ID Photo Generator

Problem

In ID photo workflows, the input is usually one real portrait, while the output must satisfy a clean background, formal clothing, half-body 3:4 framing, and a centered head. Hand-written prompts can over-embellish the subject or leak server-side style recipes. ID Photo Generator requires at least one reference image and keeps style, framing, and makeup constraints inside the server workflow.

How it works

• Analyze the reference image and fill short fields such as options.gender, options.age, and options.clothing; when text and image conflict, the image wins by default.
• Call the fixed model and capability with image_edit; the full prompt can be omitted because the server assembles the closed-source instructions.
• Poll the async task until status becomes succeeded, then return the image from data.result.data[].url.
• When payment is required, the first call returns 402; retry with the same JSON body and complete payment via the weixinpay extension.

Boundaries

Do not generate without a reference portrait. Do not compose long style prompts on the client side. Keep options to short overrides such as background, clothing, and size_hint. On failure, inspect data.error_message instead of inventing third-party endpoints.