id stringlengths 21 67 | model stringclasses 11
values | mode stringclasses 2
values | category stringclasses 15
values | tags listlengths 2 6 | title_en stringlengths 29 67 | title_zh stringlengths 7 43 | prompt stringlengths 40 14k | notes_en stringlengths 59 380 | notes_zh stringlengths 15 121 | source_name stringlengths 10 31 | source_url stringlengths 45 103 | credit stringlengths 1 38 | likes int64 3 22.5k ⌀ | posted stringdate 2025-08-21 00:00:00 2026-10-30 00:00:00 ⌀ | video_url stringclasses 0
values | preview imagewidth (px) 480 720 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
chatgpt-1980s-male-portrait-warm-lamp | ChatGPT | image-edit | portrait | [
"1980s",
"portrait",
"film-look",
"identity-preserving"
] | 1980s man in blazer, silk shirt and gold chain | 身穿西装与金链的80年代男子肖像 | Transform this photo into an authentic late-1980s male portrait while preserving my exact facial identity, facial proportions, skin tone, eyes, nose, lips, jawline, facial hair, and all natural facial features. Do not change my face or make me look like another person. Give me authentic 1980s voluminous men’s hair—thic... | The prompt locks identity with an explicit preserve list, then builds the period look in layers: teased hair, a structured blazer and patterned silk shirt, and an upscale interior. It finishes with camera cues such as soft flash, halation, a named Kodak film stock and a 35mm lens, which push the result toward a scanned... | 提示词先用明确的保留清单锁定面部特征，再分层描述发型、西装与花衬衫、高档室内布景，最后给出柔光闪光、光晕、柯达胶片和35mm镜头等相机细节，让结果更像扫描的老照片而非光滑的AI渲染。 | @lenardfrancisco on TikTok | https://www.tiktok.com/@lenardfrancisco/video/7683830698725215495 | Lenard francisco | 86 | 2026-09-10 | null | |
chatgpt-1980s-self-portrait-scarf | ChatGPT | image-edit | edit | [
"80s",
"retro",
"fashion",
"vintage-film"
] | Woman in 1980s teal blouse and headscarf | 身穿青绿色衬衫和头巾的女子，1980年代风格 | Transform my appearance into an authentic 1980s version of myself while keeping my facial features and identity recognizable. Give me an iconic 80s scarf, vintage clothing, retro accessories, and an authentic 1980s atmosphere. Make it look like a real photograph taken in the 1980s, with vintage film grain, natural ligh... | The prompt fixes identity first by asking for recognizable facial features and no face changes, then builds the period look from concrete items such as a scarf, belt and shoulder bag. Framing the result as a real 1980s photograph with film grain, natural light and realistic skin pushes the output toward a candid snapsh... | 提示词先要求保留可辨认的面部特征并且不改变长相，再用头巾、腰带和肩包等具体单品搭建年代感。把结果设定为真实的1980年代照片，并加入胶片颗粒、自然光和真实肤质，使画面更像抓拍而不是造型图。 | @ishaprayag on TikTok | https://www.tiktok.com/@ishaprayag/video/7683911490646199573 | Isha Prayag | 942 | 2026-09-10 | null | |
chatgpt-1985-family-photo-restyle | ChatGPT | image-edit | edit | [
"80s-trend",
"retro",
"date-stamp",
"photo-edit"
] | Family photo restyled as a 1985 snapshot | 家庭合影转换为1985年的老照片 | Using my uploaded photo, show me what I would have looked like around 1985. Preserve my identity, facial features, skin tone, age, and recognizable appearance. Reimagine my hair, clothing, accessories, and surroundings with bold, unmistakably mid-1980s styling—expressive silhouettes, statement accessories, layered deta... | The prompt locks identity first and then changes wardrobe, accessories and setting, so the faces stay recognizable. Analog grain, faded colour, direct flash and a specified red orange date stamp make the result read as a period snapshot. | 提示词先锁定人物身份，再改变服装、配饰和场景，因此人脸仍然可辨认。模拟颗粒、褪色色调、直闪光和指定的红橙色日期水印，让结果更像真实的时期快照。 | @katrinayao on TikTok | https://www.tiktok.com/@katrinayao/video/7683941449594932500 | Kat😺 | 844 | 2026-09-10 | null | |
chatgpt-80s-indonesia-street-couple | ChatGPT | image-edit | edit | [
"retro-80s",
"couple",
"street-style",
"hijab"
] | Couple walking a 1980s Indonesian shopping street | 1980年代印尼街头散步的情侣 | Buat foto ini menjadi foto gaya anak muda sedang pacaran saling merangkul jadul era 80an dan berpakaian anak muda memakai (jaket Levis warna warni era 80an, perempuan tetap memakai jilbab), dengan beground pertokoan jadul dengan jalanan kota tahun 80 an. Buat jangan merubah wajah asli | Starts from a full body photo and asks for an edit rather than a new image, with an explicit instruction to keep the original face. Concrete period cues (colourful Levi jacket, hijab, old shop facades, 80s street) let the model fill the scene with matching signs and props. | 以全身照为原图进行编辑而非重新生成，并明确要求保留原有面部。提示词具体列出年代元素（彩色牛仔夹克、头巾、老式店铺立面、80年代街道），让模型补全相符的招牌和道具。 | @dianaunnie.799 on TikTok | https://www.tiktok.com/@dianaunnie.799/video/7684128723154816277 | Diana Unnie | 69 | 2026-09-11 | null | |
chatgpt-adidas-windbreaker-shutter-hand-over-mouth | ChatGPT / Gemini | text-to-image | realism | [
"street-style",
"adidas",
"flash-photography",
"portrait"
] | Man in Adidas windbreaker covering mouth by shutter | 身穿阿迪达斯风衣的男子在卷帘门前遮嘴 | A raw, street-style, candid photograph of a young Asian man standing in a medium waist-up shot. He is wearing a white oversized Adidas original windbreaker jacket with iconic black three-stripes down both sleeves and a small trefoil logo on the chest, a high neck with a hoodie and drawstrings, and black pants. His dark... | The prompt pins the garment details (stripes on both sleeves, chest trefoil logo, drawstring hood), the pose (hand covering nose and mouth, lower hand with motion blur) and the background, so each one appears in the output. Harsh direct flash against a dark ribbed shutter gives the high contrast street look. | 提示词明确写出服装细节（双袖条纹、胸前三叶草标、带抽绳连帽）、姿势（手遮住口鼻，下方手部带动态模糊）和背景，因此输出中每一项都能对应。强烈的直射闪光打在深色条纹卷帘门上，形成高反差的街拍质感。 | @trungayai on TikTok | https://www.tiktok.com/@trungayai/video/7689454589904571669 | 𝓣𝓻𝓾𝓷𝓰𝓐𝓘 | 1,492 | 2026-09-26 | null | |
chatgpt-autumn-leaves-forest-portrait | ChatGPT | image-edit | edit | [
"autumn",
"fall-leaves",
"portrait",
"photorealistic"
] | Woman tossing maple leaves on a forest path | 女子在林间小路上抛洒枫叶 | Using my photo for face, hair, and body, I'm aiming for a photorealistic autumn shot. Imagine me on a forest path, covered in colorful autumn leaves, laughing and looking straight up at the camera. My arms are stretched out above, as if I'm throwing leaves into the air. I'm wearing a black leather jacket, black top, da... | Anchoring identity to the reference photo with explicit face, hair and body instructions, then specifying an upward camera angle, golden evening backlight and layered maple leaves in the foreground and background gives the scene depth while keeping the subject recognizable. The editorial photography cue sets a clean, p... | 先明确用参考照保留脸部、发型和身材，再指定仰拍视角、金色黄昏逆光以及前后景飘落的枫叶，让画面有纵深感，同时保持人物可辨认。编辑摄影的说法则带来干净专业的质感。 | @klh000o1 on TikTok | https://www.tiktok.com/@klh000o1/video/7682245250466712863 | Kelly | 196 | 2026-09-06 | null | |
chatgpt-bandaged-hands-bearded-man-monochrome | ChatGPT / Midjourney | image-edit | portrait | [
"black-and-white",
"fine-art",
"portrait",
"studio-lighting"
] | Bandaged hands surround a bearded man in monochrome | 绷带手环绕蓄须男子的黑白肖像 | Create a cinematic, photorealistic black-and-white fine-art portrait of a rugged bearded man seated against a completely dark minimalist studio background. He wears a sharp black suit jacket with an open collar, creating a sophisticated but vulnerable appearance. Several realistic human arms and hands emerge from behin... | Uploading your own photo anchors the face, while the prompt fixes lighting, lens and composition in detail. Explicit hand-anatomy rules keep the extra arms believable and free of duplicated limbs. | 上传自己的照片可以锁定面部特征，提示词则详细规定了布光、镜头和构图。针对每只手解剖结构的明确约束，让额外的手臂看起来真实可信，没有重复的肢体。 | @mcsvisuals1 on TikTok | https://www.tiktok.com/@mcsvisuals1/video/7684407883559111966 | MCs | 436 | 2026-09-11 | null | |
chatgpt-barefoot-runner-coastal-sunset-road | ChatGPT | image-edit | realism | [
"sunset",
"coastal",
"low-angle",
"reference-photo"
] | Barefoot runner on coastal road at sunset | 日落时分在海岸公路上赤脚奔跑的女孩 | Use my uploaded photo as a reference for the person and keep the face, hair, body shape, and recognizable personal features as true to the original as possible. Create a photorealistic, cinematic scene in vertical format: the person is seen from behind and is running barefoot along the center line of a winding coastal ... | A very low camera just above the road with the subject centered on the center line makes the blurred foreground sneakers loom large and sets up strong depth. Warm backlight and the golden reflection on wet asphalt carry the mood. | 镜头贴近路面，主体居中对准路中线，前景虚化的运动鞋显得很大，形成强烈纵深。逆光暖色和湿润路面上的金色反光营造出氛围。 | @deniseleadingham on TikTok | https://www.tiktok.com/@deniseleadingham/video/7687793692115012894 | ❤️Denise L ❤️ | 37 | 2026-09-21 | null | |
chatgpt-bw-editorial-portrait-hand-on-chin | ChatGPT | image-edit | portrait | [
"monochrome",
"portrait",
"reference-photo",
"editorial"
