← 返回首页
技能

AI Image Prompts · 绘画提示词库

github · 1.0.4 · 下载量 1,092 · 热度分 90.0

出处与授权

作者: YouMind-OpenLab
许可证: MIT 许可证 (原文)
版权声明: Copyright (c) 2026 YouMind-OpenLab

本站只收录项目元数据与简介,不对作者正文做转载;完整内容请到上方出处查看。商用前请自行核对原仓库的 LICENSE 与 NOTICE。

展开许可证全文
MIT License

Copyright (c) 2026 YouMind-OpenLab

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

LICENSE · 2026-09-26 23:14:14

产品介绍

Recommend curated prompts from a 10,000+ real-world image generation prompt library. Works with ANY AI image model — Nano Banana Pro, Nano Banana 2, Seedream 5.0, GPT Image 1.5, Midjourney, DALL-E 3, Flux, Stable Diffusion, and more. Use this skill when users want to: - Find proven image generation prompts (any model) - Get prompt inspiration for portraits, products, social media, posters, etc. - Create illustrations for articles, videos, podcasts, or marketing content - Browse categorized prompt templates with sample images - Translate and understand prompt techniques

资源信息

作者:YouMind-OpenLab
分类:图文音视频多模态生成
语言:en
版本:1.0.4
认证方式: 无需认证
许可证:MIT
是否付费: 官方标注免费
发布方:YouMind

许可证信息摘自官方源,可能过时或缺失。商用前请以原仓库的 LICENSE 文件为准。

标签
绘画提示词文生图营销素材

安装方式

npx skills i YouMind-OpenLab/ai-image-prompts-skill

功能说明

  • 按用途分成 11 类,不是随机堆砌
  • 全文已入库,站内直接检索、可复制
  • 语义检索(描述需求即可)
  • 内容改写模式(贴正文生成定制提示词)
  • 数据每日同步两次
  • MIT 许可,无需 API Key
适用场景
找某类图片的现成提示词给文章/视频/播客配插画做海报 / 封面 / 社媒图翻译并理解提示词写法

提示词库 · 2.3万

数据源:社区提示词库(MIT 许可,每日同步)

共 15141 条提示词,按 11 个用途归类;同一条可归入多个用途,因此分类条数相加会大于去重后的总数。

命中 2105 条

Japanese Woman Portrait Prompt for Nano Banana Pro

profile-avatar

A prompt designed to generate a close-up portrait of a clean-cut, young Japanese woman in her 20s, featuring straight long black to dark brown hair with bangs, against a white background, specifically for the Nano Banana Pro model.

展开提示词全文
A young, clean-cut Japanese woman in her 20s, close-up of the face, white background, straight long black to dark brown hair, with bangs.

Amateur Smartphone Snapshot Prompt for Nano Banana 2

profile-avatar

A detailed image generation prompt for Nano Banana 2, focusing on creating a casual, amateur-style smartphone photo of a young woman resembling Sadie Sink, wearing a Spider-Man themed bodysuit, with specific details on lighting, texture, and pose.

展开提示词全文
{
  "photo_type": "Amateur smartphone snapshot, casual indoor photo captured on an iPhone 16 Pro front camera with typical lens curvature.",
  "subject": {
    "age": "Young adult",
    "body_type": "Curvy hourglass figure with soft athletic muscle tone",
    "description": "A young adult female with a heart-shaped facial structure, fair and lightly freckled complexion, and long, wavy copper-red hair parted in the center.",
    "ethnicity": "Caucasian",
    "expression": "Playful winking expression with a slight, subtle smile",
    "gender": "Female",
    "mirror_rules": "Not a mirror selfie, direct amateur capture",
    "pose_description": "Sitting comfortably on a white sectional sofa, leaning back slightly, one arm raised with fingers grazing her hair while the other hand rests casually on the cushion.",
    "tattoos": "None"
  },
  "clothing": {
    "bottom": {
      "color": "Black and white",
      "details": "High-cut leotard style revealing the hips, featuring a white webbing pattern over a matte black base.",
      "fabric": "Spandex-polyester blend",
      "type": "Integrated bodysuit bottom"
    },
    "fit_and_physics": "Compression creases visible at the waist and armpits; fabric stretching tightly over the chest and hips with visible seam tension.",
    "outerwear": {
      "color": "None",
      "details": "None",
      "fabric": "None",
      "type": "None"
    },
    "shoes": {
      "color": "None",
      "details": "None",
      "fabric": "None",
      "type": "None"
    },
    "top": {
      "color": "Black and white",
      "details": "Long-sleeved bodysuit with a high neckline, featuring a large white stylized spider emblem on the chest and white web patterns on the sleeves.",
      "fabric": "Matte performance fabric",
      "type": "Spider-Man themed bodysuit"
    }
  },
  "hair": {
    "color": "Copper-red",
    "interaction": "One hand is gently lifting a section of a pigtail, creating natural movement and flyaway strands.",
    "style": "Long, wavy copper-red hair, parted in the center, styled into two high, voluminous loose pigtails."
  },
  "face": {
    "description": "Use the uploaded character reference image, keep all facial details 100% unchanged",
    "eyebrows": "Thick, naturally feathered eyebrows",
    "eyes": "Light blue almond-shaped eyes, one winking",
    "lips": "Full, well-defined lips with a natural satin finish",
    "makeup": "Minimal, natural-looking makeup enhancing the freckles"
  },
  "skin_texture": "Fair, lightly freckled complexion with visible natural pores and subtle skin variations.",
  "biological_features": {
    "eye_complexity": "Wetness on the tear duct and sharp corneal reflections from room lighting.",
    "hair_strand_physics": "Fin

