Joined January 2011
935 Photos and videos
Pinned Tweet
😱AI flight control is getting crazy! One red line controlled this first-person Harry Potter-inspired Quidditch POV. Ride the broom, chase the Golden Snitch, fly through Hogwarts-style towers, skim over Black Lake, dive through a bridge arch, and return for the final catch. 🧹 Workflow breakdown in the comments 👇
🤯AI aerial shots can now be “drawn”. I made a World Landmark Mashup FPV flythrough: Mt. Fuji → Eiffel Tower → Colosseum → Statue of Liberty I drew the flight path directly on the image, then let AI generate the drone movement. 画一根线,可以控制 AI 航拍镜头! 我做了一版《世界地标大乱斗》FPV 穿越: 富士山 → 埃菲尔铁塔 → 罗马斗兽场 → 自由女神像 Workflow & prompt in the comments 👇
178
575
5,535
6,985,217
This Westie accidentally set off an electric motorbike alarm… and immediately started dancing to the beat 🤣 Prompt below
用AI做了一只掌心水蜜桃猫!🍑🐱 它真的长得像极了水蜜桃,软萌到犯规… 这种视频在小红书太容易爆了~
1
1
82
Prompt: Vertical 9:16 realistic smartphone footage at night. A fluffy West Highland White Terrier stands beside a parked electric motorbike on a quiet city street. The motorbike alarm suddenly goes off, flashing lights and beeping rhythmically. The Westie gets startled, then immediately starts dancing to the beat in a funny, chaotic, meme-worthy way: head bobbing, front paws waving, body swaying, small jumps and silly turns. Casual handheld phone video, slight camera shake, the person filming is trying not to laugh. Realistic fur movement, natural night lighting, flashing alarm lights, slight motion blur, authentic viral pet video vibe.
40
🤯 Turning fashion visuals into ghost-trace editorial posters! Made with ChatGPT-Image2: 1. Soft Noise — white crinkled tulle dress 2. Black Echo — black leather coat 3. Wool Memory — grey knitwear set 4. Liquid Trace — silver perfume concept A clean face anchor, distorted fabric trails, quiet white space and tiny magazine-style typography Feels like an experimental fashion magazine, not a normal clothing poster Prompt below 👇
🤯 ChatGPT Image 2 unlocked emotional black-and-white portraits! The trick isn’t making everything blurry. It’s layered focus: a sharp emotional anchor, blurred figures, wind-blown hair, motion edges, mist and grain. Sharp where emotion lands. Blurred where memory fades. Prompt below 👇
2
8
31
3,466
Prompt: Create a vertical high-resolution experimental fashion editorial poster. 【Theme】: [your theme] 【Brand name】: [your brand name] 【Collection title】: [24SS / 24FW / Archive 01 / Capsule 02] 【Fashion / product direction】: [white crinkled tulle dress / black leather coat / grey knitwear set / silver perfume bottle / jewelry / bag / shoes] 【Main color palette】: [grey-white-black / warm beige-grey / cold blue-grey / silver-white-black / dark red-black] 【Model direction】: [adult female fashion model / adult male model / cold minimal duo] 【Editorial copy】: [write a 20–40 word English description] The image should feel like a high-end, cold, minimal and experimental fashion magazine poster. Use a large pale grey-white negative-space background with subtle paper grain, low-contrast noise and printed editorial texture. Avoid clean plastic white backgrounds or commercial studio backdrops. The main subject is [model direction], shown in a full-body or nearly full-body pose, placed slightly right of center. The pose should feel quiet, restrained, distant and editorial. The model should have a calm expression and high-fashion presence. Avoid influencer posing, exaggerated performance or commercial catalog energy. The clothing or product must have strong design value, clear silhouette and rich material details: sheer fabric, crinkled texture, soft volume, leather reflection, knit texture, structured tailoring, transparent layers, natural folds or subtle shine. Keep the person realistic, with natural skin texture and no plastic AI face. Add a controlled “ghost scan blur” effect: vertical motion trails, translucent afterimages, stretched scan lines, slight glitch displacement and soft dissolving blur around the left side of the body, lower garment edges, arms, shoes or product edges. The effect should look like the person, garment or product has been optically scanned, folded by paper texture, or dragged by memory. Important: keep the face or upper body relatively clear as the visual anchor. Do not blur the entire person. The key contrast is: clear emotional anchor blurred ghost trace around the edges. If the theme includes a product such as perfume, jewelry, bags or shoes, place it subtly near the model, at the bottom of the poster or in the foreground. Connect the product with the model through reflection, transparent blur, liquid trace, glass distortion or scan-like afterimages. Do not make it look like a normal e-commerce product image. Use a minimal fashion magazine layout: small English title at the top, such as “[Brand name] / [Collection title]”; a small paragraph of editorial copy near the body or product; and a tiny abstract logo at the bottom. Typography should be clean, restrained, small, high-end and sans-serif. Avoid large promotional text, loud fonts or ad-like design. Overall mood: low saturation, low contrast, cold, quiet, premium, experimental, editorial, archive-like, independent designer lookbook, avant-garde fashion magazine, artistic brand campaign. Avoid: cartoon, illustration, cheap poster design, e-commerce ad, promotional banner, large text, complex background, multiple random people, collage grid, over-sharpening, over-retouched skin, obvious AI look.
