SuperAPI AI Directory

designer-toy

Design & creation

View repository

Updated 10-16 09:56

About this skill

Designer Toy Blind Box Doll Series Design Skill

Overview

Generate designer toy/blind box series.Core Capabilities:

  • Character Consistency Assurance: Ensure character features are completely consistent in group and solo images (hair color, eyes, proportions)
  • Pose Diversity: Avoid all standing poses; provide rich poses such as sitting/crouching/lying/floating
  • Scene Richness: Complete scene design including lighting, background, and atmosphere

Workflow

flowchart TD
    A["1. 定义角色锚点"] --> B["2. 设计系列款式"]
    B --> C["3. 生成群像"]
    C --> D["4. 创建位置映射表"]
    D --> E["5. 逐个生成单人图"]
    E --> F{"6. 一致性检查"}
    F -->|通过| G["完成"]
    F -->|未通过| H["调整Prompt重新生成"]
    H --> E

Stage 1: Define Character Anchors (CRITICAL)

[!IMPORTANT] Mustuse in all image generationscompletely identicalcharacter anchor descriptions, otherwise it will cause inconsistency between group and solo images!

Character Anchor Template

Create for each character character-anchor.md file:

# Character Anchor - [角色名称]

## LOCKED Identity (所有图片必须保持一致)

### Base Foundation
Designer toy, vinyl figure collectible,
professional product photography, studio lighting,
chibi proportions, highly detailed collectible figure

### Proportions (LOCKED)
- Head-to-body ratio: [如 2.3:1]
- Head shape: [如 Round but not perfectly round]
- Body: [如 Slightly chubbier, approachable]

### Face (LOCKED)
- Shape: [具体描述]
- Expression base: [基础表情风格]

### Eyes (LOCKED)
- Size: [如 Large, 42% of face width]
- Shape: [如 Round with subtle upturn at corners]
- Pupils: [如 Large black with big highlights]
- Expression: [如 Curious but confident, dreamy cool]

### Skin (LOCKED)
- Tone: [颜色名 + Hex色号,如 Creamy fair #FFF5EE]
- Texture: [如 Smooth matte vinyl]

### Hair Color (可变但需明确声明)
- **发色变体表**:
  - 基础款: [颜色 + Hex,如 Warm brown #8B4513]
  - 精灵款: [颜色 + Hex,如 Golden blonde #F0C05A]
  - 隐藏款: [颜色 + Hex,如 Silver white #C0C0C0]

### Blush (LOCKED)
- Shape: [如 Small circular]
- Color: [颜色 + Hex,如 Soft pink #FFB0B0]
- Style: [如 Soft gradient edges]

Stage 2: Design Series Styles

Each style needs to define:

  • Clothing description (use clothing anchors!)
  • Expression changes
  • Pose (must be diverse!)
  • Accessories
  • Scene/Background
  • Lighting style

Clothing Anchor System (CRITICAL)

[!IMPORTANT] Mustcreate an anchor description for each outfit, and both group and solo images reference the same anchor to ensure the clothing is completely consistent!

# Outfit Anchor - [款式名称]

## LOCKED Outfit (群像和单人图必须完全一致)

### Main Garment
- Type: [服装类型,如 sweater, dress, costume]
- Color: [主色 + Hex,如 Red #C41E3A]
- Secondary color: [次色 + Hex,如 White #FFFFFF]
- Pattern: [图案描述,如 White snowflake pattern on chest]
- Material texture: [材质,如 Knitted texture, chunky yarn]

### Details (LOCKED)
- Collar/Neckline: [领口,如 Ribbed round collar]
- Sleeves: [袖子,如 Long sleeves with ribbed cuffs]
- Fit: [版型,如 Relaxed oversized fit]
- Special elements: [特殊元素,如 Small pompom at hem]

### Accessories (LOCKED)
- [配件1,如 Red santa hat with white fur trim]
- [配件2,如 Green striped scarf]

Pose Diversity Library (avoid all standing poses!)

Category Pose Type Prompt Keywords Applicable Scenes
Sitting Pose Cross-legged sitting sitting cross-legged on floor Cozy, casual
Sitting on windowsill sitting on windowsill, legs dangling Dreamy, looking up
Leaning back sitting leaning against, relaxed sitting Lazy, comfortable
Crouching Pose Crouching with chin on hand lying on stomach, chin in hands, feet up Playful, cute
Crouching on an object lying over [object], arms hanging Lazy, relaxed
Lying Pose Lying on back lying on back, arms spread, snow angel pose Joyful, snowy ground
Lying on side lying on side, curled up Sleeping, cute
Floating Floating floating in air, dreamy levitation Magic, dream
Flying flying pose with cape flowing Hero, fantasy
Dynamic Standing Pose Standing on one foot standing on one leg, playful balance Lively, mischievous
Jumping jumping with joy, feet off ground Joyful
Interactive Pose Holding an object holding [item], looking at it lovingly Storytelling feel
Hugging an accessory hugging [plush/item] Cozy

Product Form Expansion

Form Prompt Keywords Features
Blind box doll vinyl figure, collectible toy, blind box figure Standard three-dimensional figure
Plush charm plush keychain, hanging charm, soft plush toy Cute and soft
Sitting plush chubby sitting bean bag pose, round bottom Chubby Sitting Pose
Acrylic Standee acrylic standee, flat illustration style, 2D art Flat Standee

Stage 3: Generate Group Portrait

Group Portrait Scene Design Principles (CRITICAL)

[!IMPORTANT] The group portrait needs a setting, butmust not steal the spotlight! The setting serves the theme, and the characters are always the visual focus.

