The Qwen Image 2 Pro Edit API is built for changing an existing image rather than inventing every pixel again. That makes it useful for AI character consistency, ecommerce variations, outfit changes, product placement, and campaign localization.

On MakeInfluencer, the model accepts one to six reference images plus a written edit instruction.

## Model Inputs

| Field         | Requirement                                                                                  |
| ------------- | -------------------------------------------------------------------------------------------- |
| `prompt`      | Required edit instruction                                                                    |
| `imageUrls`   | Required array with 1–6 reference image URLs                                                 |
| `aspectRatio` | Defaults to `match_input_image`; also supports common square, landscape, and portrait ratios |

See the live [Qwen Image 2 Pro Edit API page](/developers/models/qwen-image-2-pro-edit-api) for the exact enum values and credit quote.

## Better Edit Instructions

Good editing prompts say what must remain unchanged as well as what should change.

Weak:

> Put her in a different outfit.

Stronger:

> Keep the same person, facial geometry, hair, pose, camera angle, and studio lighting. Replace the black jacket with a cream linen blazer and white silk top. Preserve the background and crop.

That structure reduces accidental changes to already-approved parts of the frame.

## Preview and Generate

```bash
curl --request POST \
  https://www.makeinfluencer.ai/api/v1/models/qwen-image-2-pro-edit/preview-credits \
  --header "Authorization: Bearer $MAKEINFLUENCER_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "input": {
      "prompt": "Keep the same person and lighting. Add the supplied handbag naturally to her left hand.",
      "imageUrls": [
        "https://example.com/approved-model.jpg",
        "https://example.com/product-reference.jpg"
      ],
      "aspectRatio": "match_input_image"
    }
  }'
```

After the preview is approved, POST the same body to `/generations`. The default single-copy request uses the existing image-editing path and returns one queued generation ID.

## Multi-Reference Use Cases

<FeatureGrid columns="2">
  <Feature
    icon="users"
    title="Character plus Product"
    description="Use an approved creator image and a product reference to guide a campaign edit."
    index={0}
  />
  <Feature
    icon="palette"
    title="Style Transfer"
    description="Combine a source subject with a visual mood, material, or wardrobe reference."
    index={1}
  />
  <Feature
    icon="layers"
    title="Campaign Localization"
    description="Adapt signage, setting, or styling while preserving the core approved asset."
    index={2}
  />
  <Feature
    icon="image"
    title="Composition Guidance"
    description="Supply references for objects or framing instead of relying on a long text-only description."
    index={3}
  />
</FeatureGrid>

The model does not know which image is authoritative unless the prompt explains each reference's role. Use language such as “first image is the person; second image is the product; preserve the first image's face and framing.”

## Aspect Ratio Strategy

Start with `match_input_image` when the composition is already approved. Changing aspect ratio during the same edit forces the model to solve both the creative change and a re-composition.

For platform variants, use a two-pass workflow:

1. Approve the edit at the original ratio.
2. Create separate portrait, square, or landscape adaptations from the approved edit.

This makes review easier and avoids rejecting a good product edit because the first reframe failed.

## Qwen Edit vs Fresh Generation

Choose Qwen Image 2 Pro Edit when most of the image should survive. Choose a generation model such as [Dreamina V3](/developers/models/dreamina-v3-api) or [MakeInfluencer Flagship](/developers/models/ai-influencer-image-api) when the brief needs a fundamentally new composition.

<Callout type="tip" title="Protect Expensive Downstream Steps">
  Approve the edited still before sending it to video. Catching a changed face
  or incorrect product in the image stage avoids wasting animation and review
  credits.
</Callout>

## API FAQ

### How many input images can I use?

The current MakeInfluencer registry accepts one to six reference image URLs.

### Can the output match the source dimensions?

Use `aspectRatio: "match_input_image"`. The service computes a valid output size from the selected aspect behavior.

### Is credit pricing per reference image?

Use `preview-credits` for the current quote. The model page is generated from the same pricing function used by generation.

### Can I create several copies?

The generic generation body supports image copies, but precision editing is easiest to reason about one output at a time. Queue distinct creative variations with distinct idempotency keys when you need controlled alternatives.


Use the Qwen Image 2 Pro Edit API for precision AI image edits with up to six references, controlled aspect ratios, shared credits, and production examples.

Qwen Image 2 Pro Edit API Guide

Can the output match the source dimensions?

Qwen Image 2 Pro Edit API Guide

Create Your AI Influencer

Continue Reading

Z-Image Turbo API Guide: Fast AI Image Generation

AI Image Editing Guide

Faceless YouTube Channel with AI

Ready to try it yourself?