Wan 2.5 — Edit an image

POST

/services/aigc/image2image/image-synthesis

curl --location 'https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \
  -H 'X-DashScope-Async: enable' \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "wan2.5-i2i-preview",
  "input": {
    "prompt": "Change the floral dress to a vintage-style lace long dress with exquisite embroidery details on the collar and cuffs.",
    "images": [
      "https://img.alicdn.com/imgextra/i2/O1CN01vHOj4h28jOxUJPwY8_!!6000000007968-49-tps-1344-896.webp"
    ]
  },
  "parameters": {
    "prompt_extend": true,
    "n": 1
  }
}'

{
  "output": {
    "task_status": "PENDING",
    "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx"
  },
  "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx"
}

Get an API key and set it as an environment variable. If using the SDK, install it.

Edit images using text prompts while preserving subject consistency. Supports multi-image fusion with up to three reference images.

Authorizations

string

header

required

DashScope API Key. Create one in the Qwen Cloud console.

Header Parameters

enum<string>

default"enable"

required

Enables asynchronous processing. Must be enable. HTTP requests only support asynchronous processing. Omitting this header returns a "current user api does not support synchronous calls" error.

Available options:enable

Body

application/json

enum<string>

required

Model name.

Available options:wan2.5-i2i-preview

Example:wan2.5-i2i-preview

object

required

Input data for image editing.

Show child attributes

string

required

A positive prompt describes the elements and visual characteristics that you want the generated image to include. Accepts Chinese and English. Maximum: 2,000 characters; longer content is truncated.

Example:Change the floral dress to a vintage-style lace long dress with exquisite embroidery details on the collar and cuffs.

string[]

required

An array of URLs pointing to the input images. Maximum: 3 images per request. For multiple images, array order defines the sequence (Image 1, Image 2, etc.).

Image limits:

Supported formats: JPEG, JPG, PNG, BMP, and WEBP. PNG alpha channels are ignored.
Resolution: Width and height must each be between 384 and 5,000 pixels.
Maximum file size: 10 MB.

Supported input formats:

Public URL: Both HTTP and HTTPS are supported.
Base64-encoded string: Format: data:{MIME_type};base64,{base64_data}.

Example:

[
  "https://img.alicdn.com/imgextra/i2/O1CN01vHOj4h28jOxUJPwY8_!!6000000007968-49-tps-1344-896.webp"
]

Required range:items: 1–3

string

The negative prompt. Elements to exclude from the generated image. Accepts Chinese and English. Maximum: 500 characters; longer content is truncated.

Example:low resolution, error, worst quality, low quality, disfigured, extra fingers, bad proportions

object

Controls output resolution, prompt rewriting, watermarks, and other processing options.

Show child attributes

string

default"1280*1280"

Output resolution in width*height format. Default: 1280*1280. Total pixels must fall between 589,824 (768768) and 1,638,400 (12801280). Aspect ratio must be between 1:4 and 4:1.

Recommended resolutions:

1280*1280 (1:1)
1024*1024 (1:1)
800*1200 (2:3)
1200*800 (3:2)
960*1280 (3:4)
1280*960 (4:3)
720*1280 (9:16)
1280*720 (16:9)
1344*576 (21:9)

If not specified, the system outputs an image with 1280*1280 total pixels and preserves an aspect ratio similar to the input:

Single-image input: The aspect ratio matches the input image.
Multi-image input: The aspect ratio matches the last input image.

Example:1280*1280

integer

default4

The number of images to generate. Valid values: 1 to 4. Default: 4. The n parameter directly impacts billing -- higher values mean higher costs. To control costs, it is recommended to explicitly set it to 1 for testing.

Example:1

Required range:1 <= x <= 4

boolean

defaultfalse

Whether to add a watermark in the bottom-right corner of the image that displays the fixed text "AI-generated".

Example:false

boolean

defaulttrue

Enables smart prompt rewriting. When enabled, the Large Language Model enhances your prompt for better results, at the cost of longer processing time.

Example:true

integer

The random number seed. Value range: [0, 2147483647]. If not specified, the algorithm automatically generates a random number as the seed. If specified, the algorithm generates a separate seed value for each of the n images (seed, seed+1, seed+2, ...). To reproduce specific results, use a fixed seed value. Note: Due to inherent randomness, identical seeds may not produce exactly identical results.

Required range:0 <= x <= 2147483647

Response

200-application/json

object

Show child attributes

string

The ID of the task. Can be used to query the task for up to 24 hours.

enum<string>

The status of the task.

Available options:PENDING,RUNNING,SUCCEEDED,FAILED,CANCELED,UNKNOWN

string

Unique identifier for the request. Use for tracing and troubleshooting issues.