Composition Priority (front to back)

  1. Foreground Characters - Absolute main subject, highest clarity and saturation
  2. Ground/Platform - Supports the characters, provides a sense of space
  3. Midground Props - 2-3 small props echoing the theme, restrained and not cluttered
  4. Background Atmosphere - Blurred treatment, enhances the atmosphere

Background Handling Techniques

Technique Prompt Keywords Effect
Depth of field blur soft bokeh background, blurred background elements Highlight foreground characters
Color downgrade desaturated background, muted background tones Avoid the background stealing attention
Simplify elements minimal background props, subtle scene elements Avoid clutter
Atmospheric glow soft ambient glow, dreamy atmosphere Add a sense of layering

Group Portrait Prompt Structure

## Collection Group Shot Prompt

### 1. Scene Setting (布景但不喧宾夺主)
Blind box collection display, professional product photography,
[主题场景描述,如: cozy Halloween pumpkin patch setting],
soft gradient background with subtle [主题元素] silhouettes in bokeh,
foreground figures in sharp focus, background softly blurred,
desaturated background tones to emphasize characters

### 2. Platform/Ground (承载角色的平台)
[平台描述,如: wooden crate platform with scattered autumn leaves],
platform color complementing but not competing with figures

### 3. Arrangement
[数量] designer toys arranged in a group, varied heights and positions,
some standing, some sitting, natural grouping with breathing room

### 4. Characters (从左到右,每个必须完整描述,引用服装锚点)
- Left 1: [完整角色描述 + 发色 + 服装锚点内容]
- Left 2: [完整角色描述 + 发色 + 服装锚点内容]
- Center: [完整角色描述 + 发色 + 服装锚点内容]
- Right 2: [完整角色描述 + 发色 + 服装锚点内容]
- Right 1: [完整角色描述 + 发色 + 服装锚点内容]

### 5. Background Props (克制!2-3个即可)
- [道具1,如: carved pumpkin with soft glow, slightly out of focus]
- [道具2,如: small pile of candy corn, blurred in background]

### 6. Lighting (突出角色)
Product photography lighting, soft box lighting,
rim light for character separation from background,
[氛围打光,如: warm orange ambient glow from below]

### 7. Technical
High quality render, 4K, detailed [材质] texture,
depth of field with f/2.8 equivalent bokeh

Stage 4: Create Position Mapping Table (CRITICAL)

After generating the group portrait,immediatelycreate a position mapping table,must reference clothing anchors:

Group Portrait Position Style Name Fur Color/Hair Color (Exact) Clothing Anchor Reference Pose Corresponding Single Character Image
Left 1 Pumpkin Cloak Style Cream Brown #D2B48C → Outfit-01 Standing pose holding a pumpkin lantern 01-pumpkin-cape.png
Left 2 Little Devil Style Caramel Brown #8B4513 → Outfit-02 Playful one-foot standing 02-little-devil.png
Center Ghost Cloak Style Cream Brown #D2B48C → Outfit-03 Floating and drifting 03-ghost-cloak.png
Right 2 Wizard Style Dark Chocolate #3D2314 → Outfit-04 Sitting pose flipping through a magic book 04-wizard.png
Right 1 Hidden Style - Skeleton Moonlight White #F5F5F5 → Outfit-05 Lying pose playing dead 05-hidden-skeleton.png

Stage 5: Generate Single Character Images One by One

[!IMPORTANT] The group portrait must be used as a visual reference! Text descriptions alone cannot guarantee consistency; Gemini needs to directly see the character images in the group portrait.

Single Character Image Generation Method (CRITICAL)

When generating each single character image,must simultaneously provide:

  1. the group portrait image - as a visual reference
  2. Position indication - clearly specify which character in the group portrait it is
  3. Text Prompt - supplement scene and technical details

Single Character Image Prompt Structure

## Individual Figure Prompt

### 0. Visual Reference (CRITICAL - 必须包含!)
[附上群像图片 00-collection-group.png]

请参考上面的群像图片,为我生成其中 **[位置描述]** 的角色的单人特写图。

位置描述示例:
- "从左数第1个角色"
- "最右边的角色"
- "正中间的角色"
- "从右数第2个角色(穿紫色巫师袍的那个)"

这个角色在群像中的特征:
- 服装:[简要描述,如:紫色巫师袍,尖顶帽]
- 姿势:[群像中的姿势,如:盘腿坐着]
- 位置:[精确位置,如:右边第二个]

请保持这个角色的所有视觉特征(毛色、服装颜色、配件、比例)与群像完全一致!