Golden Hour Mirror Selfie Realism

profile-avatar

A structured prompt for Nano Banana Pro to generate a raw, intimate mirror selfie taken during the golden hour, focusing on natural lighting, soft grain, and the candid moment of a person unbuttoning a shirt.

展开提示词全文
{
  "meta": {
    "camera": "iPhone 17 Pro",
    "lens": "35mm",
    "aspect_ratio": "4:5",
    "style": "raw iphone indoor realism, natural HDR, soft grain"
  },
  "scene": {
    "location": "bedroom mirror",
    "time": "golden hour sunset leaking through window",
    "atmosphere": "warm quiet room, soft shadows, intimate silence"
  },
  "lighting": {
    "primary": "golden sunlight hitting from side",
    "effect": "fabric slightly translucent, warm glow on skin, natural highlights"
  },
  "subject": {
    "pose": {
      "position": "standing in front of mirror",
      "action": "mid-unbutton shirt moment",
      "body": "one hand holding shirt open slightly, other holding phone",
      "torso": "subtle forward lean"
    },
    "expression": {
      "eyes": "locked into mirror",
      "mouth": "barely parted",
      "energy": "caught in a private moment"
    },
    "hair": {
      "style": "soft messy natural fall"
    },
    "outfit": {
      "top": {
        "type": "oversized white button-up",
        "state": "half unbuttoned",
        "effect": "light passing through fabric, shape hinted not shown"
      },
      "bottom": {
        "type": "minimal lounge shorts",
        "color": "neutral"
      }
    }
  },
  "camera": {
    "angle": "mirror selfie slight tilt",
    "distance": "medium",
    "imperfection": "tiny blur + fingerprint smudge on mirror"
  }
}

Black-and-White Fashion Collage with Face Preservation

profile-avatar

A detailed prompt for Google Gemini Nano Banana Pro to create a photorealistic black-and-white 2x2 collage of a male model, requiring the preservation of the face and outfits from attached reference images, focusing on high-contrast fashion photography style.

展开提示词全文
using the attached reference photos as sources, preserve the same male face and hairstyles from the face reference and the same outfits from the clothing references. create a photorealistic black-and-white studio photoshoot collage arranged in a clean 2x2 grid with thin white spacing between frames. neutral gray seamless studio background, soft diffused fashion lighting with gentle shadows. frame 1: waist-up portrait, man in a plain rolled-sleeve t-shirt, one hand lightly touching the chin, direct intense gaze to camera, relaxed shoulders. frame 2: three-quarter standing pose in the same t-shirt, body angled slightly sideways, one hand near the mouth in a thoughtful gesture, casual stance. frame 3: mid-shot in a denim jacket, body turned partly away, one hand holding the jacket collar or shoulder, confident look toward camera. frame 4: tight portrait in the denim jacket, fingers resting near the lips and nose bridge, fashion editorial expression. natural skin texture, realistic fabric folds and denim texture, sharp hair detail. high-contrast monochrome conversion, cinematic fashion photography, 85mm lens look, f/2.8. no text, no watermark

Emma Watson Casual Lifestyle Photo

profile-avatar

A detailed JSON-formatted prompt for Nano Banana 2 to generate a photorealistic image of Emma Watson taking a selfie in a cozy bedroom setting, focusing on lighting, composition, and specific details like freckles and accessories.

