character-designer

Character Designer — Character Designer

Design a complete character image for a video project: from the character's background, personality, and appearance to three-view (front/side/back) prompt generation, ensuring the character maintains visual consistency across all subsequent still images.

When to Use This Skill

A video project needs one or more fixed character images
Create a brand ambassador / digital human IP
Animated short / Story Video requires multi-character design
Ensure the same character is visually consistent across different scenes
Design a character from scratch or customize a character based on brand requirements

What This Skill Does

Character Profile: Create the character's complete background setting
Appearance Description: Define visual features in detail (face, body type, clothing, accessories)
Three-view Prompts: Generate AI image prompts for front/side/back
Consistency Token: Export reusable "character identity keywords"
Multiple Outfits: Support different scene outfit options for the same character
Expression Set: Define the character's key expression changes

How to Use

Start from a description

设计一个角色：30 岁左右的女性科技创业者，干练自信

From brand context

根据项目上下文，设计一个匹配品牌调性的代言人形象

Multi-character design

为这个视频设计 3 个角色：主持人、用户、技术专家

Instructions

When user requests character design:

Understand Requirements

我来帮你设计角色。需要了解：

1. **角色定位**: 这个角色在视频中的角色是什么？
   [ ] 主持人 / 讲解员
   [ ] 品牌代言人
   [ ] 故事主角
   [ ] 配角 / 对话方
   [ ] 旁白角色 (只有声音)
   [ ] 其他: ___________

2. **性别与年龄**: 
   - 性别: [男 / 女 / 中性]
   - 年龄段: [20s / 30s / 40s / 50+]

3. **视觉风格**: 
   [ ] 真实感 (Photorealistic)
   [ ] 3D 卡通 (Pixar-style)
   [ ] 2D 插画 (Illustration)
   [ ] 动漫风 (Anime)
   [ ] 数字人 (Digital Human)
   [ ] 其他: ___________

4. **品牌调性** (参考 project-context.md): [专业/友好/活力/权威/...]

5. **参考图**: 有没有已有的角色形象或风格参考？

Create Character Profile

# 角色档案: [角色名]

## 基本信息

| 属性 | 描述 |
|------|------|
| **角色名** | [Name / 代号] |
| **角色定位** | [主持人 / 主角 / 配角] |
| **性别** | [男 / 女 / 中性] |
| **年龄** | [具体年龄或范围] |
| **身高** | [cm] |
| **体型** | [纤细 / 匀称 / 健壮 / 微胖] |

## 性格特征

- **核心性格**: [3-5 个关键词]
- **说话风格**: [温和 / 干练 / 幽默 / 专业 / 热情]
- **标志性动作/习惯**: [举例]
- **情绪基调**: [乐观 / 沉稳 / 严肃 / 活跃]

## 外貌特征 (Visual Identity)

### 面部
- **脸型**: [椭圆 / 方形 / 圆形 / 心形 / 长形]
- **眼睛**: [颜色、大小、形状]
- **眉形**: [浓眉 / 柳叶眉 / 自然]
- **鼻子**: [挺拔 / 小巧 / 自然]
- **嘴唇**: [薄 / 中厚 / 厚]
- **肤色**: [白皙 / 小麦色 / 深色 / 棕色]
- **面部特征**: [酒窝 / 痣 / 雀斑 / 无]

### 发型
- **长度**: [短发 / 中长 / 长发]
- **颜色**: [黑 / 棕 / 金 / 其他]
- **造型**: [直发 / 卷发 / 扎起 / 披散]
- **刘海**: [有 / 无 / 偏分]

### 服装 (默认造型)
- **风格**: [商务 / 休闲 / 科技 / 时尚 / 制服]
- **上装**: [详细描述]
- **下装**: [详细描述]
- **鞋子**: [类型]
- **配饰**: [眼镜 / 耳环 / 手表 / 领带 / 胸针 / 无]

### 配色方案
- **主色**: [服装/外貌主色] — hex: [#code]
- **辅色**: [辅助色彩] — hex: [#code]
- **点缀色**: [配饰颜色] — hex: [#code]

## 背景故事 (可选)

[2-3 句话描述角色背景，用于理解角色动机和行为]

