Skip to main content
ClaudeWave
Skill482 repo starsupdated 3d ago

director.art-design

服化道——设计「人物设定图」与「场景环境图」的生图提示词,含顶部「风格前缀块」(摄影机/胶片/调色/画幅等烧进画面的统一风格)+ 生图 avoid(多指/文字水印/穿帮等)+ identity DNA(角色跨图一致的关键特征锁定)。Nomi 为角色/场景等视觉 anchor(`carrier: visual`)生成参考图、写它的生图 prompt 时参考。

Install in Claude Code
Copy
git clone --depth 1 https://github.com/aqm857886159/Nomi /tmp/director.art-design && cp -r /tmp/director.art-design/skills/director-art-design ~/.claude/skills/director.art-design
Then start a new Claude Code session; the skill loads automatically.

SKILL.md

# 服化道(视觉 anchor 生图提示词)

把「跨镜头要一致的角色 / 场景 / 道具」画成一张**参考图**——它长什么样、穿什么、在什么空间里。这张图是后续所有视频镜头的**视觉地基**:视频用图生视频(i2v),画面风格与长相都从参考图继承。

**在 Nomi 里这对应画布上的「视觉 anchor 节点」**(`carrier: visual` 的 `character` / `scene` / `prop`):系统会用你写的这段 prompt,经**应用内文生图模型**生成一张参考图,锁住这个元素的样子;之后引用它的镜头连一条参考边,长相就不漂移。

> **本技能只管「生图阶段」。** 视频 shot 的 prompt **不写**风格前缀、**不写** avoid——那些从参考图继承。风格与画面基调只在这里(生图)一次性烧进画面。

## 铁律:每张参考图顶部都带「风格前缀块」

这是本技能最核心的产出。每张人物 / 场景生图 prompt 的**最顶部**,先放一个英文「风格前缀块」,再接中文叙事正文。风格前缀块 = 两部分(可选第三部分):

**① 风格前缀(英文)**——把「用什么摄影机 / 什么胶片 / 什么调色 / 什么画幅」压成一串词,烧进画面。人物版与场景版分开写。选定一套后整片统一,别一张一个样。常用基线(真人写实,按项目替换机型/镜头/调色/色温):

```
# 人物版
photorealistic cinematic portrait, shot on ARRI ALEXA, prime lens, natural motivated lighting,
true-to-life skin texture with subsurface detail, film grain, shallow depth of field, color-graded cinematic look.

# 场景版
photorealistic cinematic environment, shot on ARRI ALEXA, anamorphic lens, natural light with practical sources,
atmospheric depth, film grain, restrained color grading, true-to-life material textures.
```

**② 生图 avoid(英文)**——图像生成专属负面约束,防多指 / 水印 / 穿帮。基线:

```
avoid text overlays, avoid HUD elements, avoid watermark, avoid extra fingers, avoid deformed hands, avoid distorted face.
```

按元素特性追加(如某题材的视觉禁忌)。

**③ 光影效果层(可选)**——某张图需要一个具体光效(如泳池图要「水下丁达尔光柱」)时,把该光效的英文短语追加在风格前缀之后、中文正文之前,且**只叠在用到它的那张图**上,不全局乱叠。

**组装顺序**:`[风格前缀] + [可选光效短语] + [中文叙事正文]`。示例(人物参考图顶部):

```
photorealistic cinematic portrait, shot on ARRI ALEXA, prime lens, natural motivated lighting,
true-to-life skin texture with subsurface detail, film grain, shallow depth of field, color-graded cinematic look.
avoid text overlays, avoid HUD elements, avoid watermark, avoid extra fingers, avoid deformed hands.

<下面接中文叙事式角色设定:整体定位 / 面部 / 体型 / 服装 / 配饰 / 气质……>
```

## 风格前缀预设库(选一个当基线)

不想从零写风格前缀时,从下面挑一套压缩前缀当项目基线(人物用人物版、场景用场景版)。**预设可跨风格混用**——真人项目拍空镜可叠「3A 游戏 CG」的景观质感,漫剧的环境背景也能借。