展开提示词全文
{
  "style": "casual lifestyle photography, intimate candid, soft modern realism",
  "subject": "young woman with long wavy blonde hair and glasses taking a selfie while lying in bed, wearing a yellow fitted bodysuit",
  "pose": "reclined on pillows with one arm extended holding the camera, finger gently touching lips in a playful, relaxed gesture",
  "environment": "cozy bedroom setting with patterned butterflies pillowcases, textured bedding, bedside table with lamp, one piece miniature, and small personal items",
  "mood": "warm, relaxed, personal, slightly playful",
  "lighting": "soft warm indoor lighting from bedside lamp combined with direct flash effect, creating gentle shadows and natural skin highlights",
  "details": "freckled skin, subtle natural makeup, delicate necklace worh key, casual bedroom textures, soft fabric folds",
  "camera": "close-up selfie perspective, slightly high angle, shallow depth of field",
  "color_palette": "warm yellows, soft creams, natural skin tones, warm amber light",
  "composition": "subject centered with diagonal arm line leading to face, intimate framing emphasizing facial expression and upper body",
  "influences": "contemporary social media photography, natural lifestyle portraiture",
  "quality": "high resolution, photorealistic, clean detail, subtle grain"
}

Sydney Sweeney Lazy Morning Selfie Simulation

profile-avatar

A complex, JSON-structured prompt designed to generate a highly specific, photorealistic image simulating a handheld smartphone selfie of a woman (Sydney Sweeney reference) lying on a bed in loungewear, detailing physical features, clothing, pose, and camera characteristics.

展开提示词全文
The prompt is already in English. It describes generating a high-angle selfie photo of a woman in her early 20s with a curvy athletic body, oval face, blue almond-shaped eyes, blonde wavy hair, and specific tattoos. She is wearing a white crop top and high-waisted shorts, lying prone on a white bed propped up on elbows. The photo should look like an amateur iPhone 16 Pro front camera snapshot with natural lighting, specific details like visible pores, vellus hair, and realistic skin texture. The background includes a gold-framed mirror, white bed sheets, white hydrangeas in a glass vase, and a white Prada shopping bag. Camera settings include f/11 aperture, ISO 400, 24mm equivalent focal length, and a standard iPhone HDR curve for natural, unedited look.

Stylized 3D Cartoon Man in Streetwear

profile-avatar

A prompt for Gemini Nano Banana 2 to generate a stylized 3D cartoon character, a man in modern gold and white winter streetwear, featuring specific accessories and a confident pose, rendered in a Pixar-Disney-like style.

展开提示词全文
Stylized 3D cartoon a man "image Uploaded" in modern streetwear, full-body character, black aviator sunglasses, bright gold winter hat with white pom-pom, bright gold down jacket with thick white fur trim, bright gold sweatshirt under it, silver chain with snowflakes, baggy blue pants, white gloves, white sneakers, confident posture pointing with both hands (finger guns), clean soft studio lighting, smooth gradients, high detail, clear edges, winter blue background with fine snow particles, cute but cool atmosphere, Pixar-Disney-like 3D render, centered composition

Group Selfie of Four Women in Barcelona

profile-avatar

A JSON prompt for generating a group selfie featuring four distinct individuals (Emma Stone, Anne Hathaway, Ana de Armas, and a fourth person, using reference images for identity lock) in a casual, candid style on a European city street, emphasizing wide-angle distortion and realistic skin texture.

