热门角色不仅是灵感来源,更是你的效率助手。通过精挑细选的角色提示词,你可以快速生成高质量内容、提升创作灵感,并找到最契合你需求的解决方案。让创作更轻松,让价值更直接!
我们根据不同用户需求,持续更新角色库,让你总能找到合适的灵感入口。
生成评估技能的创意设计,适用于艺术性任务场景。
Assessment prompt: Write a creative brief for a Stable Diffusion image campaign Scene-setter The city is a rain-slicked mirror just past dusk, sodium lamps breathing amber into puddles while a salt-kissed wind carries hints of citrus and incense. Down at the waterfront, a pop-up night festival unfurls—lanterns blink like new constellations, stalls hum with music and chatter, and a hush of bioluminescent glow threads the air like spilled stardust. The client wants imagery that feels like stepping into warm mist: sensory, cinematic, and human—faces lit by neon, textiles fluttering like tide foam, metal and glass beaded with rain. The work will anchor a launch campaign for “Lumen Tide,” a coastal culture festival celebrating food, craft, and sustainability. Your task Write a professional creative brief that will guide a team using Stable Diffusion to generate a coherent set of hero and supporting visuals. Your brief should be as clear as a blueprint and as evocative as a story, translating brand intent into promptable, reproducible directions. What your brief must include - Purpose and KPIs - Describe the marketing objective in one sentence, then set measurable goals (e.g., poster recall, social engagement, event sign-ups, CTR). State how success will be measured and in what timeframe. - Audience insight - Define primary and secondary audiences with 1–2 motivating insights each. Evoke context—what they see, hear, smell, crave at night by the water—and why this imagery compels them. - Core message and emotional arc - Distill the single-minded proposition and the feeling arc you want: anticipation → immersion → belonging (or your chosen arc). Identify 3 sensory motifs to carry through all assets (e.g., “lantern-amber skin tone,” “sea-glass teal reflections,” “bioluminescent threadlines”). - Visual strategy - Subject focus, environments, time-of-day, weather, and lensing. Specify lighting language (e.g., wet speculars, soft rim light from lanterns, deep indigo shadows), color palette boundaries (primary/secondary/neutral), depth, and texture cues (porous stone, lacquered wood, rippled water). - Composition guidance for negative space and copy-safe areas (avoid relying on baked-in text; reserve space for typeset overlays). - Style references - Reference styles via movements, materials, and descriptive attributes rather than living artist names (e.g., “neon noir,” “cinematic night photography,” “bioluminescent marine macro,” “hand-poured resin gloss”). If citing public-domain influences, note them as conceptual anchors. - Brand and ethics guardrails - Inclusivity guidelines, no stereotyping, no depiction of real private individuals without consent, avoid IP-infringing motifs, and avoid unsafe content. Accessibility notes (contrast targets, legibility, color-blind-friendly palette). - Deliverables and specs - Key visual poster, 3840×2160, landscape, print-safe and digital-ready. - Social story triptych, 1080×1920 each, vertical; panel-to-panel continuity. - Square social teaser, 1080×1080; thumbnail-first clarity. - Seamless pattern tile, 