Introduction

A practical guide to using the Qwen Image Edit model in our Media Generator: options, costs, prompt tips, and best practices.

Qwen Image Edit is a fast, consistent image-editing model that requires a reference image. In our platform, it supports:

• Image editing — Modify existing images via prompt
• Consistent outputs — Strong fidelity to the reference
• Single reference — One input image per edit
• matchinputimage — Preserve input aspect ratio or choose a new one

Use it when you need to edit or transform an existing image with reliable, consistent results.

Where to Use Qwen Image Edit

| Location | How | |----------|-----| | Media Generator | Tools → Media Generator → Create Job → choose "Qwen Image Edit" → upload at least one reference image | | Image projects | Create a project, add reference images, and select Qwen Image Edit |

Requirements

• Reference image required — At least one input image is mandatory
• Reference defines the base content; the prompt describes the changes

Platform Options

Aspect ratio

| Option | Use case | |--------|----------| | matchinputimage | Keep input aspect ratio (default) | | 1:1 | Square output | | 16:9 | Landscape | | 9:16 | Portrait | | 4:3 | Standard | | 3:4 | Photo |

Default:

• match

input

• image

Output format

• PNG — Lossless
• JPG — Smaller file size

Default:

• PNG

Reference images

• Required — Minimum 1 image
• Single reference — One input image per generation (unlike Pruna, which supports multiple)
• Best for focused edits on a single source image

Costs (credits)

Qwen Image Edit uses flat per-image pricing:

| Item | USD | Credits | |------|-----|----------| | Per image | $0.03 | 0.08 |

\Credits computed with default billing config. Check the model selector in the UI for current rates.

Prompt guidelines

Do

• Describe what to change — e.g., "change background to white," "add autumn leaves"
• Be precise about edits — "replace the sky with clouds," "remove the text overlay"
• Use clear, action-oriented language
• Focus on one main edit per run for best consistency

Avoid

• Vague instructions ("make it better," "improve it")
• Asking for completely unrelated new content
• Multiple conflicting edit requests in one prompt

Good practices

1. Use a clear reference image — Sharp, well-lit inputs give more consistent edits.
2. Be specific in the prompt — Describe the exact change you want.
3. Use matchinputimage when possible — Preserves composition and proportions.
4. One main edit at a time — For complex changes, run multiple edits in sequence.

Limits and constraints

| Constraint | Value | |------------|-------| | Reference images | Required, 1 per generation | | Max images per request | 5 |

Qwen vs Pruna Image Edit

| Feature | Qwen Image Edit | Pruna Image Edit | |---------|-----------------|------------------| | Reference images | 1 required | 1–10 supported | | Cost per image | 0.08 credits | 0.03 credits | | Strengths | Consistency, reliability | Multiple refs, lower cost |

Choose Qwen when you want strong consistency on a single image; choose Pruna when you need to blend multiple references or keep costs lower.

Troubleshooting

| Issue | What to try | |-------|-------------| | Missing reference warning | Upload at least one reference image | | Unexpected or inconsistent edits | Simplify the prompt; focus on one main change | | Model not available | Confirm Replicate API token is configured |