Qwen Image Edit Inpaint

Qwen Image Edit Inpaint is a Qwen Image masked reconstruction route that combines visual-semantic understanding with image synthesis. It interprets subjects, spatial relationships, typography, and appearance as connected parts of one design, allowing the generated or revised image to follow a detailed creative brief instead of merely matching isolated keywords or request settings.

Use this exact route when the production task calls for masked reconstruction. Describe the subject and intended result first, then specify composition, viewpoint, text content, material, lighting, and visual finish. For edits, identify what must change and what must remain untouched; for adapted or layered work, explain the role of each source and the consistency expected across the final asset.

Highlights

• Masked region reconstruction. Regenerates only the selected area while using the surrounding image to infer plausible content.
• Context-aware boundary blending. Matches perspective, texture, color, and illumination across the edge of the edited region.
• Semantic object replacement. Can remove, add, or transform content according to a natural-language instruction.
• Preservation beyond the mask. Retains unselected regions so localized edits do not become uncontrolled full-image regeneration.

Pricing

When to Use

Prompt Guide

Identify the primary subject and every source or condition, state the intended transformation or action, then describe composition, camera or viewpoint, lighting, materials, pacing, atmosphere, and exact preservation requirements. Refer to multiple inputs in their schema order.

{
  "prompt": "A precisely directed composition with explicit subject, transformation, camera or viewpoint, lighting, material, and preservation requirements",
  "image": "https://example.com/reference.png",
  "mask": "https://example.com/reference.png",
  "seed": 1,
  "strength": 0.93
}

Technical Specs

Related Models

• alibaba/qwen-image
• alibaba/qwen-image/edit
• alibaba/qwen-image/max