1
4
862
🤯 Mirrors instantly make fashion visuals feel more expensive! A “Mirror Luxury Campaign” system made with ChatGPT-Image2: 1. Black-gold eveningwear 2. Burgundy velvet dress 3. Pearl jewelry 4. Vintage perfume Formula: mirror reflection blurred foreground soft light leaks retro warm tones magazine-style typography Great for fashion, jewelry, perfume and luxury accessory concepts! Prompt below 👇
🤯 AI Fashion Visual Systems Vol.02: Campaign Typography! Made with ChatGPT-Image2 Typography becomes the structure of the campaign: 1. Surreal typography poster 2. Product hero campaign 3. Dark fashion week poster 4. Beauty close-up cover Oversized type brand mood editorial composition = fresh campaign visuals Great for mockups, fashion posters, luxury ads and early brand concepts Prompt in the comments 👇
4
8
45
5,367
Mirror Luxury Campaign prompt: Create a vertical 9:16 high-end fashion campaign poster for 【brand name】, featuring 【theme / product】. Main title: 【main title】 Subtitle: 【subtitle】 Color palette: 【main colors】 Outfit: 【outfit style】 Accessories / props: 【accessories / props】 Scene: 【scene】 Action: 【model action】 Style: premium, elegant, retro-luxury, editorial, magazine-style, cinematic, soft-focus photography. Use a mirror-based composition: the model appears through a mirror reflection, with a blurred foreground, soft light leaks, warm retro tones, realistic fabric texture, subtle reflections, and a polished luxury campaign feel. The model should look like an adult woman with realistic skin texture, refined styling, and a natural elegant pose, such as adjusting earrings, touching a necklace, fixing hair, holding a perfume bottle, or looking into the mirror. Include clean magazine-style typography: small editorial text at the top, a large serif headline in the lower-middle area, and small CTA / season text at the bottom. Avoid cheap commercial styling, flat lighting, plastic skin, messy typography, distorted jewelry, awkward hands, and generic e-commerce poster layouts.
1
5
1,243
💙 Sweet & energetic K-pop dance — actually moving! Made with ChatGPT Seedance from a 16-panel storyboard. Rooftop sunshine, ocean view, light blue idol look. Hair flips, leaps, spins, low poses and clean kicks. Workflow prompts below 👇
AI figure skating is getting crazy. ❄️ I used ChatGPT-Image2 Seedance to turn a 16-panel storyboard into a cinematic frozen-lake skating video. Moonlight, snow mountains, flying ice shards, flowing dress — plus Spiral, Ina Bauer, Hydroblade, Jump and a final pose. Workflow prompts in the comments 👇
12
30
185
15,362
1/ Workflow Step 1: Generate the 16-panel storyboard image first. This gives the dance clear motion beats: opening pose, walk-in, hair flip, leap, spin, kick, low pose, ending pose. Step 2: Generate a character sheet / role card as a consistency reference. This helps lock the dancer’s face, hairstyle, outfit and body proportions before moving to video.
1
4
20
2,033
2/ Video generation Step 2: Send both the storyboard the character sheet into Seedance. The storyboard controls the choreography. The character sheet helps keep the face and styling consistent. That way the result feels like a real structured dance video, not random motion or identity drift. Prompt: Generate an outdoor sweet K-pop dance video based on the 16-panel storyboard and the character sheet. Keep the same young adult East Asian female dancer, with the same face, hairstyle, body proportions, and outfit throughout the whole video. She wears a light blue bandeau top and a matching light blue mini skirt, with white socks and white sneakers. Scene: sunny rooftop / seaside terrace, blue sky, ocean view, bright natural light, light wind. Follow the storyboard in order and keep the choreography dynamic: power opening, walk-in, hair flip, side leap, drop & rise, hip hit, arm sweep, full spin, kick, lean-back wave, jump accent, low pose, rise, big open pose, high-leg accent, ending pose. The motion should feel sweet, energetic, rhythmic, and clearly structured — not random shaking or repeated posing.
