Image Subject Background Replacement

The problem

Replacing backgrounds for products, people, or pets often becomes a manual bottleneck in e-commerce images, posters, and social content: masking takes time, lighting can look unnatural, and outputs still need checks for watermarks and resolution. This skill turns that workflow into one asynchronous image task: provide subject_image and bg_img, then receive a watermark-free image, a quality report, and a presentable page. It is useful when generating several variants for manual review.

How it works

The core flow submits a task and polls for results:

• Input validation: accepts a public image URL or standard Base64, then runs basic compliance checks.
• Async submission: calls execute_task.py to start the job and returns only task_id without blocking.
• Polling: calls query_task.py to check status and immediately surfaces key fields on each round.
• Quality assessment: reports score and boolean checks for subject retention, watermark-free output, high resolution, and lighting consistency.

Boundaries

It fits clear, well-edged subjects and backgrounds with compatible lighting. If the score is low, use a sharper subject image or a more compatible background. If the subject is incomplete, heavily occluded, or the background lighting is clearly mismatched, results may be unstable. AppKey-less simulation is a preview only and does not represent real generation.