生成虚构场景描述

7 浏览
1 试用
0 购买
Oct 1, 2025更新

根据指定角色或情境创作生动的虚构场景,文字富有表现力。

示例1

以下内容包含两部分:1) 一段具象的场景描写,帮助你把握“雾雨港口的灯笼女巫”意象与氛围;2) 可直接用于Stable Diffusion/SDXL的专业提示词与参数建议(含负面提示词与多种构图版本),以确保画面在“叙事性、角色设定、质感与光影”上稳定地呈现。

一、场景描写(灵感基调)
雨丝像细针缝补着夜色,港口被雾压得低低的,潮腥味从缆绳和旧桅杆的木纹里缓缓渗出。棕褐色的木栈道浸水发亮,灯影在水洼里被划成流动的鳞片。远处,拖网渔船的铁壳轻轻撞击码头,沉闷的回声与雾号混成一口冷气。她在尽头出现——灯笼女巫,成人的身形被油布风衣包裹,湿发像水草贴在颊侧。她肩上横一只失了光泽的黄铜横担,串着大小不同的灯笼,蜡泪在金属边缘凝出白色的硬刺;暖琥珀色的火核在玻璃筒里呼吸,像海底的小心脏。她行过之处,冷蓝的雾被轻轻拨开,像潮水让道,露出皮革缝线与布料的盐霜。裙摆下沿绣着潮汐符阵,微光一息一闪,与她吐出的白雾一同升腾——咒语被雨剪成短促的音节,又被风缝回夜里。码头护栏外,水面翻出一圈圈暗绿的磷光,像被灯火唤醒的鱼群在深处转身;更远的黑浪上,零碎的人影与海鸟的影子重叠,像归航也像送别。她举高一盏盏灯,温暖与寒意在她的指间对峙,光像绸带穿过雨柱,在雾中打出一条可供灵魂靠岸的航路。

二、Stable Diffusion 提示词与参数
说明:以下为SDXL通用写法。若使用不同前端,注意权重语法可能略有差异(如Automatic1111支持( )权重)。你可从“电影海报”“角色立绘/设定图”“半身氛围肖像”中选择一个或多个版本。

1) 电影海报式全景(叙事主场景,16:9)
正向提示词:
masterpiece, best quality, cinematic realism, nighttime foggy rain harbor, drenched wooden pier, slick cobblestones, rusted trawlers moored, thick cold mist, rain streaks, puddle reflections, sodium-vapor lamps bloom, sea ropes and wooden crates
(lantern witch:1.4), adult female occult mariner, oilskin coat over layered skirts, weathered wide-brim witch hat with talismans, tarnished brass lantern cluster on a yoke, dripping wax, bioluminescent runes on hem, salt crystals on fabric, wet hair strands, steam breath
guiding drowned spirits as faint silhouettes over water, spectral gulls, green-blue phosphorescence swirling, subtle ghostly wake
lighting: warm amber lantern glow vs cold cyan fog, volumetric light through rain, wet specular highlights, cinematic contrast
composition: low-angle three-quarter, leading lines along pier, rule of thirds, deep depth, 35mm lens, subtle film grain, raindrops on lens, high detail textures, moody color grading (teal and amber)

负面提示词:
lowres, worst quality, jpeg artifacts, text, watermark, signature, extra fingers, extra limbs, deformed, mutated hands, poorly drawn hands, bad anatomy, out of frame, oversaturated, cartoonish, chibi, flat lighting

参数建议:
- 模型:SDXL 1.0(或等价高质量写实模型)
- 采样器:DPM++ 2M SDE Karras
- 步数:28–36
- CFG:4.5–6.5(SDXL建议偏低以保留细节)
- 分辨率/比例:1024x576 或 1216x704(16:9)
- 种子参考:137 或 20250101(可固定以复现实验)

可选增强:
- ControlNet Canny/Depth:锁定码头与人物的主线条,降低漂移
- 天气强度微调:heavy rain, windblown rain, thicker fog(按需替换)
- 光效微调:godrays, light shafts, bloom subtle(避免过曝)

2) 角色立绘/设定图(干净背景,便于做角色卡,3:4)
正向提示词:
masterpiece, character sheet, full body, neutral grey studio background, soft rim light
(lantern witch:1.4), adult woman, calm determined gaze, oilskin coat, layered salt-stained skirts, weathered wide-brim witch hat with pinned talismans, brass lantern cluster (different sizes) on shoulder yoke, wax drips, belt pouches, rope, small knife, wet fabric microfibers, stitched runic hem glowing faintly
front view, slight contrapposto, boots wet sheen, hands visible holding lantern, detailed textures, realistic materials, studio shadow on floor