] | Black and white portrait, man with hand on chin | 黑白肖像，男子手托下巴 | En utilisant le visage de référence ci-dessus, crée un portrait de mode noir et blanc à fort contraste d’un jeune homme portant une chemise blanche impeccable, assis dans une pose réfléchie, avec une main près du menton. Il est encadré par une silhouette floue au premier plan, créant une composition cinématographique d... | The blurred foreground silhouette creates a frame within a frame, and the prompt pins the look with an 85 mm lens, a Leica M11 rendering, soft rim light and monochrome tones. Deep shadows and bright highlights carry the mood while the reference face sets the identity. | 前景虚化的剪影形成框中框的构图，提示词用 85 毫米镜头、Leica M11 质感、柔和逆光和单色调来锁定画面风格。深阴影与强高光突出面部表情，参考脸部图像决定人物身份。 | @onsenfou390 on TikTok | https://www.tiktok.com/@onsenfou390/video/7691032466735238420 | ABC ABC on s’en fou 😜 | 14,052 | 2026-09-29 | null | |
chatgpt-couple-80s-retro-photo | ChatGPT | image-edit | edit | [
"80s",
"couple",
"retro",
"film-grain"
] | Couple recast as 1980s party portraits | 情侣变身80年代派对合影 | Com base nessa foto, crie uma imagem ultra realista de como nós seríamos se vivêssemos nos anos 80. Mantenha os traços do rosto mas adicione um penteado volumoso típico da época, roupas clássicas dos anos 80 com cores marcantes ou jeans de cintura alta, e uma iluminação com granulação de filme fotográfico antigo | The prompt tells the model to keep the facial features of the source photo while listing concrete era cues: voluminous hair, bold-colour clothing or high-waisted jeans, and aged film-grain lighting. Naming the identity constraint and the styling cues separately lets the model change the look without changing who is pic... | 提示词要求保留原图的面部特征，同时列出具体的年代元素：蓬松发型、高饱和色服装或高腰牛仔裤，以及老胶片颗粒光线。把身份约束和造型要素分开描述，模型就能只改造型、不改人物本身。 | @diegocarvalhoosteo on TikTok | https://www.tiktok.com/@diegocarvalhoosteo/video/7682999891639078164 | Dhyego | Dicas | 1,039 | 2026-09-08 | null | |
chatgpt-couple-80s-studio-portrait | ChatGPT | image-edit | portrait | [
"80s-fashion",
"retro-portrait",
"film-look",
"couple-portrait"
] | Couple posing in a 1980s film studio portrait | 情侣在1980年代胶片棚拍中合影 | Transform this couple photo into an authentic 1980s studio portrait while preserving both people’s exact facial features, face shape, identity, natural skin tone, and recognizable appearance. Do not change their faces or make them look like different people. Give the couple a classic 1980s romantic portrait aesthetic. ... | The prompt locks identity with an explicit preserve-faces instruction, then lists each person's hair and outfit item by item. Analog cues such as film grain, soft focus, mild vignette and rim light make the edit read as a real 1980s photo rather than a filtered one. | 提示词先用明确的保留五官指令锁定身份，再逐项列出每人的发型和服装。胶片颗粒、柔焦、轻微暗角和发丝轮廓光等模拟胶片的细节，让结果更像真实的1980年代照片，而不是滤镜效果。 | @aeziandlysa on TikTok | https://www.tiktok.com/@aeziandlysa/video/7683890505742109972 | Aezi & Lysa | 191 | 2026-09-10 | null | |
chatgpt-dancing-living-room-hairbrush-mic | ChatGPT | text-to-image | realism | [
"candid",
"lifestyle",
"photorealistic"
] | Woman dancing barefoot with hairbrush microphone | 赤脚女子用发刷当麦克风在客厅跳舞 | Create a candid photo of me dancing by myself in the middle of a cozy living room with music blasting. I'm barefoot, wearing distressed jeans and an oversized vintage band tee, holding a hairbrush like a microphone. My (describe your hair) hair is flying everywhere and I'm laughing uncontrollably. Warm evening lighting... | Concrete wardrobe, prop and emotion cues (barefoot, distressed jeans, hairbrush as microphone, uncontrolled laughter) make the moment read as candid. The 3:4 ratio and warm evening light set the frame and mood. | 具体的服装、道具和情绪描述（赤脚、做旧牛仔裤、用发刷当麦克风、止不住地笑）让画面像抓拍的瞬间。3:4 比例和温暖的晚间光线定下画幅与氛围。 | @jenjusticeunfiltered on TikTok | https://www.tiktok.com/@jenjusticeunfiltered/video/7690247838596353311 | JenJusticeUnfiltered⚖️💣🔥 | 532 | 2026-09-27 | null | |
chatgpt-inuyasha-character-card | ChatGPT | image-edit | character | [
"anime",
"inuyasha",
"character-sheet"
] | Inuyasha style character profile of a moonlit shrine priestess | 犬夜叉风格角色设定卡，月下神社巫女 | Using my photo as reference, turn me into character in the Inuyasha universe with the exact artwork from series. With everything you know about me Pick who my Character would be. What friends I would have if any, would I have a lover ? Who is my enemy where would I live. What ability do I have. | The prompt turns a questionnaire into a structured character sheet, so each question becomes its own labelled panel that the model fills in turn by turn. Naming the franchise and asking for the exact series artwork keeps the anime style consistent across all panels. | 提示词把一份问卷变成结构化的角色设定卡，每个问题对应一个带标题的面板，模型逐项作答。点明作品名并要求沿用原作画风，使各个面板的动漫风格保持一致。 | @kathy_scorpio13 on TikTok | https://www.tiktok.com/@kathy_scorpio13/video/7681313938935500062 | kathy_scorpio495 | 193 | 2026-09-03 | null | |
chatgpt-iris-center-nature-color-palette | ChatGPT | image-edit | collage | [
"eye",
"nature",
"color-palette"
] | Iris at center ringed by nature color elements | 虹膜居中，四周环绕自然元素配色 | Analyze the eye color in the photo I uploaded in every possible detail and create an artistic visual for me. Place my iris photo exactly at the center of the composition. Around it, visualize every color found within my iris using real elements from nature. | Uploading a close up iris photo and asking for it to sit at the exact center anchors the composition on the real eye. Asking the model to map every iris tone to a named natural object turns the color analysis into a labeled radial layout. | 上传虹膜特写照片并要求精确放在中心，让整个画面以真实眼睛为锚点。要求把虹膜中的每一种色调对应到一个具体的自然物体，于是颜色分析变成带标签的环形排版。 | @jessica_feldman_ on TikTok | https://www.tiktok.com/@jessica_feldman_/video/7687762438644600077 | jessica_feldman_ | 77 | 2026-09-20 | null | |
chatgpt-lagos-1980s-portrait-reference | ChatGPT | image-edit | photo-to-art | [
"1980s",
"retro",
"lagos",
"portrait",
"film-grain"
] | Woman in 1980s Lagos street portrait | 1980年代拉各斯街头女性肖像 | Using my uploaded photo as the facial reference, create a realistic portrait showing what I might have looked like in the 1980s. Keep my facial features and overall appearance recognizable and natural. Give me an authentic 1980s hairstyle, such as voluminous curly hair, and dress me in a stylish vintage 80s outfit with... | Uploading a reference photo and asking to keep the facial features anchors the likeness, while the prompt sets the era through hairstyle, outfit and a named Lagos setting with period cars. Film grain, scratches, faded warm tones and soft focus make the result read as a physical print rather than a modern costume. | 先上传参考照片并要求保留五官来锁定人物相貌，再用复古发型、时代服装以及带有老式汽车的拉各斯场景限定年代。颗粒感、划痕、褪色的暖色调和柔焦让画面像一张真实的老照片，而不是现代的服装扮演。 | @ewawunmi76 on TikTok | https://www.tiktok.com/@ewawunmi76/video/7688410480255175956 | Ewawunmi💍💦🌹💎 | 190 | 2026-09-22 | null | |
chatgpt-malaysian-family-studio-portrait-1980s | ChatGPT | text-to-image | portrait | [
"retro",
"family",
"vintage-film",
"malaysia"
] | Vintage Malaysian family studio portrait with sepia backdrop | 马来西亚复古家庭影棚合影，棕褐色背景 | A vintage, nostalgic studio family portrait from the late 1980s or early 1990s in Malaysia. The photo shows a traditional Malaysian family: a father standing at the back wearing a formal grey suit jacket and a striped tie, a mother sitting gracefully in the center wearing a vintage floral Baju Kurung Moden with a match... | The prompt names the era, country and exact clothing for each person, then uses scanned-photo texture words such as faded colours, analog film grain and soft warm light to make it look like an old album print. Assigning each outfit to a position in the group keeps a multi-person composition readable. | 提示词写明年代、国家以及每个人的具体服装，再用褪色、模拟胶片颗粒和柔和暖光等扫描老照片的质感词，让画面像旧相册里的照片。按站位逐一指定服装，使多人构图清晰可控。 | @nynasweety on TikTok | https://www.tiktok.com/@nynasweety/video/7683135740514946324 | Nynasweety | 381 | 2026-09-08 | null | |
chatgpt-malaysian-hijab-lakeside-golden-hour | ChatGPT | text-to-image | portrait | [
"portrait",
"golden-hour",
"fashion",
"malaysia"
] | Hijab woman walking by a lake at golden hour | 黄金时刻湖边漫步的头巾女子 | Create a highly photorealistic full-body outdoor portrait of a stylish young Malay Muslim woman in her mid-20s, walking naturally beneath large mature shady trees beside a peaceful lake in Malaysia during golden hour. FACE & EXPRESSION: A naturally attractive young Malaysian woman in her mid-20s with soft feminine faci... | The prompt is split into labelled sections for face, hijab, outfit, pose, background, lighting, camera and composition, so each detail is set separately. The fixed framing rules (trunks on the left, both shoes in frame) and the 50mm lens spec keep the full-body shot stable. | 提示词按人脸、头巾、服装、姿势、背景、光线、摄影和构图分段列出，每项细节单独设定。固定的取景规则（树干位于左侧、双脚都在画面内）和50mm镜头设定让全身构图保持稳定。 | @mariahalqibtiyyah on TikTok | https://www.tiktok.com/@mariahalqibtiyyah/video/7684813808413527314 | Mariyah Al Qibtiyyah | 51 | 2026-09-13 | null | |
chatgpt-moana-spiritual-rebirth-eagle | ChatGPT | text-to-image | portrait | [
"transformation",
"animal-spirit",
"cinematic",
"moana"