1024×1024, for banners and merch. - Optional product/food close-up, 2048×2048, macro texture emphasis. - Stable Diffusion production plan - Model: recommend Stable Diffusion XL 1.0 for composition and detail (note: if constrained to SD 1.5, call out limitations and compensations). - Sampler and steps: recommend a high-quality sampler (e.g., DPM++ 2M Karras or DPM++ SDE Karras). Typical steps 25–40 for base renders; document chosen step count. - CFG scale: propose a range (e.g., 4.5–7.0) and when to adjust for fidelity vs. freedom. - Resolution: per deliverable above; specify whether to render native or upscale. If upscaling, note tool/method (e.g., high-res fix or external 2× upscaler) and denoise strength. - Seed strategy: assign fixed seeds for each “scene” to ensure reproducibility; use variation seeds for alternates. Note: reproducibility requires same model, VAE (if applicable), sampler, steps, CFG, and resolution. - Prompt structure template: define a consistent slot order to reduce drift, e.g.: Subject + Environment + Time/Weather + Lighting + Color Palette + Mood + Composition + Camera/Lens + Material/Texture + Style Descriptors + Quality Tags + Negative Prompt - Negative prompts: specify common artifact filters (e.g., low quality, blurry, extra limbs, distorted hands, text, watermark, logo, signature, jpeg artifacts) plus any scene-specific exclusions. - Consistency tools: define if/when to use ControlNet (Canny for layout, Depth for scene coherence, OpenPose for human pose), reference image adapters, or a custom brand-motif LoRA (only with cleared training data). Include weights/strength ranges and when to lower/raise for diversity. - Post pipeline: optional SDXL Refiner pass for detail, targeted inpainting for hands or faces, color match to brand palette, final grain/halation pass for cohesion. - Shot list - 5–8 scene briefs, each a few lines: the who/what/where, sensory goal, composition, and technical parameters (AR, seed, sampler, steps, CFG). Ensure a throughline motif appears in all shots. - File and metadata hygiene - Naming convention, versioning, and embedded PNG info (prompt, negative prompt, seed, model, sampler, steps, CFG). Note source and rights for any references. - Risk and mitigation - Identify likely failure modes (e.g., text artifacts, hand anatomy, rain-bokeh overpowering faces, poster mudding at small sizes) and your plan (e.g., tighter negative prompts, pose control, inpainting, manual type set). - Feasibility and timeline - A pragmatic mini-schedule from concept test renders to final selects, including iteration checkpoints and approval gates. Include one example prompt Provide exactly one fully written Stable Diffusion prompt (plus negative prompt and parameters) for the key visual poster to demonstrate how your brief translates to generation. Use your template and include: model, sampler, steps, CFG, resolution, seed, and any ControlNet/LoRA notes. Keep it within the constraints you set. Scenario constraints you must honor - Mood: rain-wet night; lantern/LED warmth vs. deep indigo/cyan; bioluminescent hint, not sci-fi overpowering. - People: diverse, candid, joyful calm. No celebrities, no minors as primary subjects. - Text: do not rely on the