| 风格 | 一句话 | 人物版压缩前缀 | 场景版压缩前缀 |
|---|---|---|---|
| **真人写实**(默认) | 真实摄影机实拍质感,电影感、自然光、胶片颗粒 | `photorealistic cinematic portrait, shot on ARRI ALEXA, prime lens, natural motivated lighting, true-to-life skin texture with subsurface detail, film grain, shallow depth of field, color-graded cinematic look.` | `photorealistic cinematic environment, shot on ARRI ALEXA, anamorphic lens, natural light with practical sources, atmospheric depth, film grain, restrained color grading, true-to-life material textures.` |
| **3A 游戏 CG**(UE5 级) | 虚幻 5 级超写实,"比真实更好看" | `3A game CG character rendering, Unreal Engine 5 visual fidelity, subsurface skin scattering, detailed hair strands with translucent highlights, fabric micro-texture visible, cinematic rim lighting, HDR, clean sharp focus.` | `3A game CG rendering, Unreal Engine 5 visual fidelity, cinematic volumetric lighting, HDR imaging, sharp hyper-detailed textures, clean vibrant saturated colors, atmospheric perspective, clear directional light with long shadows.` |
| **皮克斯 3D** | 圆润温暖、表情夸张、一眼想亲近 | `Pixar 3D animation style, soft rounded character design, oversized expressive eyes, warm cinematic lighting, smooth subsurface skin shading, vibrant saturated color palette, shallow depth of field, emotional facial expression.` | `Pixar 3D animation style environment, warm golden hour lighting, rich saturated colors, soft ambient occlusion, detailed miniature-like textures, inviting atmosphere, gentle volumetric haze, cinematic depth of field.` |
| **迪士尼 3D** | 比皮克斯更"魔法感",更强光效、更梦幻 | `Disney 3D animation style, magical cinematic lighting with visible light rays, dramatic color palette with warm-cool contrast, expressive character animation, detailed hair dynamics, enchanted atmosphere with subtle sparkle particles.` | 同左(角色 / 场景共用一套,按需拆) |
| **国漫**(仙侠/玄幻) | 东方水墨意境 + CG 精度并存 | `Chinese fantasy animation (仙侠国漫), East Asian features with intense gaze, flowing long hair, layered Chinese robes (汉服) with jade ornaments, calligraphic chi-energy trails, jade-green / celestial-gold / cinnabar-red palette.` | `Chinese animation style (guoman), ink wash painting influence with modern 3D CG, traditional Chinese architecture (飞檐斗拱), ethereal cloud and mist, jade-toned palette with gold accents, dramatic 天光 rim light.` |
| **新海诚** | 现实场景 + 超饱和光影 + 情绪化天空 | `Makoto Shinkai anime style, stylized character over hyper-detailed background, soft anime facial features, emotional atmospheric lighting.` | `Makoto Shinkai anime style, hyper-detailed realistic backgrounds, ultra-vivid saturated sky with dramatic clouds, golden hour lens flare, rain and light particles, reflective wet surfaces.` |
| **赛璐璞 TV 动画** | 经典日漫平涂色块 + 清晰描边 | `Japanese cel-shaded anime style, clean black outlines, flat color fills with hard-edge shadow cuts, limited palette, bright even lighting, expressive simplified features.` | 同左(限背景细节、平涂色块) |
| **厚涂电影动画**(Arcane 级) | 每帧像油画,笔触可见、色彩厚重 | `Arcane/Fortiche thick-paint 3D CG style, visible brush stroke textures, rich painterly color, dramatic chiaroscuro lighting, realistic proportions, oil-painting-like blending on skin.` | 同左(环境同样厚涂笔触 + 大气分层) |
| **韩漫/Webtoon** | 柔和渐变、精致五官、梦幻光效(竖屏首选) | `Korean webtoon (manhwa) style, soft gradient shading without hard shadow edges, refined beautiful features, luminous skin with soft bloom, pastel palette with vivid accents, romantic atmospheric lighting.` | `Korean webtoon (manhwa) style, clean minimal soft-focus backgrounds, watercolor-wash environments, dreamy diffused light with soft bloom, pastel palette.` |
| **爱死机**(LDR 级 CG) | 同 3A 级精度但更暗更粗粝更成人向 | `Love Death and Robots CG, photorealistic with gritty industrial texture, pore-level skin with sweat and dirt, noir lighting with deep shadows, desaturated palette with one vivid accent, anamorphic lens.` | `Love Death and Robots CG environment, industrial dystopian setting with rust / spalling / peeling paint, volumetric particulate haze, practical light sources only, teal shadows / amber highlights, oil-sheen wet surfaces.` |

**预设没命中**:向用户要「一部代表作 / 导演 / 流派名 + 一张参考图」,按 7 维逐维提炼一套新前缀(① 摄影机/镜头 ② 构图 ③ 光影 ④ 色彩/调色 ⑤ 质感/介质 ⑥ 氛围/母题 ⑦ 运动/节奏),产
root-cause-remediationSkill

Mandatory for every Nomi corrective change: user-reported bugs, regressions, CI-only failures, flaky tests, performance or security defects, review/audit findings, and compatibility failures in any production path. Classify one_off versus recurring before implementation. Recurring and high-risk repairs require a schema-v3 contract, shared enforcement boundary, structural prevention, dependency lifecycle decision, and changed regression evidence.

docxSkill
plansSkill
brand.promoSkill

品牌宣传片 playbook。把产品文案/卖点做成一条「3 秒钩子 → 卖点 → 使用场景 → 行动号召」的短宣传片,分阶段执行(剧本审阅 → 拆镜 → 落画布 → 生成 → 排时间轴),每阶段暂停让用户审阅。

creation-editSkill

编辑创作区文档:读取当前内容,追加或替换文本,维护分镜描述格式。

director.actionSkill

非对抗性动作场景(跑酷/追逐/攀爬/特技/坠落)的动作编排知识——专业动作词汇、环境交互、动作节奏、FPV 追拍、身体力学约束、怎么把动作精确物理化地描述进视频提示词;Nomi 写动作戏 shot 时参考。

director.cinematographySkill

镜头语言与摄影技法方法论——统一景别体系/构图规则/运镜的情绪语言/打光方案/景深控制/色温光源/镜头特性,以及这些怎么翻译成视频提示词该怎么写。Nomi 拆镜头或写视频 shot 的 prompt 时参考。

director.consistencySkill

治 AI 视频多镜头「同角色换脸 / 同道具换形 / 同场景换景」三大顽疾的方法论——五维一致性检查、参考图锚定、场内状态表、段间承接。Nomi 拆镜头与跨镜生成时参考:把每个跨镜元素落成画布 anchor 节点(角色/场景/道具/风格),镜头用 anchorIds 引用、系统连参考边,段间用首尾帧承接。