] | Shirtless man merging with glowing blue eagle | 无上衣男子与发光蓝色雄鹰融为一体 | Create an image of what my spiritual rebirth would be, like grandmother Tala from Moana was turning into her glowing animal. Make the image in a cinematic and artistic style showing myself and what I would turn into. Base your answer off of all of our previous conversations and what you know about me. | The prompt gives a clear reference transformation, the Moana grandmother turning into a glowing animal, so the model has a visual template for the change. Asking it to draw on past chats and memory makes the scene personal, while the cinematic style cue sets the lighting and the mountain sunset backdrop. | 提示词以《海洋奇缘》中祖母塔拉化身发光动物的场景作为变身模板，为模型提供明确的视觉参照。要求模型参考过往对话和用户信息，使画面更具个人色彩，而电影感的风格描述则决定了光影和山间日落的背景。 | @shadstoker on TikTok | https://www.tiktok.com/@shadstoker/video/7682873877566999822 | Shad Cheyenne Stoker | 54 | 2026-09-07 | null | |
chatgpt-monochrome-cafe-portrait-iced-coffee | ChatGPT | text-to-image | portrait | [
"portrait",
"monochrome",
"cafe",
"editorial"
] | Black and white cafe portrait with iced coffee | 黑白咖啡馆人像配冰咖啡 | Create a photorealistic, cinematic black-and-white autumn portrait of a glamorous dark haired woman sitting at a cozy café table. She has shoulder length, softly curled dark hair, green eyes, natural glamorous makeup, lips glossy and a calm, confident, slightly mysterious expression. She wears a textured oversized blac... | Locks the monochrome editorial look with named light sources (natural window light, soft candle bokeh) and a single pose cue (head resting on one hand, direct eye contact). The closing framing rule to keep the hair, woman and iced coffee inside the frame keeps every key element in view. | 用明确的光源描述（自然窗光、柔焦烛光与虚化光斑）锁定黑白杂志人像质感，再用一个姿势指令（头靠在手上、直视镜头）定下亲密而随意的神态。结尾的构图规则要求保留头发、人物和冰咖啡，确保关键元素都在画面内。 | @ness_marie_ on TikTok | https://www.tiktok.com/@ness_marie_/video/7686203111882411277 | Ness_🖤🖤 | 374 | 2026-09-16 | null | |
chatgpt-monochrome-fashion-portrait-frame-in-frame | ChatGPT | image-edit | portrait | [
"portrait",
"monochrome",
"fashion",
"editorial"
] | Monochrome portrait framed by blurred foreground silhouette | 单色人像，模糊前景剪影框住主体 | Using the reference face above, create a high-contrast black-and-white fashion portrait of a young man wearing a crisp white shirt, sitting in a thoughtful pose with one hand near his chin. He is framed by a blurred silhouette in the foreground, creating a cinematic “frame within a frame” composition. Dramatic lighting... | A reference photo keeps the face consistent, and the frame within a frame composition places a blurred foreground silhouette in front of the subject for depth. Camera and lighting terms such as 85 mm lens, soft rim light, Leica M11 rendering and monochrome tones define the editorial look. | 使用参考照片保持人脸一致，框中框构图在主体前方加入模糊的前景剪影以增加纵深。85 毫米镜头、柔和轮廓光、莱卡 M11 风格和单色调等镜头与光线描述共同定义了这种杂志式质感。 | @hukamedia.zeii on TikTok | https://www.tiktok.com/@hukamedia.zeii/video/7691019172045802760 | mohamed huka fam zeii❤️ | 4,129 | 2026-09-29 | null | |
chatgpt-nigeria-jersey-football-dribble | ChatGPT | image-edit | realism | [
"football",
"sports",
"nigeria"
] | Young man dribbling football in Nigeria jersey | 身穿尼日利亚球衣带球过人的年轻男子 | Use the uploaded reference image as the exact facial identity reference. Create an ultra-realistic cinematic 8K sports photograph of the same young man confidently performing impressive dribbling skills with a football on a professional stadium pitch. He wears a Nigeria football jersey, matching shorts and white footba... | The prompt uploads a reference face and locks identity with an explicit likeness instruction, then builds the scene from exact kit details (jersey, shorts, white boots) and camera cues like mid-dribble motion, shallow depth of field and high speed sports photography. | 提示词先上传参考人脸并明确要求保持面部一致，再用球衣、短裤、白色球鞋等具体服装细节，以及带球动态、浅景深和高速体育摄影等镜头描述搭建场景。 | @adelex2468 on TikTok | https://www.tiktok.com/@adelex2468/video/7681363280471854354 | Adelex(Easy Life) | 8,859 | 2026-09-03 | null | |
chatgpt-night-street-portrait-pillar-violet-bridge | ChatGPT / Gemini | image-edit | portrait | [
"night",
"street",
"portrait",
"cinematic"
] | Man leaning on pillar under violet-lit bridge | 男子倚靠柱子，紫光大桥夜景 | Cinematic nighttime urban street portrait of a young man with natural medium-length black hair. He wears a minimalist oversized black short-sleeve shirt, casually leaning his back and shoulder against a large dark cylindrical street pillar. His body is oriented toward the left, while his head turns back toward the came... | Pairs an exact lens and eye level framing with a body angled away and a head turned back to camera, then layers light direction, a named color palette and film grain terms. The post has readers upload their own photo as the reference, so the face stays consistent while the scene is written out in detail. | 用精确的镜头焦段和平视机位，搭配身体侧转、头部回望镜头的姿态，再叠加光线方向、指定色板和胶片颗粒等描述。帖子要求上传自己的照片作参考，因此人脸能保持一致，场景则由文字细致描述。 | @trungayai on TikTok | https://www.tiktok.com/@trungayai/video/7691155856074870037 | 𝓣𝓻𝓾𝓷𝓰𝓐𝓘 | 535 | 2026-09-30 | null | |
chatgpt-personal-tattoo-meaning-sheet | ChatGPT | text-to-image | realism | [
"tattoo",
"personalized",
"memory",
"infographic"
] | Floral arm tattoo with compass, mountains and paw prints | 花卉手臂纹身，含罗盘、山脉与爪印 | create for me a tattoo that fits my personality perfectly based on everything you know and have learned about me, create an image of the tattoo and descriptions of the meaning behind each part of the tattoo design you created. tell me why the tattoo created suits my personality perfectly | The prompt draws on ChatGPT's saved memory of the user, so no personal details need to be typed in. Asking for the design, a meaning for each element and a closing rationale in one message produces a mockup with a matching legend panel. | 这个提示词依赖 ChatGPT 保存的用户记忆，无需手动输入个人信息。一次性要求设计图、每个元素的寓意以及最终理由，就能得到带配套图例面板的纹身效果图。 | @vangiegilley82 on TikTok | https://www.tiktok.com/@vangiegilley82/video/7682941647562231054 | Vangie Gilley 🫶 | 49 | 2026-09-07 | null | |
chatgpt-queen-acronym-poster-personal | ChatGPT | image-edit | poster | [
"acronym",
"typography",
"personalized",
"identity-reference"
] | Gold QUEEN acronym poster with personal collage | 金色 QUEEN 缩写海报配个人拼贴 | Using my uploaded photo as the ONLY identity reference and what you have learned about me through our conversations when available, choose ONE word that best represents who I am, then turn that word into a personalized acronym. Do NOT ask me to choose the word—surprise me. Choose the word independently based entirely o... | The prompt lets the model choose the word and build the acronym itself, so each letter gets its own trait line. Naming the uploaded photo as the only identity reference keeps the face consistent while props and layout carry the personality. | 提示词让模型自行选词并生成缩写，因此每个字母都有一行特质说明。将上传照片设为唯一身份参考，使人脸保持一致，道具和版式则承载人物个性。 | @ninersqueen76 on TikTok | https://www.tiktok.com/@ninersqueen76/video/7686313195148954894 | ❤️NinersQueen76💛🇺🇸 | 79 | 2026-09-17 | null | |
chatgpt-rancher-service-double-exposure-profile | ChatGPT | image-edit | poster | [
"double-exposure",
"monochrome",
"portrait",
"editorial-poster"
] | Cowboy profile merged with service and ranch scenes | 牛仔侧脸与服役、牧场场景的双重曝光 | Create a vertical ASPECT_RATIO monochrome editorial double-exposure portrait based on PERSON_REFERENCE. The final image must preserve the recognizable identity of the person: face shape, skin tone, hairstyle direction, facial proportions, age impression, expression, and overall presence. Transform the person into a lar... | A placeholder template fixes the layout: a large right-facing profile with a dense story collage inside the silhouette. Leaving STORY_THEME blank lets the model build a story that fits the subject's expression, while the grain, lighting and monochrome terms keep the finish gritty rather than plastic. | 模板用占位符固定了构图：右侧是大幅侧脸，轮廓内部是密集的故事拼贴。故事主题留空时，模型会自行构思与人物表情相符的故事，而颗粒、光影和单色等描述让成品保持粗粝质感，而非塑料感。 | @wranglerjf on TikTok | https://www.tiktok.com/@wranglerjf/video/7683621459196955917 | Wrangler | 1,498 | 2026-09-09 | null | |
chatgpt-rugrats-kid-reptar-bedroom | ChatGPT | text-to-image | character | [
"rugrats",
"reptar",
"80s-nostalgia",
"personalized"
] | Braided girl in Rugrats bedroom with Reptar | 编织辫女孩与 Reptar 在《Rugrats》卡通房间里 | With everything you know about me, what would I look like as a Rugrats kid. Include the toy Reptar, other characters and things from the show. | Asking for personal details lets the model fill the scene with a notepad checklist, a name necklace and other props that feel custom. Naming the show, the toy Reptar and the cast keeps the cartoon look and the characters recognizable. | 要求模型结合个人信息，让画面里出现便签清单、名字项链等定制道具。点明节目、玩具 Reptar 和角色阵容，能让卡通风格和人物都清晰可辨。 | @totalparadise on TikTok | https://www.tiktok.com/@totalparadise/video/7685035709698297102 | 👑💎 ♦️𝓒𝓱𝓻𝓲𝓼𝓼𝔂💎👑☕️♣️ | 969 | 2026-09-13 | null | |
chatgpt-rugrats-kid-reptar | ChatGPT | image-edit | photo-to-art | [
"rugrats",
"cartoon",
"reptar"
] | Girl with locs as Rugrats kid with Reptar | 扎辫女孩变身蠢蛋一族小孩抱着雷普塔 | With the picture provided and what you know about me Show me what I would look like as a Rugrats kid. Include the toy Reptar(or whatever one you liked from the show) | Naming the show and one iconic toy gives the model a clear cartoon target, while the reference photo keeps the likeness. The prompt is short, so the model supplies the scene, props and styling on its own. | 指定节目和一个标志性玩具，既给出明确的卡通目标，又借参考照片保留人物相貌。提示词很短，场景、道具和风格由模型自行补充。 | @kenyana940 on TikTok | https://www.tiktok.com/@kenyana940/video/7684885263046249742 | Kenyana940 | 997 | 2026-09-13 | null | |