model to render legible typography. Leave obvious negative space for later typesetting. - IP and safety: no copyrighted logos or brand marks, no imitation of living artists’ unique styles, no unsafe or sensitive content. - Platforms: ensure the key visual can crop cleanly to 1:1 and 9:16 without losing the subject. Evaluation criteria - Clarity: Is the brief actionable by a prompt artist without a meeting? - Consistency: Do directives ensure visual coherence across deliverables? - Specificity: Are sensory and technical details concrete enough to prompt? - Feasibility: Are model and parameter choices realistic for SDXL-era workflows? - Originality: Does the creative direction feel fresh yet on-brand for a coastal night festival? - Rigor: Are risks anticipated with credible mitigation plans? - Translation: Does the example prompt reflect the brief’s intent and constraints? Tone and length - Aim for 900–1,400 words. Write with sensory precision—let the reader smell the rain on warm stone, see the amber halo on skin, hear the soft hiss of lanterns—while staying operationally exact.
以下提示词旨在生成一幅信息密度高、叙事饱满、优点与瑕疵并存的“展览现场”图像,专门用于考察写作者对作品进行全面点评的能力(叙事理解、构图与光色分析、材质与技法描述、细节洞察与改进建议等)。提示词以强烈的感官描绘引导画面充满可被评论的“钩子”,便于写作者进行多维度评析。 一、主提示(适用于SDXL等通用模型) - 正向提示(复制使用): 独立小型画廊的黄昏开幕,金橙余晖从高窗斜落,木地板泛起温暖光泽;中央墙面挂一幅巨大的厚涂油画:翻涌的青绿海浪与孤伫的白色灯塔相抗,刀痕与堆砌笔触清晰,光泽随角度微变;画下玻璃展柜内,一枚旧黄铜罗盘蒙着浅浅雾气,氧化斑驳微绿;左侧角落竖着一根冷白霓虹管装置,发出清冷嘶嘶声般的光感,与油画温橙色调撞色,形成色温冲突;观众三人错落而立:年轻观者微皱眉、略前倾,手中折皱的宣传册在指尖投出细碎阴影;一位银发老妇轻捻耳环,目光柔和却专注;一位策展人半隐于暗部,胸前证件反出淡蓝高光,神情克制;地面有一小滩洒落的红酒,酒红在地板纹理间渗开,映出天花板灯具的破碎倒影;作品说明牌微微倾斜,字距不匀,纸角轻翘;空气中尘粒在侧逆光里细闪,远处黑帘被空调风拂起,边沿出现轻微运动模糊;构图采用三分法与对角线引导,灯塔与霓虹形成视觉张力,视线被地板木纹与光带牵引至画心;色彩主调为青绿与橙金,点缀一抹深红作为视觉制动;质感饱满:油彩厚涂纹理、玻璃细微划痕、金属氧化斑、木纹磨损;摄影参数感:35mm、f/2.8、景深层次分明、低对比胶片颗粒、轻微镜头畸变与边缘色散、柔和暗角;高动态范围但控制高光不过曝,真实室内反差,写实而富叙事张力,cinematic lighting, volumetric dust, gallery interior, storytelling - 负向提示(复制使用): 过度噪点、强烈AI水印、过曝高光、死黑阴影、低分辨率、极端畸变、面部崩坏或融化、手指数量错误、额外肢体、严重鬼影、重复人脸、涂抹感皮肤、脏污伪影、过强色带、莫名logo、清晰可读的大段文字、构图截断脸部 - 推荐参数(可按环境微调): 采样器:DPM++ 2M Karras(或同级高保真采样) 分辨率:1024×768(横构图)或768×1024(竖构图) 步数:28–36 CFG:5–7 种子:固定一个随机数以便复现(如 123456789) Hires/Refiner:可关闭;若开启,放大1.5–2.0,轻锐化 二、难度调节与可点评“钩子”开关(按需拼接到正向提示末尾) - 提高叙事复杂度:在油画表面加入细小盐析纹与裂纹;观众之间增加不经意的眼神交汇;远处门外一抹路灯影透入地面。 - 增加光色挑战:引入第三光源(天花暖黄卤素灯),制造多光源阴影方向与色温混合。 - 反射与透视测试:玻璃展柜中叠加两层反射,隐约映出观者的模糊侧影与窗格线,地板反光呈轻微桶形畸变。 - 文本元素控制:说明牌仅保留几笔无法完整辨读的假字形,避免真实文字分散注意但保留排版感。 - 技术瑕疵留白(便于点评):保留轻微边缘光晕、细小色散、微热噪;避免大面积故障感。 三、针对不同点评侧重的快速变体(任选其一替换场景主体) - 人像叙事变体(侧重表情、肤色、布料与高低光衔接): 暮色画廊改为“模特与画作同框”情境:模特半身靠墙,粗呢外套与丝质围巾材质对比,颧骨处硬边侧光,唇部微湿亮点,眼下细微反光,背景仍保留霓虹与油画的色温冲突与反射难题。 - 静物/产品变体(侧重材质、微细节与色彩和谐): 展柜内增加古书与裂纹白瓷小碟,瓷面釉裂清晰,玻璃上有指纹与雾痕;侧逆光勾出高光边,背景霓虹冷蓝与书页暖黄互补。 - 城市/空间变体(侧重构图与节奏): 把画廊外延伸至雨后街口,可透过落地窗看见湿地反光、远处红伞与蓝牌灯交错,室内暖、室外冷的透窗对比强化透视与层次。 四、使用建议(助力评测一致性) - 固定种子生成标准样张;更换少数开关词以控制难度梯度,避免话题偏离。 - 若需更强“可议之处”,适度提高场景中的冲突元素(冷暖、粗细、亮暗、动静、旧新),但避免过度夸张导致失真。 以上提示词可直接用于生成用于点评写作训练的图像:画面具备清晰叙事线索、复杂光色关系、丰富材质层次与有限可接受的技术瑕疵,能有效引导写作者进行多维度、深层次的作品点评。