### 1. Foundation
Designer toy, [材质类型] figure collectible,
professional product photography, chibi proportions

### 2. Character Anchor (从角色锚点完整复制!)
[完整的LOCKED特征描述]

### 3. Fur/Hair Color (必须显式声明!从映射表复制!)
Fur Color: [颜色名] [Hex色号]
例如: Fur Color: Cream brown #D2B48C

**MUST match the character in the reference group image!**

### 4. Outfit (从服装锚点完整复制!)
[完整的服装锚点内容,不要简化!]

**MUST match exactly as shown in the reference group image!**

### 5. Expression & Pose
Expression: [表情描述]
Pose: [姿势描述,使用姿势库关键词]

### 6. Scene & Background (不喧宾夺主)
Background: [背景描述,使用虚化和低饱和]
Scene elements: [2-3个场景元素,soft bokeh]
Atmosphere: [氛围关键词]
Platform: [承载平台描述]

### 7. Lighting
[打光描述,见打光库]
rim light for figure separation

### 8. Technical
3/4 front view, high quality render, 4K,
detailed [材质] texture, soft shadows,
depth of field with background blur

Operation Steps When Calling Gemini

  1. Upload the group portrait image - use 00-collection-group.png as the input image
  2. Clearly indicate the position at the beginning of the Prompt - "Please refer to this group portrait and generate for methe 2nd character from the right's single character image"
  3. Emphasize consistency requirements - "Must maintain an appearance fully consistent with that character in the group portrait"
  4. Provide a complete scene description - The background, lighting, etc. can differ, but the character itself must be consistent

Scene and Lighting Library

Atmosphere Scene Elements Background Description Lighting
Cozy Warmth Fireplace, wooden floor, Christmas stockings, blanket warm living room warm orange glow, fireplace lighting
Dreamy Starry Sky Stars, moon, falling stardust, window night sky through window soft blue-purple gradient, moonlight
Sweet Candy Gingerbread house, candy decorations, lollipops candy land background pink warm lighting, sweet ambiance
Festive Joy Christmas tree, gift boxes, string lights, gold decorations festive holiday scene colorful bokeh lights, warm festive
Fresh Snow Scene Snowflakes falling, ice crystals, snowy ground, white environment snowy winter scene cool blue-white lighting, soft diffused
Halloween Mystery Jack-o'-lantern, spider webs, dry leaves, bat silhouettes spooky halloween night warm orange underglow, purple ambient
Cute Halloween Candy basket, cartoon ghosts, colorful pumpkins cute halloween party soft orange-purple gradient, playful
Autumn Forest Fallen leaves, mushrooms, tree stump, pinecones autumn forest floor golden hour warm light, dappled shadows

Stage 6: Consistency Check

Required Check Items

Check Item How to Check Common Issues
Hair Color Consistency Compare the hair color of the same character in the group portrait and the single-character image Group portrait blonde hair → single image brown hair
Eye Consistency Check eye shape, size, and highlight position Eyes become smaller / highlights disappear
Proportion Consistency Check the head-to-body ratio In the single image the head becomes smaller / the body becomes longer
Clothing Match Compare clothing color and pattern Pattern disappears / color shifts
Expression Match Check whether the expression matches the design Expression does not match the setting

When inconsistency is found

  1. Identify the specific difference: hair color / eyes / proportions / other
  2. Check the single-image Prompt: Was the hair color declaration omitted?
  3. Modify the Prompt:
    • Place the hair color description earlier
    • Use a more emphatic tone:Hair Color: MUST BE Golden blonde #F0C05A
    • Add negative prompts:NOT brown hair, NOT dark hair
  4. Regenerate

Common Errors and Solutions

Problem Cause Solution
Inconsistent hair color between group portrait and single image Hair color was omitted from the Prompt Each imagemustexplicitly declare hair color + Hex color code
All poses are standing poses The pose library was not used Refer to the pose library and choose a different pose for each version
The background is too simple Only "white background" was written Add scene elements + atmosphere + lighting description
Character proportions are inconsistent The head-to-body ratio was not locked Fix the proportions in the character anchor and repeat them every time
Eye style changes Eye features were not described in detail Lock eye size, shape, and highlight description

Output file structure

[系列名称]/
├── character-anchor.md           # 角色锚点定义
├── series-design.md              # 系列款式设计
├── position-mapping.md           # 群像位置映射表
├── prompts/
│   ├── 00-collection-group.md    # 群像Prompt
│   ├── 01-xxx.md                 # 各款式Prompt
│   └── ...
├── 00-collection-group.png       # 群像
├── 01-xxx.png                    # 单人图
├── ...
└── REPORT.md                     # 设计报告

Start execution

  1. Ask the user for the series theme they want to design
  2. Create character anchor file
  3. Design series styles (note pose diversity!)
  4. Generate group portrait and create position mapping table
  5. Generate individual portraits one by one (each must declare hair color!)
  6. Perform consistency check
  7. If inconsistent, adjust Prompt and regenerate
  8. Output the complete series and report