chatgpt-seated-man-reaching-hands-smoke-portrait | ChatGPT | image-edit | portrait | [
"portrait",
"dark-studio",
"conceptual",
"smoke"
] | Seated man surrounded by reaching hands in smoke | 坐在烟雾中被伸来的手包围的男子 | Create a cinematic, photorealistic conceptual portrait of a young man sitting alone in a dark minimalist studio. He sits centered on a simple chair or low seat, leaning slightly forward with his head bowed and hands clasped between his knees. His posture communicates emotional exhaustion, isolation, anxiety, and deep i... | Asking the model to preserve facial anatomy after a photo upload keeps the subject's face stable while the prompt builds the scene. Describing a narrow overhead spotlight on black, with background hands kept softer than the man, holds the figure as the visual focus. | 上传照片后要求保留面部结构，使人物面部保持稳定，同时由提示词搭建场景。描述纯黑背景上的窄幅顶光，并让背景中的手比人物更柔和，使人物保持视觉焦点。 | @mcsvisuals1 on TikTok | https://www.tiktok.com/@mcsvisuals1/video/7684298113783303455 | MCs | 2,575 | 2026-09-11 | null | |
chatgpt-yacht-sequin-gown-night-portrait | ChatGPT / Kling | image-edit | portrait | [
"fashion",
"night-portrait",
"yacht",
"photo-reference"
] | Woman in sequined gown on yacht corridor | 游艇走廊中身穿亮片长裙的女子 | Use my uploaded photo as the identity reference and preserve my recognizable facial features, blue eyes, hair characteristics, natural body proportions and authentic skin texture ULTRA-PHOTOREALISTIC • LUXURY NIGHT EDITORIAL • BLACK GLAMOUR • CINEMATIC FLASH PHOTOGRAPHY • VERTICAL 4:5 Create an ultra-photorealistic lux... | Pinning the identity to an uploaded photo, then fixing pose, gown, setting and lens in a numbered brief keeps the result on-subject. The flash, warm yacht lights and a 35 to 50mm lens with shallow depth of field give it the editorial look. | 先用上传的人像照片锁定身份，再把姿势、礼服、场景和镜头写成分项说明，结果更贴合要求。闪光灯、暖色游艇灯光，加上 35 到 50 毫米镜头与浅景深，营造出时尚大片的质感。 | @amandakravyn on TikTok | https://www.tiktok.com/@amandakravyn/video/7684346251210542349 | Amanda Kravyn🥀🖤 | 722 | 2026-09-11 | null | |
chatgpt-young-man-earphones-dusk-portrait | ChatGPT / Gemini | text-to-image | portrait | [
"portrait",
"sunset",
"cinematic",
"backlight"
] | Young man with earphones at dusk | 黄昏时戴耳机的年轻男子 | Photorealistic cinematic dusk portrait of a young Asian man , medium-length naturally tousled black hair with soft loose bangs falling across his forehead, delicate refined facial features, wearing an oversized black hooded jacket with a subtle light-colored inner collar and simple white wired earphones. He has a quiet... | The prompt places the subject in the lower right and the sun at left, leaving the upper sky empty as negative space, with foliage entering from the top edge as a foreground frame. An 85mm telephoto at f/2.8, a backlit silhouette body and a muted palette make the shot read as candid and cinematic rather than posed. | 提示词把人物放在右下、太阳放在左侧，上方天空留白，树枝从画面上缘伸入形成前景框架。85mm 长焦、f/2.8 光圈、逆光剪影的身体和低饱和色调，让画面呈现抓拍般的电影感而非摆拍。 | @trungayai on TikTok | https://www.tiktok.com/@trungayai/video/7690052053447626004 | 𝓣𝓻𝓾𝓷𝓰𝓐𝓘 | 1,580 | 2026-09-27 | null | |
gemini-heartbreak-storm-kneeling-man-woman-steps | Gemini | image-edit | realism | [
"heartbreak",
"cinematic",
"rain",
"poster"
] | Heartbroken man kneels in rain as woman turns away | 心碎男子雨中跪地，女子背身而坐 | Photorealistic cinematic heartbreak artwork in a wide **16:9 composition**, depicting a devastated young adult man kneeling in heavy rain on the left and a young adult woman sitting turned away from him on stone steps on the right. A violent thunderstorm and branching white lightning fill the sky above a distant city, ... | The frame splits into a cool storm-lit side and a warm amber side, so the emotional contrast reads at a glance. Symbolic keepsakes like the photo, teddy bear, roses and ring box sit in the foreground and tell the breakup story without any text. | 画面分为冷色暴风雨侧与暖色琥珀光侧，情绪对比一目了然。照片、泰迪熊、玫瑰和戒指盒等象征物置于前景，无需文字即可讲述分手的故事。 | @sirus_official on TikTok | https://www.tiktok.com/@sirus_official/video/7689906533546691848 | prompt | 1,199 | 2026-09-26 | null | |
gemini-hooded-man-hillside-cafe-night | Gemini | image-edit | portrait | [
"portrait",
"night",
"streetwear",
"reference-photo"
] | Hooded man relaxing on hillside cafe terrace at night | 头戴连帽的男子在山坡咖啡露台夜间放松 | Make a photo showing a man that I attached who is sitting relaxed on the terrace of a cafe or an open area in a highland area at night, with the background of hills and glittering city lights. Subject and Pose A man sits leaning slightly with his back to the camera with a relaxed gesture on an outdoor folding chair (ca... | Attaching the reference photo and describing clothing, pose and background in labelled sections keeps the person and outfit consistent. Direct front flash separates the subject from the dark hills, while the cool palette makes the warm village lights stand out. | 附上参考照片，并用分段方式描述服装、姿势和背景，可以保持人物与穿着一致。正面直闪将主体从黑暗的山景中分离出来，冷色调与暖黄色灯光形成对比。 | @geminiaiprompt on TikTok | https://www.tiktok.com/@geminiaiprompt/video/7685850288816753940 | JASA EDIT AI | PROMPT AI | 1,049 | 2026-09-15 | null | |
gemini-vm-letter-couple-portrait | Gemini | image-edit | photo-to-art | [
"couple",
"typography",
"3d-letter",
"luxury"
] | Couple posing inside a navy VM letter | 情侣置身深蓝 VM 字母造型人像 | Masterpiece, ultra-photorealistic, premium luxury editorial portrait, vertical composition. Create a large three-dimensional capital letter “VM” as the main centerpiece. The V and M must be clearly recognizable, perfectly formed, elegant and symmetrical, with the couple naturally integrated into the structure of the le... | Upload the couple's photo as a reference so Gemini keeps their faces, then state where each person sits along the V and M strokes. The prompt also fixes the material, lighting and exact text, which keeps the letter readable and the typography clean. | 先上传情侣照片作为参考以保持人脸一致，再说明两人在 V 和 M 笔画上的位置。提示词同时固定了材质、光线和文字内容，使字母清晰、排版干净。 | @gemini_photo on TikTok | https://www.tiktok.com/@gemini_photo/video/7687294178681818381 | gemini AI | 300 | 2026-09-19 | null | |
gpt-image-100-elements-scene-inventory | GPT Image 2 | text-to-image | infographic | [
"inventory",
"dense-scene",
"text-list",
"counting",
"illustration"
] | Town scene with numbered 100 element list | 附100项编号清单的小镇场景 | Create a scene with 100 elements, and list the 100 elements within the image. | A fixed element count plus a request to list every item makes the model plan the scene as an inventory. The result is a dense illustrated town under a numbered text list. | 固定元素数量并要求逐项列出，让模型把画面当作清单来规划。结果是一幅密集的插画小镇，上方配有编号文字清单。 | @umesh_ai on X | https://x.com/umesh_ai/status/2046569229189849228 | umesh_ai | null | 2026-04-21 | null | |
gpt-image-1942-arcade-city-pilot-walk | GPT Image 2 | image-edit | realism | [
"wwii",
"arcade",
"cinematic",
"worldbuilding",
"face-reference"
] | Pilot walking through a wartime city under arcade dogfights | 飞行员走过街机空战主题的战时城市 | Ultra-realistic cinematic “1942 City” scene, retro WWII arcade-shooter-inspired metropolis in real life, intense aerial-war atmosphere, vertical 4:5 editorial composition
Male, exact face, vintage military-pilot-inspired streetwear from head to toe, bomber jacket with retro aviation patches, fitted cargo trousers, rug... | Sections for worldbuilding, comedic details, lighting and camera give the model a clear checklist, so the scene fills with layered background gags. A reference face is locked with the phrase exact face while the outfit is spelled out piece by piece. | 提示词按世界观、喜剧细节、光线和镜头分段列出清单，让背景塞满层次分明的小细节。用 exact face 锁定参考人脸，同时把服装逐件写明。 | @Xaroon_x on X | https://x.com/Xaroon_x/status/2056940561148363010 | Xaroon_x | null | 2026-05-20 | null | |
gpt-image-1969-festival-crowd | GPT Image 2.5 | text-to-image | realism | [
"historical",
"crowd",
"photorealism"
] | Festival crowd in Bethel, New York, August 1969 | 1969年8月纽约贝瑟尔音乐节人群 | Create a realistic outdoor crowd scene in Bethel, New York on August 16, 1969.
Photorealistic, period-accurate clothing, staging, and environment. | Naming only a precise place and date lets the model infer era clothing, hairstyles, stage scaffolding and film colour without listing them. Asking for period accuracy pushes it toward documentary realism rather than costume pastiche. | 只给出精确的地点和日期，模型就能自行推断当时的服饰、发型、舞台脚手架和胶片色调，无需逐项列出。要求符合时代细节，能让画面更像纪实摄影，而不是戏服式的模仿。 | OpenAI image prompting guide | https://developers.openai.com/api/docs/guides/image-prompting#use-historical-and-real-world-context | OpenAI | null | null | null | |
gpt-image-1980s-fashion-editorial-red-suit-portrait | GPT Image 2.5 | image-edit | portrait | [
"1980s",
"fashion-editorial",
"film-grain",
"reference-image",
"vintage"
] | 1980s editorial portrait in a red suit | 红色西装的八十年代时尚杂志人像 | Create a vertical, ultra-photorealistic portrait of the same clearly adult woman from the reference image, preserving her facial identity, facial structure, skin tone, hair characteristics, and overall appearance.