负面提示词:
text, labels, diagram arrows, watermark, logo, extra limbs, extra fingers, deformed hands, bad anatomy, busy background, low detail

参数建议:
- 分辨率/比例:896x1152 或 1024x1344(3:4)
- 采样器/步数/CFG:同上
- 提示:若想要“多视图回转”,可改写为 “character turnaround, front and back views, orthographic, prop close-ups”,并加强负面词中的text/labels,结果随机性较高

3) 半身氛围肖像(近景情绪,1:1 或 4:5)
正向提示词:
masterpiece, moody portrait, chest-up, shallow depth of field, 85mm lens, rain droplets on lens, bokeh
lantern witch, adult, rain-wet strands of hair across cheek, amber lantern glow on skin, cyan fog backlight, steam breath, freckles, raindrops on oilskin, reflective highlights, bioluminescent runes soft glow, cinematic color grading, intimate framing

负面提示词:
lowres, blurry eyes, cross-eye, overexposed, text, watermark, extra fingers, bad hands, neck artifacts

参数建议:
- 分辨率/比例:1024x1280(4:5)或 1024x1024(1:1)
- 步数/CFG:同上

素材与细节控制词(可选拼接)
- 材质细节:salt-stained fabric, frayed edges, patina on brass, rain beading, wet wood grain, fine leather pores
- 海港元素:mooring bollards, coiled ropes, barnacles, distant lighthouse beam, gulls silhouettes, tugboat haze
- 魔法/叙事:faint spirit silhouettes, phosphorescent tide, sigils flicker, candle smoke mingling with rain

风格切换(按需替换末尾风格短语)
- 写实电影感:cinematic realism, natural color science, subtle film grain
- 绘画质感:painterly, oil-and-ink blend, impasto brushwork, textured canvas
- 黑暗奇幻插画:dark fantasy illustration, high octave contrast, dramatic chiaroscuro

常见问题与小贴士
- 灯与雾的对比是氛围关键:用“warm amber vs cold cyan”锁定冷暖对冲;将“volumetric light”与“wet specular”同时出现能显著增强雨夜质感。
- 手部质量控制:负面提示里优先压“extra fingers, bad hands”,必要时加手部LoRA或局部重绘。
- 反文字污染:若不希望出现随机标签或水印,确保负面提示包含 text, watermark, signature,并避免“poster, magazine cover”等易诱发文字的词。

直接复制所需版本的正向与负向提示词到你的UI中,结合建议参数即可开始迭代。通过固定种子与微调“雨量、雾厚、光比”,你可以在同一角色设定下,快速产出叙事场景、立绘与肖像的成组作品。祝你在雾雨与灯火之间,捕获那个在港口为灵魂点路的身影。

示例2

Night Run, Weightless Impact — hero visual for a lightweight running shoe promo

Creative brief (evocative guide)
Rain freckles the asphalt like scattered stardust. A lone runner cuts through the velvet-blue night, each footstrike sparking neon echoes across wet streets. The shoe hangs in a razor-sharp freeze-frame against a river of magenta-cyan light trails; breath turns to silver mist; a halo of rim light carves the silhouette as if speed itself were sculpting the air. The city becomes a blur of electric calligraphy, and the shoe—featherlight, aerodynamic, all breath and mesh—feels like a comet skimming the ground.

SD prompt set A — Product hero, cinematic freeze-frame (primary key visual)
Positive prompt:
ultra-detailed lightweight running shoe, airy knit mesh upper, micro-perf ventilation, sculpted midsole, translucent outsole with fine grooves, dynamic laces, zero-gravity feel, mid-stride freeze over rain-slick street, neon city night, cyan and magenta light trails, long-exposure streaks behind, droplets suspended around the shoe, crisp specular highlights on wet asphalt, dramatic rim light, volumetric mist, deep bokeh skyline, rule of thirds composition, generous copy space to the left, cinematic contrast, photoreal product photography, 85mm lens, shallow depth of field, f/2.0, sharp subject, background motion blur, hyper-clean edges, no logo, no text

Negative prompt:
text, logo, watermark, brand mark, duplicated shoe, extra soles, warped laces, deformed, melted, low-res, noisy, sketch, cartoon, flat lighting, overexposure, underexposure, blown highlights, muddy blacks, incorrect perspective, lens distortion, bad reflections, artifacts, motion jitter on the shoe, soft focus on shoe