展开提示词全文
{
  "image_generation_prompt": {
    "subject": {
      "description": "Four young women taking a group selfie outdoors, Use uploaded reference images, keep identity exact for each person",
      "identity": {
        "person_1": "Use uploaded reference image 1, keep identity exact — no mixing",
        "person_2": "Use uploaded reference image 2, keep identity exact — no mixing",
        "person_3": "Use uploaded reference image 3, keep identity exact — no mixing",
        "person_4": "Use uploaded reference image 4, keep identity exact — no mixing"
      },
      "consistency_rules": "Each face must strictly match its corresponding reference image — same facial structure, eyes, nose, lips, proportions, and skin tone. No blending or identity drift.",
      "hair": {
        "person_1": "match reference image — natural strands, slight movement",
        "person_2": "match reference image — realistic texture and flyaways",
        "person_3": "match reference image — natural volume and separation",
        "person_4": "match reference image — soft realistic flow"
      },
      "face": {
        "details": "Use uploaded reference images, keep identity exact — visible pores, natural skin texture, no beautify smoothing",
        "expressions": {
          "person_1": "playful pout",
          "person_2": "winking smile",
          "person_3": "soft confident smile",
          "person_4": "side-leaning relaxed expression with closed eyes"
        }
      },
      "body": "natural proportions, close group pose, shoulders touching"
    },
    "attire": {
      "style": "casual streetwear / tourist outfits",
      "details": "simple tops, neutral tones, minimal branding, realistic fabric wrinkles and folds"
    },
    "composition": {
      "type": "group selfie",
      "framing": "tight upper-body framing with slight wide-angle distortion",
      "arrangement": "four faces close together, slightly overlapping, filling frame",
      "camera_position": "held by one subject at arm’s length"
    },
    "environment": {
      "setting": "European city street (Barcelona-style architecture)",
      "background": "ornate historic building with balconies and stone details, slightly blurred depth",
      "details": "tourist environment, soft crowd presence, urban textures"
    },
    "lighting_and_mood": {
      "lighting": "natural daylight, soft diffused light with slight directional highlights",
      "effects": "even skin tones, subtle shadows, realistic outdoor color balance",
      "mood": "fun, friendly, travel lifestyle, candid"
    },
    "camera_and_technical": {
      "style": "smartphone selfie photography",
      "lens": "24mm wide-angle equivalent",
      "aperture": "f/1.8",
      "quality_tags": [
        "high resolution",
        "natural depth of field",
        "slight lens distortion",
        "realistic skin texture",
  "
      ]
    }
  }
}

Sadie Sink Close-Up Flash Photography Collage

profile-avatar

A structured JSON prompt for Nano Banana Pro to generate a 3-frame vertical collage of Sadie Sink using flash photography. The prompt emphasizes strict identity consistency across frames, natural skin texture, specific attire (black thin-strap top), and a candid, intimate nighttime aesthetic in a bedroom setting.

展开提示词全文
{
  "image_generation_prompt": {
    "subject": {
      "description": "Young woman appearing in a 3-frame vertical collage, Use uploaded reference image, keep identity exact across all frames",
      "hair": "long straight light brown hair, natural shine, visible strand separation, slight flyaways, realistic density and soft layering",
      "face": "Use uploaded reference image, keep identity exact — consistent facial structure in all frames, visible pores, subtle skin texture, natural highlights, no smoothing or beautification",
      "consistency": "same identity in all 3 frames, no face drift, no feature alteration"
    },
    "attire": {
      "clothing": "black thin-strap top, soft fabric with natural folds and slight stretching",
      "accessories": "minimal gold bracelet and delicate necklace with realistic reflections and minor imperfections"
    },
    "composition": {
      "layout": "3 vertical stacked frames",
      "framing": "top: close-up portrait, middle: macro crop of lips and hand, bottom: medium portrait with full face and upper body",
      "spacing": "thin borders separating each frame"
    },
    "pose": {
      "variation": [
        "top frame: resting face on one hand, lips slightly pursed",
        "middle frame: close-up of lips and hand resting on shoulder",
        "bottom frame: both hands framing face, direct gaze at camera"
      ],
      "expression": "soft, intimate, slightly playful, natural variations across frames"
    },
    "environment": {
      "setting": "bedroom",
      "background": "soft bed linens, pillows, bedside table with warm lamp light, slightly blurred depth"
    },
    "lighting_and_mood": {
      "lighting": "direct on-camera flash combined with warm ambient lamp light, strong highlights on skin, soft shadows",
      "effects": "flash reflections on skin, slight overexposure on highlights, warm background glow",
      "mood": "intimate, cozy, candid nighttime aesthetic"
    },
    "camera_and_technical": {
      "style": "flash photography, candid lifestyle portrait",
      "lens": "35mm equivalent",
      "aperture": "f/2.0",
      "quality_tags": [
        "film grain",
        "flash highlights",
        "slight noise",
        "natural depth of field",
        "realistic skin texture"
      ]
    }
  }
}

Transparent Glass Social Media Profile Effect

profile-avatar

A prompt template for Nano Banana 2 designed to create an image featuring a transparent glass effect overlaying a social media profile, inspired by another user's work. The user is instructed to screenshot and use the shared prompt.

展开提示词全文
Transparent glass effect with your social media profile

Nightlife Flash Mirror Selfie Prompt

profile-avatar

A structured prompt for Nano Banana 2 to generate an ultra-photorealistic mirror selfie taken in a luxury hotel elevator at night, simulating an iPhone flash effect with hard highlights, deep shadows, and a playful, dominant expression.