Give her an authentic 1980s fashion-editorial look with a confident, effortless attitude. She is posing c... | A reference photo locks the face while a detailed list of period outfit pieces and accessories sets the 80s look. Studio backdrop, warm analog light, film grain and lens softness give the magazine feel, and a closing list of negatives keeps modern elements out. | 用参考图锁定面部身份，再逐项写出年代服装与配饰来定义八十年代造型。影棚背景、暖色模拟光、胶片颗粒和镜头柔化营造杂志质感，结尾的排除清单避免现代元素混入。 | @mehvishs25 on X | https://x.com/mehvishs25/status/2098090489811746853 | Meem | 74 | 2026-09-10 | null | |
gpt-image-1990s-anime-character-styles-catalog | GPT Image 2 | text-to-image | infographic | [
"anime",
"infographic",
"character-design",
"style-catalog"
] | 1990s anime character styles catalog with 16 labeled archetypes | 1990年代动漫角色风格图鉴，16种原型标注网格 | Anime character styles are incredibly diverse and researching them one by one can get exhausting fast. I'm using a simple GPT Image 2 prompt that turns this process into a visual catalog by generating infographic-style posters with different anime character archetypes, visual languages and era-specific styles all in a ... | Naming the era, the language, a labeled grid layout and one archetype per panel gives a consistent, scannable reference sheet from a single sentence. Specifying the 16:9 ratio fixes the canvas so the grid fills the frame. | 一句话写明年代、语言、标注网格以及每格一个原型，就能得到结构统一、便于对照的参考图。指定 16:9 比例可以让网格填满画面。 | @aimikoda on X | https://x.com/aimikoda/status/2058152463031488981 | Kōda | 590 | 2026-05-23 | null | |
gpt-image-1994-infomercial-crashed-saucer-frame | GPT Image 2.5 | text-to-image | poster | [
"infomercial",
"retro-tv",
"scanlines",
"ufo",
"typography"
] | Retro 1994 infomercial frame selling a crashed saucer | 1994年电视购物画面里兜售坠毁的飞碟 | a frame grab from a 1994 infomercial, standard definition video with scanlines. A host in a teal blazer, a battered saucer on a rotating plinth with Antarctic ice frozen around one leg, a price starburst, a phone number, and three lines of legal small print. | Naming the medium and era first (1994 infomercial, standard definition, scanlines) sets the whole look. A short list of concrete props then drives the layout: host, plinth, price starburst, phone number and legal small print. | 开头先点明媒介和年代（1994年电视购物、标清、扫描线），整体质感就定下来了。再用一串具体元素列出主持人、转台、价格爆炸贴、电话号码和法律小字，直接决定版面构成。 | @ImagineArt_X on X | https://x.com/ImagineArt_X/status/2097520474226147462 | ImagineArt_X | null | 2026-09-09 | null | |
gpt-image-1998-coastal-summer-travel-editorial | GPT Image 2.5 | text-to-image | portrait | [
"travel-editorial",
"film-photography",
"1990s",
"coastal",
"fashion"
] | Late 1990s coastal summer travel portrait | 九十年代末海岛夏日旅行杂志人像 | Create an ultra-realistic late-1990s summer travel editorial featuring a stylish young woman and a stylish young man in two separate portrait frames with the exact same visual theme and color grading.
The setting is a dreamy coastal destination inspired by Jeju Island and the Greek islands, with crystal-blue sea, rock... | Fixes an era, a real-world location mood and a film look (Kodak grain, faded highlights, golden light) so the whole frame reads as one period magazine shoot. Detailed outfit, pose and setting lists plus a negative list keep the result candid and realistic. | 先定年代、地域氛围和胶片质感（柯达颗粒、褪色高光、金色阳光），让画面统一成一本旧杂志的拍摄风格。服装、姿态、场景逐项写清，再用排除清单保持自然写实。 | @AIwithWania on X | https://x.com/AIwithWania/status/2098773641303310647 | Wania | 118 | 2026-09-12 | null | |
gpt-image-2003-digital-camera-family-snapshot | GPT Image 2 | text-to-image | realism | [
"family-photo",
"nostalgia",
"digital-camera",
"vintage",
"snapshot"
] | 2003 Digital Camera Family Snapshot in Courtyard | 2003年数码相机拍摄的小区庭院全家福 | Generate a photo from 2003, shot with a digital camera, of five-year-old me with my mom and dad in the courtyard of our residential compound in China. | Naming the year and the capture device makes the model reproduce period traits such as soft focus, flat colour and an orange date stamp. A concrete setting, a residential courtyard in China, with a simple family grouping keeps the scene believable. | 点明年份和拍摄设备，模型就会还原那个年代的柔焦、平淡色彩和橙色日期戳。再给出具体场景，即中国小区庭院，配上简单的一家三口构图，画面显得真实可信。 | @pangyusio on X | https://x.com/pangyusio/status/2046991223395950810 | pangyusio | null | 2026-04-22 | null | |
gpt-image-2027-watercolor-calendar | GPT Image 2 | image-edit | poster | [
"calendar",
"watercolor",
"character-consistency",
"korean-illustration"
] | 2027 watercolor calendar with one seated man per month | 2027 水彩日历海报，每月一位坐姿男子 | {
"prompt": "Create a premium 2027 desktop calendar poster in A4 vertical format featuring the same young man from the reference image, illustrated in a Korean hand-drawn illustration style. Design a 3x4 monthly grid layout from January to December 2027. Each month should feature the character sitting in a different ... | The same character is held consistent across twelve panels by fixing the face, curly hair, glasses and beard in every scene. A 3x4 month grid with seasonal doodle borders on a clean white ground keeps the layout orderly. | 在十二个月份场景中固定同一角色的脸部、卷发、眼镜和胡须，让人物保持一致。3x4 月历网格配合季节性涂鸦边框和留白背景，使版式清晰有序。 | @Fujimoto_hina on X | https://x.com/Fujimoto_hina/status/2055895559429644702 | Aegon | 516 | 2026-05-17 | null | |
gpt-image-3d-chibi-doodle-sticker-portrait | GPT Image 2 | image-edit | edit | [
"chibi",
"doodle",
"sticker",
"portrait",
"reference-photo"
] | Portrait surrounded by 3D chibi doodle stickers | 被3D Q版贴纸和涂鸦环绕的人像 | Use the original face from the reference photo. Background same as the photo. Warm, soft, clean lighting with soft shadows. Surround it with several 3D cute style mini chibi versions of themselves, keeping the original facial features. The chibi characters have various poses and expressions: cheerful jumping, waving, s... | The prompt locks the original face and background first, then adds mini 3D chibi copies with listed poses and expressions. White handwritten doodles, a sticker outline and short casual words finish the pastel look. | 提示词先锁定原脸和原背景，再加入若干指定姿势与表情的3D Q版小人。白色手绘涂鸦、贴纸描边和随手写的短词完成粉彩风格。 | @ChillaiKalan__ on X | https://x.com/ChillaiKalan__/status/2051209193336545713 | K | null | 2026-05-04 | null | |
gpt-image-90s-apartment-scene-reference-board | GPT Image 2 | text-to-image | architecture | [
"scene-reference",
"production-design",
"cinematic",
"interior",
"35mm"
] | 90s apartment night scene reference board | 90年代公寓夜景场景参考板 | {
"type": "scene reference board — 90s apartment living room, cinematic night",
"style": "cinematic film photography, 35mm grain, warm amber shadow fill, deep chiaroscuro lighting, hyper-detailed interior, production design reference quality",
"layout": {
"main_panel_center_left": {
"label": "CAMERA A —... | A JSON layout fixes the panel positions: two wide camera views, a six tile prop strip and a footer line of colour temperature and lens. Naming each camera angle and listing the props per tile keeps the room consistent across every panel. | 用 JSON 固定版面：两个宽幅机位、六格道具特写条和色温镜头页脚。逐一写明机位角度并按格列出道具，让各画格里的房间保持一致。 | @Iancu_ai on X | https://x.com/Iancu_ai/status/2051287273581203888 | Iancu_ai | null | 2026-05-04 | null | |
gpt-image-abbey-road-wire-line-art-poster | GPT Image 2 | image-edit | photo-to-art | [
"line-art",
"split-layout",
"minimalist",
"concept-poster"
] | Abbey Road crossing photo above, wire line art below | 上半为披头士过斑马线照片，下半为线条人像 | 请将我上传的每一张照片分别制作成一张独立的高级设计海报，不多图拼接，每张照片单独输出。整体采用3:4竖版构图，上下两个区域高度严格1:1，各占画面 50%。
上半部分保留原始照片，保持主体结构、真实质感、自然光影和原有色彩氛围，仅进行轻微高级摄影调色，使其具有杂志摄影、艺术出版物和公共议题影像的克制质感。为适配画幅，可自然扩展天空、地面或环境背景，但不得拉伸、扭曲或改变主体。
下半部分提取照片中最具识别性的主体、轮廓、姿态与叙事关系，重构为一幅极简、克制、带有公共议题气质的观念线条插画。不机械描边，不复制背景、光影和复杂细节，而是提炼原图中的方向、连接、矛盾、压力与转机，将其压缩成一组连续的黑色手绘线条。线条如被拉紧的细铁丝、路... | The 3:4 frame is split into two equal halves, keeping the photo intact on top and rebuilding the figures below as continuous, looping ink lines that enter from the side edges. Small monospaced text in the margin keeps the cream field quiet and minimal. | 将3:4画幅严格分为上下两半，上半保留原照片，下半用连续的墨线重构人物与行走关系，线条从两侧进入、缠绕回折。细小的等宽字点缀在留白处，使冷白底色保持克制。 | @xiaoxiaodong01 on X | https://x.com/xiaoxiaodong01/status/2089959872876228689 | 小小东 | null | 2026-08-19 | null | |
gpt-image-adriano-emperor-doodle-football-poster | GPT Image 2 | text-to-image | poster | [
"football",
"poster",
"doodle",
"typography"
] | Adriano career stats doodle football poster | 阿德里亚诺涂鸦风足球生涯数据海报 | Create a scrapbook doodle-style football poster of [PLAYER_NAME].
AI auto-generates realistic career stats (club and national team).
Main photo: realistic, unchanged ([PLAYER_NAME] in action or iconic pose).