Suggested parameters:
- Model: SDXL (recommended) or SD 1.5
- Sampler: DPM++ 2M Karras
- Steps: 30–40
- CFG: 6.5–7.5
- Aspect ratio: 16:9 (hero banner) or 9:16 (story/reel)
- Seed: set a fixed seed for reproducibility; explore ±50 around the base seed
- Hires pass: 1.5–2x, denoise 0.30–0.40, Latent or 4x-UltraSharp upscaler

SD prompt set B — Lifestyle momentum shot (runner in scene, dynamic and emotive)
Positive prompt:
night run through neon-lit city, low-angle tracking shot of a runner wearing lightweight running shoes, mid-stride sprint, wet pavement mirror reflections, cyan-magenta neon signage glows, long shutter drag trails behind limbs, sharp shoe silhouette with crisp rim light, breath vapor, rain mist catching light, raindrops splashing at footstrike, moody cinematic fog, selective focus with depth map, rich contrast, dynamic composition, leading lines, 35mm lens, sense of speed and determination, copy space above and to the right, no logo, no text

Negative prompt:
text, logo, watermark, brand mark, extra limbs, deformed feet, duplicated shoes, motion blur on shoe, plastic skin, waxy, low-res, flat, overprocessed, bad anatomy, incorrect pose, distorted perspective, cluttered background

Suggested parameters:
- Model: SDXL or SD 1.5 + ControlNet OpenPose (for precise stride pose)
- Sampler: DPM++ SDE Karras
- Steps: 28–36
- CFG: 6–7
- Aspect ratio: 21:9 (cinematic) or 4:5 (poster)
- ControlNet: OpenPose for runner posture; Depth or Canny to lock street perspective

SD prompt set C — Abstract CGI speed concept (campaign support visual)
Positive prompt:
featherlight running shoe dissolving into ribbons of neon cyan and magenta, speed vectors, aerodynamic airflow lines, translucent mesh surfaces, microfibers lifting like feathers, energy arcs, particle trails, glossy black void with subtle fog, hero shoe centered, studio-grade product lighting, hard rim and soft fill, reflective base plane, pristine edges, futuristic minimalism, high-polish CGI, 50mm lens look, negative space for headline, no logo, no text

Negative prompt:
text, logo, watermark, clutter, low-res, jagged edges, noisy particles, plastic look, muddy reflections, messy composition, off-center crop

Suggested parameters:
- Model: SDXL or a photoreal/CGI-focused checkpoint
- Sampler: DPM++ 2M Karras
- Steps: 32–44
- CFG: 6.5–8
- Aspect ratio: 1:1 (square KV) or 3:2 (print)
- Optional: ControlNet Canny from a clean shoe silhouette to preserve geometry

Color and lighting palette (add as needed)
- Colors: electric cyan, hot magenta, graphite black, wet asphalt gray, hints of sodium-vapor amber
- Lighting: key rim light from behind, soft fill from front, practical neon sources left/right, volumetric haze at ground level, specular kicks on water droplets
- Optional tokens: volumetric lighting, chromatic bloom, high microcontrast, cinematic tonemap, wet look

Composition controls
- Place shoe on the upper-right intersection of the rule-of-thirds grid; leave negative space for headline and CTA.
- Emphasize long exposure in the background only: “sharp subject, motion-blurred background, long-exposure light trails.”
- Use “copy space, clean background, uncluttered” to avoid text-like artifacts.
- If SD drifts into signage text, reinforce “no text, no letters, no words.”

ControlNet and reference tips
- OpenPose: feed a dynamic running pose reference to nail the stride.
- Canny/Lineart: from a shoe side-profile to maintain silhouette accuracy.
- Depth: lock the street plane and shoe hover height for consistent parallax.
- IP-Adapter (if available): guide overall shoe shape while remaining brand-agnostic.

Post-processing workflow
- Hires Fix pass to regain crisp laces, mesh weave, and outsole grooves.
- Inpaint small artifacts on edges and water droplets at 0.25–0.35 denoise.
- Grade: cool cyan shadows, magenta highlights; keep skin tones neutral if a runner is visible.
- Reserve clean areas for headline, subcopy, and CTA; add branding in design tools, not in the model prompt.

Troubleshooting nudges
- If the model creates logos or signage: add “blank signage, no words, smooth panels.”
- If the shoe warps: increase ControlNet Canny weight or raise CFG slightly; reduce denoise on hires pass.
- If the scene is too dark: add “high dynamic range, bright speculars, reflective puddles, lifted shadows.”

One-line campaign hook (optional for layout, not in prompt)
Run the night. Wear the weight of nothing.