下面是一条专为Stable Diffusion/SDXL设计的可操作提示词,用于生成一张“教学型色彩搭配评估海报”。它像一块被晨光打亮的画室墙面:九格井然,静物在每个格子里保持一致,只是色彩法则换了呼吸与温度。观者能在同一构图里,感受从海盐蓝到石榴红的流转,听见饱和与明度像潮汐推拉,嗅到亚麻与木面的清淡气息。此图可用于考察被试对色彩和谐、主次分配、明度/饱和度/色温对比等原则的阐述能力。 主提示词(可直接复制使用) - 中文版本: 一张教学型色彩理论海报,3x3九宫格,同一简洁静物在九个格子中重复出现(白瓷碗、一只柠檬、一支玻璃瓶、折叠亚麻布放在木板上),正面视角,构图稳定,软质棚拍光;画面为干净的平面化海报风,清晰边缘,微妙纸纹,哑光质感,高分辨率,无文字无数字。自左上到右下九格分别呈现不同配色法则,仅色彩方案变化,造型与光位基本一致: 1 单色调(蓝):以群青/深海蓝为主,明度层次清晰,背景冷灰中性化; 2 类似色(暖):红—橙—黄范围柔和过渡,以琥珀为主,点缀一处极小冷色; 3 互补色:青绿 vs 珊瑚橙,主次比例约60(青绿)/30(中性灰)/10(珊瑚),形成明确对比; 4 分裂互补:以紫为主,黄绿与黄橙为两侧辅色,平衡张力与和谐; 5 三分配色:红、黄、蓝等距分布,中等饱和,三色清晰分区不混浊; 6 四方形(矩形)配色:蓝–橙与红–绿两对互补,一色主导,其余为辅与点睛,并以中性灰舒缓; 7 冷暖对比:主体受暖光(偏金),背景冷青板岩色,温度反差鲜明但不过曝; 8 明度对比优先:限定色相,强调强烈黑白灰结构,背景较暗,焦点物体高明度显形; 9 饱和度对比:整体低饱和雾化,唯有一处小面积高饱和亮点牵引视线; 每格下方附一条对应色板(3–5个色块)显示该格主色与辅色;整体整洁对称,均匀留白,高级感,印刷级细节。无文字、无数字、无水印、无Logo。 - 英文辅助(可与中文混用提高泛化): color theory teaching poster, 3x3 grid, same simple still life repeated in each cell (white ceramic bowl, a lemon, glass bottle, folded linen on wood), front view, soft studio light, flat poster design, crisp edges, subtle paper texture, matte finish, high resolution, no text no numbers. From top-left to bottom-right apply: 1 monochromatic blue, 2 warm analogous (red–orange–yellow) with tiny cool accent, 3 complementary teal vs coral 60/30/10, 4 split complementary violet with yellow-green and yellow-orange, 5 triadic red–yellow–blue mid saturation, 6 tetradic rectangle blue–orange & red–green with one dominant hue and neutrals, 7 warm–cool temperature contrast (warm key on subject, cool cyan-slate background), 8 value-contrast-first with limited hues and strong light–dark pattern, 9 saturation contrast with muted field and a single vivid accent. Each cell has a small palette strip beneath (3–5 swatches). Clean, symmetrical, print-ready. 负面提示词 - 英文负面(常用更稳):text, letters, numbers, watermark, logo, signature, messy layout, extra panels, distorted objects, low contrast, oversaturated everywhere, noisy background, gradients banding, chromatic aberration - 中文补充:错误文字、脏污、过度锐化、混乱笔触、版式拥挤、色块溢出 参数建议(SDXL) - 分辨率:1024×1024 或 1536×1536(更利于细节与干净边界) - 采样器:DPM++ 2M Karras(或同级稳定采样器) - 步数:28–40 - CFG:4.5–7(海报风建议中等) - 种子:固定以便复现;生成多张后筛选边界最清晰者 - 风格提示可加权:poster design(1.2), clean layout(1.2), crisp edges(1.2), soft studio light(1.1) 可选微调(按需拼接到主提示词) - 更几何的构图网格:thin white dividers between panels, equal spacing - 更强的平面化:vector poster, minimal shading, shape-based color blocking - 更具材质对比:subtle linen texture on cloth, matte ceramic glaze, faint glass refraction - 若出现文字幻写:add stronger negative “no text, no letters, blank labels” 使用与评估建议(与提示词配套) - 生成后,要求被试逐格阐述:配色关系识别(单色、类似、互补、分裂互补、三分、四方形、冷暖、明度、饱和度);主—辅—点睛比例(如60/30/10)与中性色的缓冲作用;明度/饱和/色温如何影响层次、焦点与情绪;为何相同造型在不同配色下呈现不同空间感与节奏。 - 合格图像特征:九格主题与光位基本一致,差异主要来自颜色组织;色板条清楚、无文字;互补/三分/四方形等关系一眼可辨,不“跑调”。 这条提示词在视觉上像将色轮拆成九首乐章:有的低声呢喃(单色与低饱和),有的铜管嘹亮(互补与三分),有的以温度升降构出昼夜气息(冷暖对比)。面向评估场景,它既提供可比对的统一构图,又把色彩原则化作触手可及的图像证据,便于被试用准确术语与清晰逻辑加以阐述。
快速生成标准化创作考题,按岗位侧重定制要求,多语言发布;对候选人进行同题对比,缩短筛选周期并提升用人准确度。
为课堂作业与期末考核创建情境化题面,分层设置难度;沉淀题库与评分要点,提升教学效率与评估公正性。
围绕色彩、构图、风格切换等模块批量出题,构建实训任务库;跟踪学员阶段表现,形成可展示的项目作业集。
为Stable Diffusion创作设定明确题面与限制条件,测试风格库表现;批量生成A/B实验素材,统一创作基线。
用定向评估题自测短板,针对目标职位训练关键能力;跨题材练习并产出高质量案例,迭代个人作品集。
在活动视觉与品牌探索前,通过提示题快速验证方向;与设计同事共享题面,减少沟通成本并提升落地速度。
将“要评估的某项技能或能力”快速转换为一条可直接用于AI图像生成(如Stable Diffusion)的高质量测评任务描述。通过专家化的表达与生动的感官描绘,输出既专业聚焦又富有画面张力的创作指令,帮助招聘、教学与团队训练场景快速搭建标准化评测题,降低出题成本,提高评估的一致性与可比性,并支持多语言输出,便于跨团队与对外发布。
将模板生成的提示词复制粘贴到您常用的 Chat 应用(如 ChatGPT、Claude 等),即可直接对话使用,无需额外开发。适合个人快速体验和轻量使用场景。
把提示词模板转化为 API,您的程序可任意修改模板参数,通过接口直接调用,轻松实现自动化与批量处理。适合开发者集成与业务系统嵌入。
在 MCP client 中配置对应的 server 地址,让您的 AI 应用自动调用提示词模板。适合高级用户和团队协作,让提示词在不同 AI 工具间无缝衔接。
免费获取高级提示词-优惠即将到期