Doodles and handwritten notes: white and [PRIMARY_COLOR] neon ink (no warm colors unless specified), arrows, ... | Keep the real player photo intact and wrap him in a glowing neon outline, then surround him with doodles and boxed stat panels. Fixed sections for club, highlights and national team make the layout easy to scan. | 保持真人照片不变，并在人物外加一圈发光霓虹轮廓，周围配以涂鸦和数据框。俱乐部、高光与国家队三个固定区块让版面一眼就能读完。 | @ryanpp27 on X | https://x.com/ryanpp27/status/2048602248524214542 | ryanpp27 | null | 2026-04-27 | null | |
gpt-image-aftermath-survival-streetwear-poster | GPT Image 2 | image-edit | poster | [
"fashion",
"editorial",
"post-apocalyptic",
"streetwear"
] | Man in survival streetwear beside wrecked car poster | 身穿末日街头装的男子倚靠废弃汽车海报 | A contemporary editorial fashion poster inspired by the Post-Apocalyptic Street aesthetic, blending futuristic survival themes with modern high-end streetwear fashion. The overall atmosphere should feel cinematic, dramatic, and visually immersive, portraying a once-thriving metropolitan city that has fallen into decay ... | Pairing an editorial fashion frame with a ruined city lets the wardrobe stay sharp while debris, dust and signage carry the story. Naming the fabrics, weathering and display typeface gives concrete controls for each layer. | 将时尚大片构图与废墟城市并置，服装保持清晰，碎石、尘埃和标识承担叙事。明确面料、做旧程度和展示字体，为每一层提供具体控制。 | @Xaroon_x on X | https://x.com/Xaroon_x/status/2071179557731008802 | Zar⭕on | 193 | 2026-06-28 | null | |
gpt-image-ai-filmmaker-monochrome-panel-editorial-collage | GPT Image 2.5 | image-edit | collage | [
"editorial",
"collage",
"monochrome",
"identity-consistency",
"fashion"
] | Monochrome panel editorial collage with color cutout | 黑白分格编辑拼贴搭配彩色全身抠图 | Use my attached/reference portrait as the strict facial identity reference. Keep my real facial structure, skin tone, eyes, nose, lips, hairline and overall likeness highly consistent across every appearance. Do not beautify me into a different person.
Create a premium modern editorial collage portrait of a female AI ... | Four stacked black and white panels each get a distinct pose and framing, while one full color figure overlaps them for depth. A strict identity consistency clause, explicit outfit details and a no text rule keep the layout clean. | 四个黑白分格各自指定不同姿势和景别，再用一个彩色全身人像叠在上层营造层次。严格的人物一致性条款、详细的服装描述和禁止文字的要求让版面保持干净。 | @FutureVibesAi on X | https://x.com/FutureVibesAi/status/2104874557345841520 | Future Vibes AI - Testing & Educating | 158 | 2026-09-29 | null | |
gpt-image-ai-fitness-coach-analysis-poster | GPT Image 2 | image-edit | infographic | [
"fitness",
"poster",
"dashboard-ui",
"glassmorphism",
"selfie"
] | AI fitness analysis poster from a gym selfie | 根据健身房自拍生成的 AI 健身分析海报 | I will upload a gym mirror selfie/body photo.
Create an ULTRA PREMIUM AI FITNESS ANALYSIS POSTER that looks like a luxury futuristic wellness brand advertisement mixed with a high-end fitness app dashboard.
IMPORTANT:
The uploaded face and identity must remain EXACTLY THE SAME.
Do not change:
face structure
eyes
j... | The prompt locks the uploaded face and identity, then lists every dashboard section, food card and exercise panel by name, so the model fills a dense glassmorphism layout around a centered subject. A bright luxury palette with explicit bans on dark gym looks keeps the poster clean and premium. | 提示词先锁定上传照片的人脸与身份，再逐一列出每个分析模块、食物卡片和训练面板，让模型围绕居中人物排出密集的玻璃拟态版面。明确的明亮奢华配色并禁用暗黑健身房风格，使整体干净高级。 | @frametheory058 on X | https://x.com/frametheory058/status/2054461225413870014 | Nexora | 454 | 2026-05-13 | null | |
gpt-image-ai-memory-anime-self-infographic | GPT Image 2 | text-to-image | infographic | [
"infographic",
"anime",
"self-portrait",
"chatgpt-memory",
"personalized"
] | Anime cartoon infographic about the user, from memory | 根据聊天记忆生成的动漫卡通个人信息图 | Ask ChatGPT knowing everything it knows about you to generate an image of yourself with GPT Image 2
prompt: Knowing everything you know about me, make an infographic of me. Make a cartoon anime style image.
I'd love to see what you got. | The prompt hands the content to ChatGPT memory, so the model fills in a personal bio, values and hobbies without any details typed in. Naming two formats, infographic and cartoon anime, gets a layout of labeled panels around a character portrait. | 提示词把内容交给 ChatGPT 的记忆，不用手写任何个人细节，模型自己补全简介、价值观和爱好。同时指定信息图和卡通动漫两种形式，得到围绕角色肖像的分区标注版面。 | @icreatelife on X | https://x.com/icreatelife/status/2047042187683066100 | Kris Kashtanova | 209 | 2026-04-22 | null | |
gpt-image-ai-model-portrait-comparison | GPT Image 2 | text-to-image | realism | [
"comparison",
"street-portrait",
"photorealistic",
"social-media"
] | Two AI image models compared as street portraits | 两款 AI 图像模型的街头人像对比 | Goal: Create a clean side-by-side comparison graphic showing two AI image model outputs as realistic street portraits.
Canvas: Horizontal 3:2 social media comparison image, white background, high resolution. Use a simple two-column layout with a narrow vertical white divider between the image panels.
Layout: Top head... | One detailed subject description is repeated in both panels so the comparison is about the model rather than the sitter. Camera feel, lens, motion blur and a fixed element count with a white divider keep the layout controlled and the two photos consistent. | 两个画面重复同一位女性的详细描述，让对比聚焦于模型本身而非人物。镜头感、焦段、运动模糊，以及固定的元素数量和白色分隔线，使版式可控、两张照片风格一致。 | @TimJayas on X | https://x.com/TimJayas/status/2086455464930459760 | Tim Jayas | null | 2026-08-09 | null | |
gpt-image-airpods-model-case-infographic | GPT Image 2 | image-edit | infographic | [
"airpods",
"infographic",
"reference-identity",
"product-ad"
] | Model holding AirPods case in feature infographic | 模特手持 AirPods 充电盒的功能信息图 | High-end e-commerce infographic featuring Apple AirPods (latest generation). A young woman (same face and identity as the provided reference image) is the central subject, shown in a clean studio setting with soft, diffused lighting. She is wearing AirPods in her ears, naturally integrated into the scene, and holding t... | Naming the product, a reference image for identity, and a fixed set of labelled callouts gives an ad layout with clear negative space. Stating the palette, lighting and 1:1 ratio keeps the result consistent in a commerce ad style. | 明确产品名称、参照图身份锁定，以及固定的图标标注和指引线布局，使画面呈现清晰的广告式构图与留白。指定配色、布光和 1:1 比例，让输出风格统一。 | @mehvishs25 on X | https://x.com/mehvishs25/status/2049513437542461551 | Meem | 227 | 2026-04-29 | null | |
gpt-image-alphabet-object-grid-a | GPT Image 2 | text-to-image | infographic | [
"grid",
"alphabet",
"objects",
"layout",
"text-rendering"
] | Ten by ten grid of objects starting with A | 十乘十网格，展示以A开头的物品 | Create an image with a 10x10 grid of objects whose names all start with the letter A | A fixed grid size plus a single naming rule gives the model a clear layout and a checkable constraint. Short and open ended, it tests how well the model keeps labels and objects aligned across 100 cells. | 固定的网格尺寸加上单一的命名规则，给模型明确的版式和可检查的约束。提示词很短且开放，用来测试模型在100个格子里能否让标签与物体保持对应。 | @umesh_ai on X | https://x.com/umesh_ai/status/2046510988367945983 | umesh_ai | null | 2026-04-21 | null | |
gpt-image-amigurumi-doll-from-photo-diorama | GPT Image 2.5 | image-edit | photo-to-art | [
"amigurumi",
"crochet",
"plush",
"diorama",
"photo-to-doll"
] | Crocheted amigurumi doll in a cozy diorama | 温馨迷你场景中的钩织娃娃 | Turn your photo into a cozy hand-crocheted amigurumi doll using GPT Image 2.5
Prompt: Transform the uploaded photo into a hand-crocheted amigurumi style doll. Soft chunky yarn with visible stitches, oversized rounded head and tiny body, embroidered facial features with gentle blush, pastel color palette, cozy handmade... | Names the yarn medium and its visible stitches, then fixes the proportions with an oversized head and tiny body. Embroidered features, a pastel palette and a miniature diorama setting with warm light keep the photo conversion consistently handmade. | 先点明毛线材质和可见针脚,再用大头小身比例固定造型。刺绣五官、粉彩配色、迷你场景和暖光,让照片转换保持统一的手工质感。 | @MrDasCreates on X | https://x.com/MrDasCreates/status/2115351638353613255 | Mr Das | 32 | 2026-10-28 | null | |
gpt-image-ancient-pyramid-construction-blueprint-diorama | GPT Image 2 | text-to-image | architecture | [
"miniature",
"diorama",
"blueprint",
"ancient-architecture",
"template"
] | Pyramid construction diorama on architectural blueprints | 铺在建筑蓝图上的金字塔施工微缩场景 | A highly detailed miniature diorama of an ancient civilization under construction, displayed directly on top of large rolled-out architectural blueprints spread across a realistic wooden drafting table. The scene features iconic historical architecture from [CIVILIZATION OR LOCATION], including partially completed monu... | A bracketed civilization slot makes the template reusable, while the miniature terrain is told to blend into the printed plans beneath it. Window sunlight, shallow depth of field and a listed set of drafting props give a tactile scale model look. | 用方括号文明占位符让模板可复用,并要求微缩地形与下方印刷图纸融为一体。窗边暖光、浅景深和一组绘图道具共同营造出触感真实的比例模型质感。 | @Naiknelofar788 on X | https://x.com/Naiknelofar788/status/2054221110372405534 | simeon-sanai | 426 | 2026-05-12 | null | |