1
5
1,748
🤯用Imange2生成《香水爆炸主视觉海报》,香味也能看见! 把香调、材质、情绪和氛围一起可视化✨ 白花的清透、玫瑰的浓郁、柑橘木质的清爽、乌木皮革的深沉 都可以变成一张品牌级香氛海报! 很适合香水品牌视觉、产品 KV、社媒海报和广告创意探索! 提示词放评论区👇
🔥生产力UP! 我用ChatGPT-Image2生成《餐饮主打爆炸海报》,视觉系食欲! 4个美食主视觉海报:汉堡、拉面、奶茶、寿司卷 把食材层次、酱汁动态、热气氛围和口感卖点融进一张图 适合做餐饮新品主KV、门店海报、外卖主图和朋友圈物料! 通用提示词在评论区👇
6
5
40
3,412
提示词: 请生成一张 9:16 竖版的「香水主视觉海报」。 【产品信息】 香水名称:{填写香水名称,例如:晨光白花 / 夜色玫瑰 / 柑橘雪松 / 乌木皮革} 香调类型:{填写香调,例如:白花清透调 / 玫瑰琥珀调 / 柑橘木质调 / 乌木皮革调} 核心香材:{填写 3-5 个香材,例如:茉莉、橙花、白麝香、玫瑰、琥珀、佛手柑、雪松、乌木、皮革、焚香} 情绪氛围:{填写视觉气质,例如:清透柔光、浓郁夜色、清爽中性、深沉奢华} 主色调:{填写颜色,例如:奶白金色、深酒红、柑橘黄绿、黑金深棕} 【画面要求】 生成一张高级香水品牌广告海报,不是普通产品图,也不是电商详情页,而是一张具有奢侈香氛 campaign 质感的主视觉作品。 画面采用 9:16 竖版构图,香水瓶作为中心 hero object,瓶身居中偏大,具有高级玻璃质感、精致瓶盖、真实反光、柔和高光和品牌广告级打光。 香调元素围绕瓶身自然展开,形成“气味被可视化”的感觉。元素可以包括花瓣、果皮、木片、树脂、皮革、烟雾、水珠、光粒子、丝绸、雾气等,但要根据香调选择,不要全部堆满。 画面重点不是拆开香水瓶,而是把香气、材质、情绪和香调层次表现出来。香调元素要有空间层次、前后景、轻微悬浮感和动态感,像香味从瓶身周围缓慢扩散开来。 【文字设计】 画面可以加入少量中文文字,但不要太多。 顶部或中上部加入: 主标题:{香水名称} 副标题:{核心香材,例如:茉莉 · 橙花 · 白麝香} 底部可以加入一句短文案,例如: 让每一次呼吸都如晨光般清澈 或: 让夜色在呼吸间缓缓盛开 文字要有高级品牌海报感,简洁、克制、留白充足,不要像说明书或电商参数页。 【风格要求】 高级香氛广告、奢侈品海报、真实摄影质感、电影级光影、精致玻璃瓶、香调可视化、强氛围感、柔和景深、细腻材质、品牌主视觉、9:16 竖版。 【避免】 不要电商白底图 不要普通静物摆拍 不要元素乱飞 不要花瓣过多像婚庆图 不要文字太多 不要过度赛博风 不要廉价塑料质感 不要背景杂乱 不要让香调元素抢走香水瓶主体 最终效果要像一张可以用于香水品牌发布、社媒广告、产品 KV、门店海报或香氛 campaign 的高级主视觉海报。
1
1
3
1,335
🤯 AI Fashion Visual Systems Vol.02: Campaign Typography! Made with ChatGPT-Image2 Typography becomes the structure of the campaign: 1. Surreal typography poster 2. Product hero campaign 3. Dark fashion week poster 4. Beauty close-up cover Oversized type brand mood editorial composition = fresh campaign visuals Great for mockups, fashion posters, luxury ads and early brand concepts Prompt in the comments 👇
🤯 I used ChatGPT-Image2 to create a fashion editorial visual system! It can turn different brand moods, products and layouts into fresh campaign directions. 4 directions: 1. Giant typography campaign 2. Couture magazine cover 3. Accessories editorial page 4. Lookbook collage spread Change the brand, product, model, headline or color palette — and the whole campaign direction changes. Great for moodboards, early brand concepts and visual exploration. Prompt in the comments 👇
3
5
55
9,342
Prompt: Create a high-end fashion campaign typography poster. Theme: [your campaign theme] Brand / event name: [your brand name] Direction: [surreal typography poster / product hero campaign / dark fashion week poster / beauty close-up cover] Product or subject: [product / model / event / beauty concept] Brand mood: [luxury / dark / minimal / romantic / avant-garde / modern / cinematic] Color palette: [main colors] Main headline: [headline] Subheadline: [subheadline] Small text: [optional short details] Aspect ratio: [9:16] The image should feel like