示例3

以下は、Stable Diffusionで「砂漠の王都の試練関門と石像守衛」を高い没入感で描くための、実運用向けプロンプトと調整の指針です。語彙は視覚要素を前面に、感覚描写は視界・音・温度・質感を増幅する言い回しで構成しています。

- メインプロンプト(写実・映画的)
  砂漠の王都の試練関門, 焼けつく砂と乾いた空気, 風化した砂岩の巨門, 青金石と金の象嵌が夕陽に火の粉のように煌めく, 門を挟む二体の石像守護者, 身の丈十数メートル, 黒曜石の眼が鈍く光る, 彫りの深い碑文と浸食されたレリーフ, 砂に半ば埋もれた階段, 巻き上がる砂塵が衣擦れのような音を立てる, 低い太陽が長い影を引く, 熱気の蜃気楼が輪郭をゆらめかせる, 門前に立つ小さな巡礼者, インディゴの外套と真鍮の灯火, 炎の反射が石肌を温かく舐める, 荘厳で神秘, 圧倒的スケール, 低アングルの広角構図, 24mmレンズ, ディープ被写界深度, cinematic lighting, volumetric dust, god rays, 高精細ディテール, 石のざらつきと微細な欠け, 陽炎の柔らかなボケ, filmic color grading, masterpiece, best quality

- ネガティブプロンプト
  低品質, ぼやけ, ノイズ, 過度なコントラスト, 過度な露出, 画面の切断, フレームアウト, 歪んだ透視, テキスト, ロゴ, 署名, ウォーターマーク, 現代の建築, 車, 電線, プラスチック, サイバーパンク要素, メカ, アスファルト, 不自然な光源, 手足の本数エラー, 余分な指, 変形した顔, 反復パターンの破綻, jpeg artifacts

- 推奨設定(SDXL想定、一般的な目安)
  - 解像度: 1024x1344(縦構図)または 1216x832(横広がり/門全景)
  - サンプラー: DPM++ 2M Karras または DPM++ SDE Karras
  - ステップ: 28–40
  - CFG: 5.5–7.5(石の彫りや象嵌を出したいときはやや高め、蜃気楼や空気感重視ならやや低め)
  - Hires. fix/リサイズ: 1.5–2.0倍, デノイズ強度 0.35–0.55(細部強化)
  - シード: ランダム→気に入った絵で固定
  - ControlNet(任意): Depth/Lineartで門のシンメトリと人物の位置を固定

- スタイル別バリエーション
  1) 金砂の刻印(荘厳・写実強め)
     砂漠の王都の試練関門, 巨大な砂岩の城門と左右対称の石像守護者, 青金石と金の象嵌が砂の呼吸のように微光を撒く, 彫りの深い碑文, 砂塵のヴェール, 熱気の揺らぎ, 夕陽の逆光リムライト, 巡礼者の小さな灯火が石肌に揺れる反射, 低アングル広角, 24mm, ディープ被写界深度, ultra-detailed textures, cinematic, volumetric light, masterpiece

  2) 装飾画風(平面的な華やかさ・意匠重視)
     砂漠の王都の試練関門, 金箔の縁取り, 平面的で精密な装飾文様, 豊かな幾何学パターン, 深い群青と燻んだ金の対比, 二体の石像守護者が記号のように静止, 砂塵は筆触の細い流線, 巡礼者はシルエットでサイズ対比, フラットな影と均整美, 細密描写, 装飾的輪郭強調, high detail illustration, clean linework, muted warm palette

  3) 月光の審判(夜景・神秘演出)
     夜の砂漠, 満月に洗われる試練関門, 青白い月光が砂岩を冷たく磨く, 石像守護者の黒曜石の眼が星を映す, 静かな風が砂を細い帯にして滑らせる, 巡礼者の灯火は琥珀色の点, 冷温の二色照明, 霧のような砂塵, 長秒露光のような光の残像, deep blue and silver palette, moonlit, soft shadows, specular highlights on stone, cinematic night

- 構図のコツ(生成を安定させるための語彙と指示)
  - スケール感: 「門は画面中央」「守護者は左右対称」「巡礼者は右下の三分割交点に小さく」と明示(あるいは“rule of thirds, symmetrical gate, tiny human for scale”)
  - 石の質感: 「風化した砂岩, 欠け, 亀裂, ざらつく微粒子, 浮き出た彫り」など、触覚語を複数束ねる
  - 空気感: 「砂塵のヴェール, 熱気の蜃気楼, 逆光のリムライト, 体積光, ゴッドレイ」
  - 色設計: 夕景は「焦げた金・砂色・ラピスの群青・黒曜石の黒」/ 夜景は「銀青・群青・琥珀」