gpt-image-anime-airship-nine-scene-storyboard | GPT Image 2 | text-to-image | comic | [
"storyboard",
"anime",
"airship",
"infographic-layout",
"cinematic"
] | Nine Scene Anime Airship Storyboard Sheet | 九场景动漫飞艇分镜头设计稿 | 一份专业的动漫院线分镜头（16：9），标题为《天空之击》，副标题为“战士与圣石”。干净的浅米色背景，圆角矩形面板排列成三列网格，共 9 个场景。每个场景面板显示：橙色“场景 XX”标签、时间码、场景标题，然后是两个并排的画面，分别标有“开始帧”→“结束帧”，通过白色箭头连接，下面有注释（镜头/对白/特效/动作）。艺术风格：清晰的卡通渲染 新海诚的×工作室机智，×进击的巨人——柔和的画风光线、立体云朵、温暖的金色阳光、电影般的广角镜头、动态动作姿势、浓郁的天蓝和翡翠绿调，搭配黄铜和铜色点缀。需要描绘的场景：1. 广角景——一艘雄伟的蒸汽朋克飞艇，黄铜螺旋桨和飘扬的帆，在高空穿透积云海洋（慢速无人机推入）2.低角度特写一名女战士的靴子和... | The prompt fixes the sheet layout first, with a three column grid and a start frame to end frame pair in each panel. It then lists every scene with its camera move, which keeps the story beats and notes consistent across all nine panels. Footer blocks for characters, tone, music plan and tech specs make it read like a ... | 提示词先定下三列网格和每格开始帧到结束帧的版式，再逐个场景写明镜头运动，让九个画面的剧情节奏与注释保持一致。底部的角色、基调、音乐计划和技术规格栏让整张图像真实的制作文档。 | @0xluffy_eth on X | https://x.com/0xluffy_eth/status/2047224465097134172 | 路飞 🏴☠️ AI 研究员🧐 | null | 2026-04-23 | null | |
gpt-image-anime-angel-girl-close-up-portrait | GPT Image 2.5 | text-to-image | illustration | [
"anime",
"angel",
"close-up",
"bokeh",
"backlight"
] | Anime Angel Girl Close-up Portrait | 动漫天使少女特写肖像 | A high-resolution anime-style close-up portrait of a young girl with long, flowing white hair. The art style is soft and ethereal, emphasizing the texture of individual hair strands to create a silky, smooth appearance rather than blocky clumps. She has large, expressive eyes that are a gradient of teal and light blue,... | The prompt asks for individual silky hair strands instead of blocky clumps and describes the eye gradient and highlights in detail. Strong backlighting and a bokeh foliage background give the soft, airy mood. | 提示词要求发丝根根分明、柔滑而非块状，并细致描述眼睛的渐变色与高光。强烈逆光加上虚化的树叶光斑背景，营造出柔和通透的氛围。 | @magu_ai_h on X | https://x.com/magu_ai_h/status/2099072433664004307 | magu@ai | null | 2026-09-13 | null | |
gpt-image-anime-bob-portrait-sketch-collage | GPT Image 2 | text-to-image | illustration | [
"anime",
"fashion-illustration",
"sketch-collage",
"full-body",
"magazine-cover"
] | Anime Fashion Portrait with Sketch Collage Backdrop | 动漫风时尚人像与手绘素描拼贴背景 | Semi-realistic anime-inspired portrait of a stylish young Korean woman with a short glossy black bob tucked behind one ear, delicate round-frame glasses, and a gentle confident expression. She wears an oversized pastel lilac blouse with rolled sleeves paired with a flowing ivory A-line midi skirt. Full-body composition... | The prompt fixes the subject first with hair, glasses, outfit and full-body pose, then describes the backdrop as a scatter of monochrome sketches, manga panels and handwritten notes. Naming the ink and pencil texture, cel shading and a bright white ground keeps the colored figure separate from the grey collage. | 提示词先用发型、眼镜、服装和全身姿态锁定主体，再把背景写成散落的单色素描、漫画分格和手写笔记。点明墨线铅笔质感、赛璐璐阴影和纯白底，让彩色人物与灰调拼贴清楚分层。 | @john_my07 on X | https://x.com/john_my07/status/2062204692072198580 | Johnn | 316 | 2026-06-03 | null | |
gpt-image-anime-candid-snapshot-grid-nine-panels | GPT Image 2.5 | image-edit | collage | [
"anime",
"grid",
"candid",
"character-consistency",
"slice-of-life"
] | Nine candid anime snapshots of one character | 同一角色的九宫格动漫随手拍 | Using the provided reference image as the character base, generate a 3x3 grid of 9 distinct square panels. Each panel should depict the same character in different candid, slice-of-life scenarios, styled to look like casual snapshots or social media posts.
Layout: A perfectly aligned 3x3 grid with thin white borders s... | A reference image anchors the character, while a numbered panel by panel list gives each cell its own framing, such as motion blur, a finger over the lens, a silhouette or a rear view. Amateur photo cues like grain and mixed daylight, warm indoor and dusk light, plus a locked outfit constraint, keep the set cohesive. | 以参考图锁定角色,再按编号逐格指定构图,如动态模糊、手指遮镜头、剪影和背影。颗粒感与多种光线营造业余随手拍质感,并用统一服装约束保证整组一致。 | @8co28 on X | https://x.com/8co28/status/2099164852560539975 | 852話(hakoniwa) | null | 2026-09-13 | null | |
gpt-image-anime-character-line-art-setting-sheet | GPT Image 2 | image-edit | character | [
"character-sheet",
"line-art",
"anime",
"reference-image",
"japanese"
] | Elf Girl Anime Character Line Art Sheet | 精灵少女动画角色线稿设定资料 | 参照画像をキャラクターデザインのベースとして使用し、「商業アニメ制作現場で使用される、プロ仕様のキャラクター線画設定資料」を描く。
キャラクターの顔、髪型、髪色、瞳、輪郭、体型、シルエット、衣装デザイン、頭身、全体の印象を正確に維持すること。
キャラクターのアイデンティティを絶対に崩さない。
別キャラクターとして描き直さない。
完成イラストではない。
コンセプトアートではない。
ポスターではない。
目的は、
アニメーター・作画監督・動画担当へ向けた、
作画共有用の設定資料。
画面は、
白背景の設定資料レイアウト。
複数の線画を、
資料ページとして整理して配置する。
以下を含める：
・正面顔
・斜め顔
・横顔
・表情差... | The prompt names the production use, anime staff as the audience, and lists every required panel, from face angles to shoes and accessories. It bans color and rendering, and asks for handwritten notes, arrows and scan texture so the page reads as a real 90s to 2000s setting book. | 提示词点明用途与读者是动画制作人员，并逐项列出脸部角度到鞋子配饰的全部分栏。它禁止上色与渲染，同时要求手写批注、箭头和扫描质感，让页面像90到2000年代的真实设定集。 | @Eris_Create_Lab on X | https://x.com/Eris_Create_Lab/status/2058807276828565803 | Eris Create Lab | null | 2026-05-25 | null | |
gpt-image-anime-character-reference-sheet-kana | GPT Image 2 | text-to-image | character | [
"character-sheet",
"anime",
"turnaround",
"json-prompt",
"layout"
] | Anime Character Reference Sheet With Turnaround | 带三面图的动漫角色设定表 | {
"type": "anime character reference sheet",
"style": "clean Japanese character design board, elegant watercolor and ink illustration, soft neutral paper background, thin decorative borders, editorial layout, refined fantasy-fashion concept art",
"character": {
"name": "{argument name=\"character name\" defau... | A JSON prompt lists each layout section with its position, item count and exact Japanese labels, so the sheet comes out as a tidy grid. Named argument placeholders for hair, jacket and dress colors make it easy to swap the character. | JSON 提示词为每个版块写明位置、数量和日文标签，生成的设定表排版整齐。头发、外套和裙子颜色用命名参数占位，方便替换角色。 | @REd8358 on X | https://x.com/REd8358/status/2047465795005849818 | あーる@AI Artgrapher | null | 2026-04-24 | null | |
gpt-image-anime-character-rooftop-livestream-screenshot | GPT Image 2 | image-edit | character | [
"anime",
"livestream",
"smartphone-ui",
"mixed-reality",
"reference-image"
] | Anime Elf Maid in Rooftop Livestream Screenshot | 屋顶直播截图中的动漫精灵女仆 | 参照画像をキャラクターデザインのベースとして使用し、「アニメキャラクターが現実世界のライブ配信に現れてしまった瞬間」を描く。
キャラクターのアニメイラストとしての空気感を正確に維持する。
キャラクターのアイデンティティを絶対に崩さない。
実写人間化しない。
キャラクター部分はアニメイラスト表現を維持する。
縦型スマートフォン配信画面。
TikTok LIVE、
YouTube Live、
Instagram Live、
一般的な縦型ライブ配信風UI。
背景は固定しない。
現実的な都市空間、
または生活感のある現実空間をランダムに生成する。
例:
ビル屋上、
コンビニ前、
ゲームセンター、
駅前、
住宅街、
雑居ビルの階段... | A reference image locks the character design while a photoreal hand enters the foreground with shallow depth of field. Smartphone artifacts such as sensor noise, JPEG compression and handheld shake, plus bans on cinematic polish, make it read as an accidental live capture. | 用参考图锁定角色设计,前景加入浅景深的真人手部。手机传感器噪点、JPEG压缩和手持抖动,再加上禁止电影感与过度打磨,让画面像偶然截下的直播画面。 | @Eris_Create_Lab on X | https://x.com/Eris_Create_Lab/status/2057294055911116941 | Eris Create Lab | null | 2026-05-21 | null | |
gpt-image-anime-characters-real-living-room | GPT Image 2 | image-edit | collage | [
"mixed-media",
"reference-face",
"anime-crossover",
"interior"
] | Anime characters inside a real living room | 真实客厅里的动漫角色 | A high-fidelity, wide-angle interior shot captures a surreal, mixed-media composition within a modern living room. It features a man resembling the face in the reference photo—using an uploaded face as a reference—wearing a dark green sweatshirt, sweatpant and white sneakers, sitting centrally on a plush light gray sof... | A reference face anchors the real man while each anime character is placed by its exact spot and pose, which keeps the photoreal room and the cel shaded figures in clearly separate layers. Naming the light direction through sheer curtains and the landing spots on sofa, table and pouf keeps the layout consistent. | 用参考人脸锁定真人主体，再逐一指定每个动漫角色的具体位置和姿势，让写实客厅与卡通角色保持清楚的分层。明确窗光方向以及沙发、矮桌、布艺凳上的落点，使整体布局和光线保持一致。 | @codewithhajra on X | https://x.com/codewithhajra/status/2072304737949474960 | H A J R A | 112 | 2026-07-01 | null | |
gpt-image-anime-classroom-losing-heroines-key-visual | GPT Image 2 | text-to-image | poster | [
"anime",
"classroom",
"key-visual",
"japanese-text"
] | Anime classroom key visual with three students and posters | 动漫教室主视觉，三名学生与海报 | Goal: Create a cinematic anime key visual for a high-school romantic comedy titled 負けヒロインが多すぎる!, set in a classroom with gentle daylight and a slightly melancholic but humorous mood.