a premium fashion campaign, luxury editorial poster or magazine-style brand visual. Typography should not be just decoration — it should become the structure of the campaign. Use oversized type, strong visual hierarchy, editorial composition, negative space, realistic fashion photography, refined lighting and a clear brand mood. Choose one direction: 1. Surreal typography poster Use oversized cut-up letterforms, mirror / flip / layered composition, dark luxury mood, strong contrast and experimental fashion styling. 2. Product hero campaign Make the product and model both important. The product should feel like the hero object, supported by premium campaign typography and elegant brand layout. 3. Dark fashion week poster Use dramatic dark background, strong model pose, bold event typography, runway / show poster mood, date and small event details. 4. Beauty close-up cover Use a close-up face or side profile, refined skin texture, soft lighting, elegant magazine-cover typography and a premium beauty campaign feeling. Keep the image stylish, readable and high-end. Avoid cheap e-commerce style, clutter, weak typography, random decorations, fake plastic skin, bad anatomy, distorted products, messy text and real copyrighted magazine logos. Generate one finished poster only.
1
7
1,088
💘用 AI 做元气女友感牵手跟拍视频,有恋爱内味了! 从一张牵手回眸照片,做成 8 秒生动视频: 慢慢走 → 小跑 → 回头笑 → 头发裙摆被风带起 画面关系感写出来: 牵手拉扯、镜头前冲、发丝甩动、眼神停留、俏皮回头… 像被她拉着跑进夏天~! 提示词在评论区👇
📸是心动啊?用AI实现写真自由:牵手回眸直拍,女友感拉满! 4个女生,不同发型、穿搭和场景,同一个高光瞬间: 回头、牵手、风吹乱头发,阳光从侧后方打进来 发丝边缘像会发光一样,眼神也像在说话 既甜、又真、又有电影抓拍感!✨ 提示词放评论区👇
12
11
71
5,640
提示词: 基于首帧图片生成一段竖版 9:16 的高清写实短视频,时长 8 秒。 【一致性要求】 以首帧图片为准,保持同一位成年女性、同一张脸、同一发型、同一条白色连衣裙、同一花园小路、同一阳光树影环境、同一牵手关系。不要换脸,不要换发型,不要换服装,不要改变场景结构,不要让人物五官漂移,不要让牵手动作变形。 【分镜】 镜头1:牵手慢慢往前走,阳光透过树叶落在头发、肩膀和白色裙子上。 镜头2:她突然加快脚步,小跑起来,镜头被她带着轻微前冲。 镜头3:她一边小跑一边回头笑,头发和裙摆自然晃动。 镜头4:她轻轻拉一下“我”的手,像在说“快点跟上”。 镜头5:她继续拉着我往前跑几步,稍微放慢速度,回头露出一个俏皮又开心的笑。 【镜头语言】 真实恋人视角跟拍,轻微手持感,轻微呼吸感镜头,自然前冲感,自然对焦变化,浅景深,轻微运动模糊。镜头要像被她牵着手带着走,画面可以有轻微晃动,但主体始终清晰可见。 【真实细节】 长发随着小跑和转头自然甩动,白色裙摆随着步伐轻轻摆动,牵手的手部有真实拉扯感,肩膀、手臂、身体重心随着走路和小跑自然变化,背景树叶和花草轻微晃动,阳光形成斑驳光影。 【表情】 表情从轻轻笑 → 开心 → 调皮 → 有点撒娇地回头笑。笑容要真实自然,有感染力,不要僵硬假笑,不要夸张大笑,不要一直盯镜头营业。 【避免】 避免剧烈跳跃、舞蹈动作、夸张表演、动作过猛、广告感、MV 感、人物换脸、服装变化、手部错误、牵手姿势异常、手臂拉长、身体比例异常、脚步不接地、背景扭曲、发丝闪烁、运动模糊过重。
1
4
1,202
🔥用 ChatGPT-Image2 做了一组《火影忍者》情绪破碎海报,太戳了! 把 4 个角色定格在动画里的高光情绪瞬间: 1、鸣人:九尾失控前 2、佐助:终结谷决裂 3、卡卡西:来迟的守护 4、鼬:最后一次点额头 黑色负空间、锐利碎片、剧情残片、角色特写、破碎文字… 静态海报,也能把一段动画里的情绪高潮定格下来! 提示词在评论区 👇
🔥 Demon Slayer emotion posters hit hard! I froze 4 characters at one emotional moment: 1. Tanjiro 竈門炭治郎 2. Nezuko 竈門禰豆子 3. Zenitsu 我妻善逸 4. Rengoku 煉獄杏寿郎 Black negative space, sharp fragments, close-up details, and one controlled color burst. A single poster can freeze motion, emotion, and impact in one frame. Prompt in the comments 👇