Generate Three-view Prompts

Generate one set of prompts each for front, side, and back:
```
## 三视图 AI 生图提示词

### 通用风格标签 (所有视图共享)
```
[art style], [quality keywords], [lighting], character reference sheet, white background, clean studio lighting, full body shot, [specific style tokens]
```
### 正面图 (Front View)

**Prompt**:
```
[character description], front view, facing camera, [detailed facial features], [hair description], [clothing top], [clothing bottom], [shoes], [accessories], [pose: standing naturally / confident pose], [art style], character design, reference sheet, neutral white background, studio lighting, [quality markers: 4K, sharp focus, detailed, etc.]
```
**Negative Prompt**:
```
blurry, low quality, distorted face, extra limbs, bad anatomy, deformed, watermark, text, multiple characters, cropped, partial body
```
---

### 侧面图 (Side View / Profile)

**Prompt**:
```
[character description], side view, profile view, facing right, [detailed facial profile], [hair from side], [clothing visible from side], [body proportions], [art style], character design, reference sheet, neutral white background, studio lighting, [quality markers]
```
**Negative Prompt**:
```
[same as front + ] facing camera, front facing, twisted pose, unnatural angle
```
---

### 背面图 (Back View)

**Prompt**:
```
[character description], back view, facing away from camera, [hair from back], [clothing back details], [any back accessories or details], [art style], character design, reference sheet, neutral white background, studio lighting, [quality markers]
```
**Negative Prompt**:
```
[same as front + ] face visible, front facing, side view, looking at camera
Export Consistency Token

This is the most critical step — extract the character identity keywords that can be reused in all subsequent still image prompts:
```
## 角色一致性 Token (Consistency Reference)

### 简短版 (嵌入其他提示词)
```
[age] [gender] [ethnicity] with [hair description], [key facial feature], wearing [core outfit description], [body type], [defining accessory]
```
### 完整版 (需要高一致性时使用)
```
[Full character prompt with ALL visual details, excluding pose/angle/emotion — those are determined by the scene]
```
### 使用示例

**在剧照提示词中引用角色**:
```
[SCENE DESCRIPTION], [CHARACTER TOKEN: 30-year-old asian woman with shoulder-length black hair and side bangs, wearing navy blazer over white blouse, wire-rimmed glasses, slim build], [POSE/ACTION for this shot], [SCENE LIGHTING and ENVIRONMENT]
```
> **规则**: 
> - 每次生成包含此角色的剧照时，必须注入此 Token
> - 不要修改 Token 中的外貌描述 (除非剧情需要换装)
> - 换装时只替换服装部分，保持面部和体型描述不变
```

Multiple Outfit Options (if needed)

## 服装方案

### Look 1: 商务正装 (默认)
- 上装: [描述]
- 下装: [描述]
- 适用场景: [办公室、会议、专业演讲]
- Token 替换: `wearing [outfit 1 description]`

### Look 2: 休闲装
- 上装: [描述]
- 下装: [描述]  
- 适用场景: [户外、生活化场景]
- Token 替换: `wearing [outfit 2 description]`

### Look 3: [场景特定]
- [描述]
- 适用场景: [特定场景]
- Token 替换: `wearing [outfit 3 description]`

Expression Set Definition

## 表情参考

| 表情 | 提示词追加 | 适用情境 |
|------|-----------|---------|
| 自信微笑 | `confident smile, warm expression` | 开场、介绍 |
| 认真思考 | `thoughtful expression, slight frown` | 分析、讲解 |
| 兴奋 | `excited expression, bright eyes, wide smile` | 亮点展示 |
| 关切 | `concerned expression, empathetic look` | 讲述痛点 |
| 专注 | `focused expression, serious look` | 演示操作 |
| 友好招手 | `waving hand, friendly smile` | 打招呼、结尾 |

Save Character Profile

Save to ./output/characters/[character-name]/ directory:
- profile.md — Complete character profile
- three-view-prompts.md — Three-view prompts
- consistency-tokens.md — Consistency Token
- expressions.md — Expression set (if defined)
Update ./output/characters/character-index.md:
```
# Character Index

| 角色名 | 定位 | 风格 | 目录 |
|--------|------|------|------|
| [Name 1] | 主持人 | 真实感 | `./[name-1]/` |
| [Name 2] | 配角 | 3D 卡通 | `./[name-2]/` |
```

Output Summary

✅ 角色设计完成！

**角色**: [Name]
**风格**: [视觉风格]
**档案**: `./output/characters/[name]/profile.md`
**三视图提示词**: `./output/characters/[name]/three-view-prompts.md`
**一致性 Token**: `./output/characters/[name]/consistency-tokens.md`

**Quick Reference (一致性 Token)**:

[Short version Token — can be copied directly into other prompts]


**Next Steps**:
1. 生成三视图: `用三视图提示词生成角色图片`
2. 设计更多角色: `设计另一个角色: [描述]`
3. 编写场景剧本: `为 [角色名] 编写场景剧本`
4. 换装: `为 [角色名] 设计休闲装方案`

Tips

Define the style before designing the character: Ensure style consistency (e.g., all Pixar style or all photorealistic)
The more precise the Token, the better the consistency: Use specific descriptions instead of vague words
Three-view white background: This makes later cutout and compositing easier
Name clearly: Character names will be referenced frequently in scene scripts
Save the Token: Every subsequent still image needs this Token injected

Related Skills

prompt-generator — Combine character Tokens to generate scene still image prompts
batch-image-generator — Batch generate character three-views and still images
scene-scriptwriter — Reference characters in scene scripts
digital-human-integrator — Apply character settings to digital humans