Canvas: Wide 16:9 composition, 1200×675 style framing, like a polished TV anime promotional still. Use soft realistic lighting, detailed... | The prompt fixes every element by count (3 students, 3 desk items, 2 posters, 3 captions) so nothing extra appears, and it sets a desk-height camera angled slightly upward. Writing the exact Japanese title, poster and caption text into the prompt keeps the on-image typography readable. | 提示词逐项限定数量（3名学生、3件桌面物品、2张海报、3条手写注释），避免画面出现多余元素；同时指定桌面高度、略微仰拍的机位，并直接写出日文标题、海报和台词，使画面文字保持可读。 | @kiyoshi_shin on X | https://x.com/kiyoshi_shin/status/2090292842992488516 | 新清士@AIコンテンツ開発者 | null | 2026-08-20 | null | |
gpt-image-anime-driver-split-screen-mountain-pass | GPT Image 2 | image-edit | illustration | [
"split-screen",
"racing-game",
"anime",
"driver-view",
"reference-character"
] | Anime driver split screen on mountain pass | 峠道漂移的动漫车手分屏画面 | 参照画像のキャラクターを完全に維持する。キャラクターの顔、髪型、髪色、瞳、耳、体格、骨格、肩幅、首の太さ、腕のシルエット、年齢感、性別表現、雰囲気を維持すること。画面は左右二分割されたスプリットスクリーン構成。左右は同じ瞬間を異なるカメラで撮影している。左右は同じ車両、同じコーナー、同じ時間軸である。【車両】左右画面に登場する車両は同一車両。車種：スポーツカーまたはチューニングカー。左右画面は同一車両を異なる視点から映している。【左画面】ドライバーカメラ。車内から見た斜めローアングル腰上ショット。キャラクターは車両を運転している。両手でハンドルを握る。コーナリング中。視線は進行方向へ向けられている。真剣な表情。集中している。身体には... | The prompt locks the character to a reference image, then splits the frame into a cockpit view and a rear chase camera. It states repeatedly that both halves show the same car, corner and moment, which keeps the two views consistent. | 提示词先用参考图锁定角色，再把画面分成车内视角和后方追尾视角。它反复强调左右是同一辆车、同一弯道、同一瞬间，让两个镜头保持一致。 | @Eris_Create_Lab on X | https://x.com/Eris_Create_Lab/status/2069652470331228244 | Eris Create Lab | null | 2026-06-24 | null | |
gpt-image-anime-enchanted-forest-inner-landscape | GPT Image 2 | text-to-image | illustration | [
"anime",
"double-exposure",
"watercolor",
"forest",
"typography"
] | Anime Girl Enchanted Forest Double Exposure | 动漫少女心象森林双重曝光 | Create a luminous vertical anime illustration visualizing a girl's “inner landscape” through double exposure. Canvas: tall portrait 9:16, soft white background with airy watercolor splashes and glowing particles. Subject: one smiling anime girl, half-body seated pose, turned slightly toward the viewer, long flowing fre... | The prompt blends the hair and dress with a forest scene through double exposure, and fixes the layout by naming where the palm, stream and flowers sit. Exact counts for the butterfly, headline and profile lines, plus quoted text strings, keep the typography overlay accurate. | 提示词用双重曝光让头发和裙子融入森林场景,并指明手掌、溪流和花朵的位置来固定构图。蝴蝶、标题和资料行都给出确切数量与原文字符串,让文字排版更准确。 | @ArtistaNozomu on X | https://x.com/ArtistaNozomu/status/2057099863494652378 | のぞむ＊AIイラスト | null | 2026-05-20 | null | |
gpt-image-anime-gender-swap-before-after-poster | GPT Image 2 | text-to-image | poster | [
"anime",
"magazine-cover",
"gender-swap",
"before-after",
"japanese-typography"
] | Anime Gender Swap Before After Magazine Poster | 动漫性别反转前后对比杂志封面海报 | Goal: Create a polished anime magazine-cover style “gender swap before/after” poster. The main subject is the transformed After character: an extremely handsome teenage boy who preserves the soft facial resemblance of the original character, with light ash-blonde tousled medium-short hair, sparkling blue eyes, pale lum... | The prompt separates the canvas into a small Before card and a dominant After portrait, giving exact counts for cards and stamps plus the Japanese text strings to keep typography legible. A restricted blue and white palette with watercolor lighting keeps the layout cohesive. | 提示词把画面拆成小尺寸的前图卡片和占主体的后图人像,并明确卡片与印章的数量以及日文文案原文,保证排版文字清晰。蓝白限定配色加水彩光效让整体统一。 | @ArtistaNozomu on X | https://x.com/ArtistaNozomu/status/2085314580130197811 | のぞむ＊AIイラスト | null | 2026-08-06 | null | |
gpt-image-anime-girl-streetwear-sixteen-pose-sheet | GPT Image 2.5 | text-to-image | character | [
"anime",
"character-sheet",
"pose-sheet",
"streetwear",
"grid"
] | Anime Girl Streetwear Sixteen Pose Sheet | 动漫女孩街头风十六姿势设定表 | A character turnaround sheet featuring a cute anime girl in a dynamic streetwear outfit, arranged in a 4x4 grid of 16 distinct poses. The character has messy, shoulder-length blue-grey hair with subtle pink highlights and large, expressive blue eyes. She wears an oversized bright blue bomber jacket over a white hoodie,... | The prompt fixes one detailed outfit and hair description, then fixes a 4x4 numbered grid on a white background. A numbered list of 16 short pose lines gives each panel its own action while keeping the character consistent. | 提示词先锁定统一的服装与发型描述，再限定带编号的4x4网格和白色背景。随后用16条简短的编号姿势逐格指定动作，让角色在各格中保持一致。 | @muku_sns on X | https://x.com/muku_sns/status/2102743895843156425 | むく | AIアートのプロ | null | 2026-09-23 | null | |
gpt-image-anime-idol-dance-storyboard-grid | GPT Image 2.5 | text-to-image | character | [
"anime",
"storyboard",
"dance",
"idol",
"character-sheet"
] | Anime Idol Dance Storyboard Grid | 动漫偶像舞蹈分镜网格 | A storyboard grid of 16 numbered frames (4x4 layout) showing a cute anime idol girl performing a continuous dance sequence. The character has shoulder-length wavy brown hair with small blue and white flower accessories, large expressive eyes, and wears an elaborate idol outfit: a white short-sleeved top with blue plaid... | A detailed outfit description is fixed once, then 16 numbered frames each name a pose, camera angle and effect such as motion arrows, hearts or confetti. Layout rules for a 4x4 grid, panel borders, number labels and a shared floor keep the character consistent across frames. | 先一次性写定详细的服装设定,再用 16 个编号画格逐一指定姿势、视角和动态箭头、爱心、彩纸等效果。4x4 网格、画格边框、编号标签和统一地面的版面要求让角色在各帧保持一致。 | @OiiOii_AI_JP on X | https://x.com/OiiOii_AI_JP/status/2099738831755092209 | OiiOii AI JP | null | 2026-09-15 | null | |
gpt-image-anime-pop-art-street-style-conversion | GPT Image 2 | image-edit | photo-to-art | [
"pop-art",
"street-art",
"anime",
"halftone",
"neon"
] | Anime girl restyled as neon pop art | 动漫少女转成霓虹波普街头艺术风 | このイラストを現代アニメ・漫画の影響を基調とした、ポップアート×ストリートアート風スタイルに変換してください。シャープでクリーンな線画とフラットシェーディングをメインに色彩豊かに表現。複雑な幾何学模様など含むスタイリッシュな雰囲気。ベン・デイ・ドットでグラデーションとテクスチャを加味。ネオンカラー中心の鮮やかでコントラストの強い配色で、グラフィカルかつスタイリッシュな雰囲気に。ただし文字は入れないこと。元のイラストの構図・キャラクターを維持。 | The prompt names the target style, then specifies line art, flat shading, Ben Day dots, a neon palette and geometric patterns. It ends by banning text and locking the original composition and character. | 提示词先点明目标风格,再逐项限定线稿、平涂、网点、霓虹配色和几何图案,最后禁止出现文字并要求保留原构图与角色。 | @notoro_ai on X | https://x.com/notoro_ai/status/2051154269064278132 | のとろ | null | 2026-05-04 | null |
End of preview. Expand in Data Studio
Awesome Image and Video Prompts
2,212 prompts for the leading image and video models, each with a preview of the output it made and a link to where it came from. From ai-boost/awesome-image-video-prompts.
- image: 1,272 prompts for GPT Image 2 and 2.5, Nano Banana 2, 2.1 and Pro, and ChatGPT.
- video: 940 prompts for Seedance 2.5, MiniMax H3, Wan 3.0, Kling 3.0 and 4.0, Gemini Omni Flash and Veo 3 and 3.1.
Every row also has a short note on why the prompt works, in English and Chinese. The whole library can be browsed and searched online at agentcraft.dev, with video prompts and image prompts grouped by use.
from datasets import load_dataset
image = load_dataset("Adrian2016/awesome-image-video-prompts", "image", split="train")
video = load_dataset("Adrian2016/awesome-image-video-prompts", "video", split="train")
Fields
| Field | Meaning |
|---|---|
id |
File name in the GitHub repo, without extension |
model |
Model the prompt was written for, such as Seedance 2.5 or GPT Image 2 |
mode |
text-to-image, image-edit, text-to-video, image-to-video, reference-to-video, video-edit, keyframes-to-video or extend |
category |
Use, such as portrait, poster, infographic, cinematic or commercial |
tags |
Free-form tags |
title_en, title_zh |
Short title |
prompt |
The prompt, quoted word for word from its source |
notes_en, notes_zh |
Why the prompt works: the structure or wording that shapes the output |
source_name, source_url |
Where the prompt was published |
credit |
Creator's display name |
likes |
Likes on the source post when captured, if known |
posted |
Date the source was posted, if known |
video_url |
Link to the output clip (video subset) |
preview |
The output image, or a frame of the output clip |
Counts
| image category | rows | video category | rows |
|---|---|---|---|
| portrait | 259 | action | 171 |
| poster | 199 | vlog | 144 |
| photo-to-art | 107 | cinematic | 124 |
| illustration | 107 | commercial | 124 |
| realism | 98 | vfx | 82 |
| collage | 96 | scifi-fantasy | 82 |
| infographic | 89 | anime | 76 |
| character | 80 | dialogue | 55 |
| product | 73 | music-dance | 45 |
| other | 164 | other | 37 |
How it was made
Prompts are collected from creators' posts on X, TikTok and Reddit, and from model makers' prompting guides (Kling, Gemini Omni, Veo, OpenAI). Each candidate is checked by Claude Haiku against the list's rules, and Haiku drafts the bilingual title, notes and tags. The GitHub repo is updated weekly; this dataset is a snapshot of October 30, 2026.
Uses and limits
Good for: studying which prompt structures models follow, few-shot examples for prompt writers, evaluation sets for image and video models, and a bilingual reference for prompt translation.
Limits: each prompt was run once by its creator, so the preview shows one result. Models change, and the same prompt can give a different output today. Some prompts are in Japanese, Chinese or other languages, as written by their creators.
License
The prompts and previews are quoted from their creators for reference, with the creator and source linked in every row. Rights stay with the creators. To have an entry removed, open an issue in the GitHub repo.
Titles, notes, categories and tags are written for this collection and released under GPL-3.0, the license of the GitHub repo.
- Downloads last month
- 184