，体积光明显，光粒漂浮，柔光泛光，梦幻氛围

背景干净浅色渐变，带微光与粒子效果，整体氛围空灵、梦境、神圣

风格：高端CG插画，超精细，电影级光影，柔光渲染，8K细节，artstation 热门作品风格

苏妲己古风写真，红色纱衣半透，狐耳若隐若现，媚态撩人

狐妖小红娘涂山雅雅写真大片，粉色九尾狐裘紧身裙，媚眼如丝，红唇微张，极致妖媚

9:16 竖版，极致武侠风，绝美东方女侠，20岁出头，冷艳锐利丹凤眼，眉宇英气逼人，肤白如玉，长直黑发湿漉漉随狂风剧烈飞舞，几缕发丝贴在脸颊和颈侧，穿着湿透的深黑改良武侠劲装，外披宽袖玄色长袍，衣袍和长袖被风吹得剧烈飘扬翻飞，紧身劲装勾勒身材，腰束软剑带，足踏长靴，右手持一把古剑，剑身散发幽蓝剑气光芒，动态姿势：身体微侧回眸，衣袂猎猎，背景为月夜雨雾笼罩的竹林古道，巨大明月高悬，石板小径，古灯笼，薄雾雨丝，戏剧性冷月光与蓝光剑气结合，湿身水光效果，超强动态感，细腻布料褶皱、头发丝飘动、真实水珠反光，电影级光影，8k，masterpiece, best quality, ultra realistic, cinematic, dramatic atmosphere

GPT IMAGE 2 on ChatGpt Prompt: Transform the provided reference image into a cozy aesthetic scrapbook-style composition while strictly preserving the original subject, identity, pose, lighting, and background. Add multiple small "mini version" characters of the same person (chibi / doll-like style), placed naturally around the scene (on objects, table, shoulder, etc.). These mini figures must match the subject's face, hairstyle, outfit, and vibe consistently, styled as cute 3D collectible figurines. Show them doing different activities (reading, posing, taking photos, relaxing). Overlay handwritten-style doodles and annotations across the image: arrows, hearts, stars, sparkles, icons, and playful captions connected to elements in the scene. Use a soft pastel color palette (white base with pink, peach, blue accents). Keep the frame visually rich and filled but balanced and clean. Style: warm, cozy lighting, dreamy Instagram scrapbook aesthetic, soft depth of field, highly detailed, polished but playful. The final result must look like the SAME original image enhanced with mini alter-egos and aesthetic annotations — not a recreated or different scene

请根据上传的人像照片，生成[分析类型]图卡，
包含：[维度1]、[维度2]、[维度3]、[维度4]，
每个维度对比：推荐 / 普通 / 不建议，
保留真实五官，视觉为主，文字简短，[比例]

A side-by-side comparison graphic on a black background,
[对比条件说明] at the top in large white text,
Source image: [场景A描述],
Output image: [场景B描述],
Preserve [不变的元素], change only [变化的元素]

[风格名称]，[核心气质关键词3-5个]，
[主体描述：人物/物品/场景]，
[色彩体系：主色调 + 对比色 + 点缀色]，
[光影定义：光源方向 + 冷暖对比]，
[材质质感：2-3种材质描述]，
[审美底线：3-5个"不要xxx"]

一颗/一块/一枚【物品名称】，以[分析方式]解剖，
一半保持原状，另一半剖开，
[N个结构标注，每个包含：名称 + 数据 + 解释]，
背景：[纯色背景]，
整体气质：[审美参考A] × [审美参考B]