infinimo-ai-design-virtual-model

该 Skill 通过脚本调用 Infinimo AI Design 的虚拟模特工具：先把真人或模特底图及可选背景图上传取得地址，再提交生成任务（可设提示词、面部提示词、数量、宽高比），随后轮询日志直到返回输出图片链接，主要用于时装画册与电商模特换装展示；需配置访问令牌。

anyunzhong

热度 9.67

最近更新 2026-07-06 10:19:56

分类：创作依赖：代码执行 / 读写文件 / 第三方 API应用场景：图像生成 / 品牌素材

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add infinimo-ai-design-virtual-model

Metadata

name: infinimo-ai-design-virtual-model
description: Generate virtual model showcase images via Infinimo AI Design from a source photo, optional background, and prompts. Use for fashion lookbooks, model swaps, and on-model product presentation without a live shoot.

Virtual Model

About Infinimo AI Design

Infinimo AI Design is an AI design platform for e-commerce visuals. This skill calls the Virtual model tool (same as /virtual-model in the design studio).

Web page: https://design.infinimo.ai/?source=q-i-d-clawhub

Authentication & base URL

Base URL: https://www.clawec.com/api
Token: https://design.infinimo.ai/?source=q-i-d-clawhub · https://design.infinimo.ai/api-key?source=q-i-d-clawhub
Headers: Token: <TOKEN>

Use INFINIMO_TOKEN or INFINIMO_API_KEY. Common params: platform=1, terminal=4, language=en.

End-to-end flow

1. POST /upload/image                          → upload source (and optional background)
2. GET  /aigc/image/virtual_model_create       → submit generation job
3. GET  /aigc/log/list?type=101                → poll results (log type 101)

Generation is async; poll logs until output images appear in the newest record.

1. Upload images

POST /upload/image

Image	Required	Description
Source / base	yes	Real person or mannequin photo
Background	no	Optional scene reference

bash scripts/upload.sh /path/to/source.jpg
bash scripts/upload.sh /path/to/background.jpg

2. Submit virtual model generation

GET /aigc/image/virtual_model_create

Parameter	Required	Description
base_image_url	yes	Source image URL
base_image_type	yes	`1` = real person, `2` = mannequin
bg_image_url	no	Background reference URL
prompt	no	Model look / styling prompt
face_prompt	no	Face-specific prompt
num	no	Output count, e.g. `1`–`4`
aspect_radio	no	Aspect ratio (H:W), e.g. `1:1`, `9:16`, `比例不变` (keep original)

Allowed aspect values include: 比例不变, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16, 1:2.

bash scripts/create.sh \
  --base-url "https://cdn.../source.jpg" \
  --base-type 1 \
  --bg-url "https://cdn.../bg.jpg" \
  --prompt "Professional fashion model, studio lighting" \
  --face-prompt "Natural smile, soft makeup" \
  --num 1 \
  --aspect "9:16"

3. Poll results

GET /aigc/log/list

Parameter	Value	Description
type	`101`	Virtual model log type
agent_id	`0`	Default
group_id	`0`	Default

bash scripts/logs.sh

Response `data.logs[]`

Field	Description
logId	Record id
createTime	Timestamp
param	Submit parameters
data.type	`image` when complete
data.data	Output image URL array

Poll every 3–5s until the latest log has data.type === "image" and non-empty data.data.

Delete: GET /aigc/log/delete?log_id=<logId>

See references/response-schema.md.

Workflow

Confirm source type (real person vs mannequin) and prompts
upload source and optional background
create with URLs and options
Poll logs (type 101) until images return
Return output URLs and parameters

Example

Input: Generate 9:16 showcase from mannequin photo with luxury boutique background

upload.sh mannequin.jpg → base URL
upload.sh boutique.jpg → bg URL
create.sh --base-url ... --base-type 2 --bg-url ... --aspect 9:16 --prompt "Luxury boutique fashion shoot"
logs.sh → image URLs