| name | zhuoran-selfie |
| description | Zhuoran AI selfie photo generation skill using WaveSpeed AI with reference image.
Creates contextual self-portrait photos for office, cafe, gym, airport, and scenic locations.
Supports one-step, two-step, and smart generation modes for optimal quality.
Use when: 卓然自拍, AI照片生成, 角色照片, 场景自拍, 垫图生成,
zhuoran selfie, AI photo generation, character photo, scene selfie,
reference image, professional portrait
Related: zhuoran-video-selfie, qizhuo-selfie, clawra-selfie
Part of UniqueClub toolkit. Learn more: https://uniqueclub.ai
|
卓然自拍技能 (Zhuoran Selfie)
基于垫图使用 WaveSpeed AI 生成场景化自拍照片,支持一步直达、两步法和智能判断三种生成模式。
When to Use
适用于以下场景:
- 需要生成卓然角色的自拍照片
- 专业、干练风格的场景展示
- 办公室、咖啡厅、机场等商务场景
- 穿搭展示和日常分享
- 深夜互动或工作场景
Do NOT use this skill if:
- 需要生成守护型风格的照片(使用 qizhuo-selfie)
- 需要动态视频(使用 zhuoran-video-selfie)
- 场景涉及镜子自拍或反射场景
- 需要全身镜自拍照
触发关键词 / Trigger Phrases:
- 卓然自拍 / zhuoran selfie
- AI照片生成 / AI photo generation
- 角色照片 / character photo
- 场景自拍 / scene selfie
- 垫图生成 / reference image
- 专业肖像 / professional portrait
- 办公室照片 / office photo
- 穿搭展示 / outfit showcase
Workflow
快速开始
openclaw skill run zhuoran-selfie office
openclaw skill run zhuoran-selfie cafe --mode selfie
openclaw skill run zhuoran-selfie beach --method two_step
./skills/zhuoran-selfie/scripts/zhuoran-selfie.sh office --mode direct
命令行参数
./skills/zhuoran-selfie/scripts/zhuoran-selfie.sh <scene> \
--mode <direct|selfie|portrait> \
--method <one_step|two_step|smart> \
--target <@user|#channel> \
--caption "配文" \
--output /path/to/output.png
生成模式详解
一步直达 (one_step)
流程: 垫图 → 场景
适用: 变化小的场景(gym, cafe, westlake, bookstore)
两步法 (two_step)
流程: 垫图 → 中性背景 → 场景
适用: 变化大的场景(office, beach)
智能判断 (smart)
流程: 自动根据 change_level 选择一步或两步
逻辑:
minimal / low → 一步直达
medium / high → 两步法
API 调用流程
- 上传参考图 → WaveSpeed AI 媒体上传
- 提交编辑任务 →
POST /api/v3/x-ai/grok-imagine-image/edit
- 轮询结果 →
GET /api/v3/predictions/{task_id}/result
- 下载图片
Python API
from skills.zhuoran_selfie.zhuoran_selfie import generate_smart, generate_one_step, generate_two_step
image_url = generate_smart("office", mode="direct")
image_url = generate_one_step("cafe", mode="selfie")
image_url = generate_two_step("beach", mode="portrait")
Available Scenes
| 场景 | 代码 | 变化幅度 | 推荐模式 |
|---|
| 办公室 | office | high | 两步法 |
| 咖啡厅 | cafe | low | 一步直达 |
| 机场 | airport | medium | 智能判断 |
| 西湖 | westlake | low | 一步直达 |
| 书店 | bookstore | low | 一步直达 |
| 健身房 | gym | minimal | 一步直达 |
| 海滩 | beach | high | 两步法 |
| 深夜加班 | selfie_late_night | medium | 智能判断 |
Guardrails
禁用场景
以下场景已被禁用(高风险):
mirror_selfie - 全身镜自拍,存在逻辑破绽
mirror_reflection_selfie - 镜子反射自拍,破绽明显
beach_selfie - 涉及比基尼,可能不合适
Prompt 模板
direct (普通肖像):
{scene_description}, direct eye contact with camera,
looking straight into lens, natural lighting,
photorealistic, 8k quality
selfie (自拍模式):
a close-up selfie taken by herself at {scene_description},
direct eye contact with camera,
looking straight into the lens, eyes centered and clearly visible,
not a mirror selfie,
phone held at arm's length but phone not visible in frame,
face fully visible, natural lighting
portrait (专业肖像):
professional portrait of {scene_description},
soft natural lighting, half-body shot,
clean background, photorealistic
与 v1.0 (clawra) 的区别
| 特性 | v1.0 (clawra) | v2.0 (zhuoran) |
|---|
| 参考图 | clawra.png (卡通) | zhuoran_portrait_base.png (真人) |
| API 平台 | fal.ai | WaveSpeed AI |
| 生成模式 | mirror / direct | 一步 / 两步 / 智能 |
| 场景管理 | 无 | 场景模板库 |
| 安全控制 | 无 | 禁用高风险场景 |
| 触发方式 | Bash/TS | Python CLI + Bash 包装 |
Environment Requirements
WAVESPEED_KEY: WaveSpeed API 密钥
- Python 3.8+
- 依赖: requests
File Structure
skills/zhuoran-selfie/
├── SKILL.md # 本文档
├── zhuoran_selfie.py # 核心逻辑
└── scripts/
├── zhuoran-selfie.py # Python CLI 入口
└── zhuoran-selfie.sh # Bash 包装脚本(OpenClaw 入口)
Related Skills
About UniqueClub
Part of UniqueClub toolkit - AI-powered creative tools for character photo generation.
Learn more: https://uniqueclub.ai
注意:本技能仅供非凡产研内部使用