展开提示词全文
{
  "meta": {
    "camera": "iPhone 17 Pro",
    "lens": "48mm",
    "aspect_ratio": "9:16",
    "quality": "ultra photorealistic",
    "style": "flash nightlife realism, slight grain, imperfect highlights"
  },
  "scene": {
    "location": "luxury hotel elevator",
    "time": "2:21 AM",
    "atmosphere": "post-party silence, reflective metal walls"
  },
  "lighting": {
    "type": "iPhone flash + elevator LEDs",
    "temperature": "mixed",
    "effect": [
      "hard flash highlights",
      "deep shadows",
      "mirror reflections"
    ]
  },
  "subject": {
    "pose": {
      "position": "standing near mirror",
      "hips": "pushed slightly back",
      "one_hand": "holding phone low",
      "other_hand": "touching neck"
    },
    "expression": {
      "eyes": "direct eye contact",
      "mouth": "subtle smirk",
      "energy": "playful dominance"
    },
    "hair": {
      "style": "high ponytail with loose strands"
    },
    "outfit": {
      "top": {
        "type": "tight black crop top"
      },
      "bottom": {
        "type": "micro mini skirt"
      }
    }
  },
  "camera": {
    "angle": "mirror eye-level",
    "distance": "mid shot",
    "imperfection": "flash glare"
  }
}

High-Fidelity Mirror Selfie Prompt

profile-avatar

A highly detailed prompt for Nano Banana 2 focusing on generating a raw, high-fidelity mirror selfie with an iPhone simulation, emphasizing strict identity lock, realistic skin textures (micro-pores, vellus hair), and specific fashion details like a marble print dress and gold accessories.

展开提示词全文
(9:16) Raw high-fidelity mirror selfie, eye-level, medium shot, iPhone 17 Pro sim, 35mm, f/2.8, sharp subject, natural grain.

Framing: Subject ~65% frame, centered in mirror, large wooden door background, slight white wall edge, ~0.6m distance.

Identity Lock: Preserve ALL facial features (structure, skin, hair, marks).
Enhancement: Natural skin clarity (visible pores), even tone, soft glow, subtle light-based contour, refined highlights (eyes/lashes/brows), salon hair.
Bio-Fidelity: Micro-pores, fine vellus hair, hydrated skin.
Expression: Confident/poised gaze into phone via mirror, lower face partially occluded by phone.

Hair: Same color/texture, long sleek waves, middle part, behind shoulders + over left shoulder, warm highlights.
Makeup: Polished (filled brows, eyeliner, mascara), neutral nails.

Accessories: Gold watch (left), layered gold necklaces, gold ring (right), beige/gold iPhone, small off-white handbag (left hand).

Outfit: Strapless rib-knit bodycon dress, abstract marble (charcoal/olive/cream), tight fit, realistic folds, defined S-curve.

Pose: Torso angled ~30° left, hips toward mirror, right hand holding phone at face level, left holding bag, natural articulation.

Mood/Color: Chic, confident; palette (mahogany, charcoal, olive, cream, gold), ~3500K warm indoor light, moderate contrast.

Environment: Indoor dressing/hallway, mahogany door w/ silver handle, white wall edge, soft overhead lighting, all sharp.
上一页 73 / 176 下一页

提示词模型无关,可直接用于 Midjourney / DALL·E / Flux / Stable Diffusion 等任意文生图模型。

依赖说明

- Node.js 18+
- 无需任何 API Key(数据全部公开)
- 提示词已完整同步到本地与数据库,站内检索不依赖第三方

适用平台

claude-codecursorcodexgemini-cliopenclaw

更新记录

官方源监视中
内容更新 2026-10-21

内容有更新(细节请以官方文档为准)

内容更新 2026-10-20

内容有更新(细节请以官方文档为准)

内容更新 2026-10-19

官方源有内容更新(首次纳入监控)

信息边界说明

本页信息摘自官方文档,仅作技术信息呈现,不代表本站的判断或推荐。价格、配额、可用性以官方最新说明为准。你离开本站后的任何行为(申请 Key、付费、部署、商用),本站不做任何承诺与担保。