3
1
16
4,243
通用提示词: 请生成一张 9:16 竖版、高完成度、高清锐利的二次元情绪破碎海报。 【角色】 {{角色名}} 【高光剧情瞬间】 {{角色在动画里的代表性情绪高光片段}} 【画面文字】 角色名大字:{{角色名}} 副标题:{{高光片段标题}} 短句:{{一句情绪台词 / 内心独白}} 这不是普通角色卡,不是安静站姿,也不是单纯展示大招的战斗图。请把画面聚焦在角色最有情绪张力的一个动画高光瞬间,让整张海报像动画里最冲击的一帧被突然撕裂、放大、定格。 人物必须有强烈情绪:眼神、眉眼、嘴角、肩颈、手部、身体姿态都要参与表达。不要平静摆 pose,要像刚经历重大冲击后的瞬间停顿。 画面采用黑色 / 深色负空间,高级动漫主视觉海报质感,高清、干净、锐利,不要灰蒙、脏旧、低清、过度颗粒化。 使用大量有设计感的不规则玻璃碎片、斜向裂隙、局部 close-up 窗口、透明切片和飞散碎片。但碎片不能只是装饰,每块碎片里都要尽量有剧情信息,比如:眼睛特写、手部特写、武器 / 忍术细节、角色伤痕、护额、家族符号、重要人物残影、战斗环境残片、记忆碎片、情绪证据。 文字排版要有明显风格感,不能太规整,不能像普通印刷体。角色名、副标题和短句要融入海报结构,可使用竖排、错位、切割、局部断裂、裂隙化、拼贴感、局部遮挡、能量侵蚀感等处理。文字要有破碎情绪海报的设计感,但必须清晰、完整、可读,不要乱码。 整体风格:高清动漫主视觉海报、视觉小说封面感、成熟漫画排线、黑色负空间、锐利碎片拼贴、剧情残片、角色特写、强情绪眼神、破碎字体设计、一个情绪爆点色。 避免:普通角色卡、普通站姿、表情太平、单纯大招图、碎片空洞无内容、文字太规整、文字乱码、官方 logo、编号、系列名、Q版、低幼卡通、3D 渲染感、全图脏污、灰蒙低清、错误手指、五官崩坏。
1
3
2,304
📸用 ChatGPT-Image2 实现写真自由:酒店微醺,性感与松弛! 4 个酒店夜拍方向: 1. 深酒红 minibar 2. 香槟白窗边椅 3. 珍珠灰真丝衬衫 4. 黑色缎面吊带裙 这组写真最想留下的是夜里的松弛感: 酒杯、落地窗、真丝裙、眼神、微醺状态和夜景光线控制 Prompt 在评论区 👇
📸 AI Photoshoot: Beach Lounge Portraits. The realism is getting crazy. Same camera logic, same composition, different models, outfits, chairs and beach moods. 1. Blush sheer cover-up 2. Cream knit vacation look 3. Champagne rose resort mood A clean beach portrait system for summer fashion shoots. Prompt in the comments 👇
7
18
177
22,488
酒店夜景微醺写真 提示词: 请生成一张竖版 9:16 的高清写实夜景人像写真,单张图片,不要拼图,不要文字。 主题:【高层酒店夜景微醺写真】 人物气质:【都市轻熟 / 冷感时装 / 东方轻奢 / 温柔克制】 服装:【黑色缎面吊带长裙 / 深酒红抹胸长裙 / 香槟白挂脖长裙 / 珍珠灰真丝衬衫 缎面半裙】 场景:【落地窗单椅 / minibar 高脚椅 / 窗边休闲椅 / 床边扶手椅】 道具:【红酒杯 / 香槟杯 / 鸡尾酒杯】 画面像真实纪实感的时尚夜景肖像,不是普通酒店打卡照。成年年轻女性坐在【场景】中,身体微微斜靠,肩膀放松,腰线轻轻转向镜头,一只手拿着【道具】,另一只手自然落在椅边或裙摆旁。 人物有微醺后的安静状态:眼睑微垂,视线略微失焦,表情轻倦、松弛、若有所思。服装为【服装】,面料真实,版型稳定,有自然包裹感和垂坠感,顺着身体形成干净流畅的线条。不加外套,不加丝袜。 皮肤为自然哑光妆面,真实相机拍摄质感,保留细腻肌理和真实肤色层次,不过度反光。光线采用低反差中性柔光和室内漫反射光,只塑造人物轮廓、面部明暗和服装层次。窗外是虚化城市夜景灯海,酒杯、玻璃和少量服装材质可以有细腻反射。 构图为中景偏全身,人物占画面约 70%,头顶只留少量空间。腿部姿态简单清晰,一条腿自然前伸,另一条腿轻轻收回,腿部与裙摆、椅子保持清晰分离。 避免:游客照构图、头顶大空白、油亮皮肤、蜡像感、手指错误、酒杯变形、腿部模糊、廉价夜店感。
1
17
2,941
🤯ChatGPT-Image2 can finally turn emotion into pure visual structure! I tested a dark anime illustration system: 1. Rain-sealed memory 2. Snow lantern vigil 3. Deep-sea silence 4. Ember night The system: black negative space emotional character symbolic prop hard-edged fragments Feels like a commercial-grade visual novel cover system! prompt below👇
5
33
3,582