- 微調整の実務ヒント
  - 守護者の威圧感が弱い→「低アングル」「広角(24mm)」「小さな人物」を強調
  - 彫りが甘い→「ultra-detailed, crisp relief, fine chisel marks, high microcontrast」を追加
  - 砂がうるさい→「粒状感は保ちつつsoft haze, gentle dust」を加え、ネガティブに「heavy grain, dirty smudge」
  - 現代物が混入→ネガティブに「modern elements, vehicles, cables, plastic」を強める

上記プロンプトは、石の冷たい重み、砂の乾いた音、光の刃のような逆光を視覚語彙で密に重ね、モデルに「質感・スケール・空気」を同時に学習させる構成です。必要に応じて、強調したい語を()で包み重みづけ(例: (青金石:1.2), (低アングル:1.3))し、狙いの要素を際立たせてください。

适用用户

独立插画师与概念艺术家

用本提示快速生成绘画提示与沉浸式场景描述,完成角色设定、气氛图与风格探索,缩短从草图到成品的距离。

品牌营销与电商设计师

一键产出促销主题场景与卖点文案,引导海报主视觉与详情页素材制作,支持多语言投放,提升点击与转化。

游戏策划与美术主管

将剧情、关卡与世界观要素转为统一风格的画面引导词,批量迭代场景概念与道具设计,加速评审与共识。

小说作者与编剧

把人物关系与冲突抽象为可视化场景,获得细腻描写与分镜提示,辅助节奏控制、伏笔铺陈与命题写作。

短视频创作者与海报设计师

生成镜头语言与氛围描述,指导构图与布光,快速做出吸睛封面、开场和主题画面,提高完播率与转化。

教育培训讲师与高校导师

以模板演示提示词写作框架,组织课堂实操与作品点评,沉淀可复用教案与案例库,提升教学效率。

产品经理与创业者

把产品卖点转为视觉故事线,生成路演主视觉与A/B方案,低成本测试用户反馈,验证市场方向。

解决的问题

将任意角色/情境,快速转化为“可直接喂给图像生成工具”的高质场景描述与专业级提示词建议,帮助创意、设计、营销与内容团队: - 快速搭建有画面感、沉浸式的虚构场景,提升出图的表现力与一致性 - 获得围绕Stable Diffusion创作的专业指导与优化建议,少走弯路、显著降低试错成本 - 按指定语言输出,兼顾国际团队协作、跨地区投放与本地化创作 - 从“概念到成片”更顺畅:用于情绪板、广告KV、封面设计、世界观搭建、脚本分镜与社媒视觉 - 建立可复用的创作标准,持续稳定地产出高质量视觉与文案素材

特征总结

一键生成以画面为中心的AI绘图提示词,精准传达主题与氛围,快速把灵感转成可落地的视觉方案
自动创作充满感官细节的虚构场景描述,声光气味与触感并呈,让读者与画面同步沉浸其中
支持多语言输出与风格切换,按需指定语言与笔触,方便跨团队协作与海外项目本地化落地
可将指定角色或情境无缝融入场景,一键延展背景动机与环境,使形象更立体更易于二次创作
提供专家级提示词写作建议与示例,涵盖构图光影与质感表达,帮助新手也能稳定产出好图
支持多版本快速迭代与风格对比,自动给出替代措辞与强化选项,让你迅速锁定最佳表达方案
内置可复用模板与可调参数位,团队可沉淀统一风格库,减少沟通成本并保持作品一致性
内置校对与约束提醒,避免跑题与信息偏差,确保描述贴合需求且符合主流创作标准
面向营销海报游戏美术与IP设定等场景,直接输出可投产文案与绘画引导词,缩短制作周期

如何使用购买的提示词模板

1. 直接在外部 Chat 应用中使用

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。

2. 发布为 API 接口调用

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。

3. 在 MCP Client 中配置使用

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。

¥10.00元
平台提供免费试用机制,
确保效果符合预期,再付费购买!

您购买后可以获得什么

获得完整提示词模板
- 共 250 tokens
- 2 个可调节参数
{ 角色或情境 } { 输出语言 }
自动加入"我的提示词库"
- 获得提示词优化器支持
- 版本化管理支持
获得社区共享的应用案例
限时免费

不要错过!

免费获取高级提示词-优惠即将到期

17
:
23
小时
:
59
分钟
:
59