同类推荐

视频短视频图片音乐音频配音口播数字人全能创作

官方标注收费

通过同一个AI创作工具完成AI图片、视频、音乐与语音创作,查找公开社交数据,编辑视觉成果,并集中查看和管理生成资产。Beatra 全能创作支持AI图片生成(文生图)、AI视频生成(图生视频)和AI视频编辑,通过AI音乐生成把想法做成歌曲或纯音乐,也可用AI语音生成(文字转语音)、多语言旁白和AI语音克隆完成跨媒介内容;还能查找抖音数据、小红书数据,以及 YouTube、TikTok、Instagram 与 X 上的公开帖子、笔记、评论、账号和趋势,把社交媒体数据用于竞品调研和创作。它是一款面向AI内容创作与多模态内容创作的AI创作套件,用起来就是一个把图片、视频、音乐与语音收在同一处的AI创作助手,适用于社交内容、产品展示、营销素材、课程、播客、短片和跨媒介项目;还能在不同媒介间复用原始素材、已有成果和查到的公开资料,查看制作进度、返回文件、实际用量和积分记录。

AI数字人口播与虚拟主播视频

官方标注收费

用一张人像和一段短脚本或语音,制作有明确讲述内容的数字人口播视频。可使用现成录音,也可以先选择音色,把脚本生成旁白,让画面中的人物随声音完成稳定、自然的口播。适合 AI 主播、虚拟主播视频、数字人讲解、让照片说话、照片说话视频、产品讲解视频、数字人培训视频、公告和社交口播,并在交付后检查人物一致性、声音清晰度、口型同步和动作稳定性。

架构图一键生成

官方标注收费需申请 Key

使用 ContextWeave 生成或修改架构图、流程图、思维导图和复杂信息图。适用于需要从代码、文件或自然语言中提取结构与关系并实际产出 CW、SVG、HTML、可编辑 PPTX 或原生 VSDX 的请求;未特别说明时,PPTX 与 Visio 使用原生形状和连接器导出。不适用于统计图表、手绘插画或像素级排版。

罗斯蒙德可视化图形生成

官方未说明

【罗斯蒙德可视化图形生成】南京罗斯蒙德基于已交付项目沉淀的可视化图形生成场景:根据业务描述或系统需求,自动生成可编辑的架构图、流程图、ER 图、时序图、状态图与 C4 图。采用「统一图模型」一次建模、多格式输出:结构化文本(Mermaid / PlantUML,可嵌入 Markdown、进 Git 版本管理)、可编辑图形文件(drawio / Excalidraw,可拖拽修改节点与样式)、静态图(SVG 自渲染,PNG 给转换命令),并输出图形说明(图的含义 + 关键节点解释)。内置样式系统统一配色与命名,内置布局引擎自动排布,校验脚本拦截孤立节点、悬挂连线与语义缺漏。触发词:画架构图、画流程图、画 ER 图、画时序图、画状态图、C4 图、系统架构图、部署架构图、数据流程图、实体关系图、生成 drawio、生成 Excalidraw、导出 mermaid、导出 PlantUML、导出 SVG、导出 PNG、可编辑的图、图转可编辑、架构图可编辑、流程图二次修改、画图、出图、画个图。

海报设计工作室

免费增值需申请 Key

将一段活动描述、一张产品照片或品牌参考图,转化为抓人眼球的活动海报、宣传 banner 或社交媒体配图。这款 AI海报设计与制作工具生成图片丰富的海报,搭配清晰视觉层级、按品类匹配的风格和预留文字区域,覆盖活动、音乐节、电影、产品发布和促销场景。你可以把它当作 AI宣传单制作工具来制作促销宣传单,当作活动海报设计工具,当作电影海报生成器,当作音乐节海报设计工具,当作产品发布会海报的 banner设计 AI,或当作平面设计海报工作室。从活动主题出发,用品牌风格参考图变换产品或场景照片,或在已认可草稿上做定向精修,完成可印刷或可分享的成品。

架构图一键生成

官方标注收费需申请 Key

使用 ContextWeave 生成或修改架构图、流程图、思维导图和复杂信息图。适用于需要从代码、文件或自然语言中提取结构与关系并实际产出 CW、SVG、HTML、可编辑 PPTX 或原生 VSDX 的请求;未特别说明时,PPTX 与 Visio 使用原生形状和连接器导出。不适用于统计图表、手绘插画或像素级排版。

什么是技能

技能(Skill)是封装好的单项能力,比如文字识别、文档摘要、关键词提取。开发者把调用逻辑和提示词打包,使用者像用插件一样加载即可,无需了解底层模型。启枢技能 收录大量技能,可直接试跑或编入方案。