Prompt: Create a vertical 【aspect ratio】 original anime-style emotional illustration poster. 【Theme】: 【Character】: 【Emotional concept】: 【Outfit】: 【Main prop】: 【Symbolic elements】: 【Main color palette】: 【Accent color / emotional color leak】: 【Fragment style】: 【Extra requirements】: The poster should feature one original adult character placed in the center or slightly off-center, shown from half-body to knee-up. The character must carry a clear emotional state, such as suppressed sadness, silent pressure, emotional collapse before it happens, lonely defense, frozen memory, deep-sea silence, or quiet anger. The face should be calm but expressive, with tension in the eyes, mouth, shoulders, neck, and hands. The character wears 【Outfit】 and holds 【Main prop】. The prop should not be decorative only; it must support the emotional theme. For example, an umbrella can suggest sealed memory, a lantern can suggest waiting, a sword can suggest restrained anger, a book can suggest silence, a mask can suggest self-protection, and a glass bottle can suggest trapped words. Use a large area of deep black, charcoal gray, or near-black negative space as the background. Avoid a realistic detailed scene. Add only subtle paper grain, low-contrast texture, mist, dust, rain, ash, water ripples, or old poster texture. Build a strong fragmented composition around the character using 【Fragment style】: irregular rectangles, square cutouts, hard-edged slices, fractured panels, broken visual windows, or mirror-like detail fragments. These fragments should feel like emotional cracks, memory pieces, visual noise, or broken space. Some fragments should partially cover the character, while others should zoom in on details such as eyes, hands, hair, clothes, prop structure, flowers, cracks, water, smoke, or light. Add 【Symbolic elements】 around one side or the lower part of the character. These elements should interact with the character through wrapping, growing, invading, floating, cracking, burning, dissolving, or surrounding. Do not spread them evenly across the whole image. They should feel like the character’s emotion becoming visible. Use 【Main color palette】 as the base, with one precise 【Accent color / emotional color leak】 appearing only in key areas, such as a scarf, flower petal, light source, crack, eye highlight, weapon reflection, water glow, or burning edge. The accent color should feel controlled and memorable, not overly saturated. Style: mature anime illustration, dark emotional poster, visual novel cover, character concept art, manga linework, watercolor wash, low saturation, paper grain, asymmetric collage layout, symbolic emotional storytelling, sharp fragments, soft atmosphere, strong cover design. Avoid: cute chibi style, childish anime face, generic character standing pose, empty expression, 3D render, realistic photo style, messy background, random collage, oversaturated colors, fragments used only as decoration, broken hands, bad anatomy, distorted face, watermark, logo, text.
5
1,336