Create an Image or Change a Reference Photo · Alibaba

Describe a new scene, or supply your own image and say what to change, along with an explicit output size. Up to three references can guide an edit, and successful results provide downloadable images.

Your AI

Use with your AI

For tasks with several steps, continue in the AI you already use.

CodexClaude CodeCursorWindsurfCline

View instructions

Use Toolmu generate-and-edit-images-fast-alibaba (Alibaba Qwen Image 3.0 standard, ID 4212699e-e0b3-48b8-9275-ab3ef7c5c0c1) to change my photo background. Reuse the existing connection; if setup is needed, follow https://toolmu.com/SKILL.md within my existing authorization and budget. First ask me for a real, publicly accessible HTTPS URL of my photo; do not invent a URL or use a sample asset. Once supplied, put that URL first in body.images. Set body.prompt to: Replace the background of the first photo with plain light gray; try to retain the subject outline, colors and main details, and add no text. Set body.size=1024*1024, body.n=1 and body.prompt_extend_mode=direct. Use an asterisk between width and height. Perform this one edit within the authorized task and budget. Show the actual successful image and available download link, ask me to inspect the subject and background, and remind me to save the temporary download promptly. Do not guarantee unchanged identity or a fixed completion time.

Task:
Replace [the cluttered room behind me] in [my photo] with [a plain light-gray background]. Keep [my face and clothing] close to the original. I will share [a public link that opens the photo itself without signing in]. Make [one square image] and give me a link to save it.

Paste into your connected AI's conversation.

Connection guide

Up to 0.2268 credits / call

The final charge is based on confirmed usage. Unused reserved credits return to your available balance.

Alibaba

Create an Image or Change a Reference Photo · Alibaba

You receive

Receive images generated from a description or an edit of supplied references. For a photo, you can request a light-gray background while asking to retain the subject. Choose 1–6 requested outputs, default one, then review and save the actual images returned.

Start with your task

Replace [the cluttered room behind me] in [my photo] with [a plain light-gray background]. Keep [my face and clothing] close to the original. I will share [a public link that opens the photo itself without signing in]. Make [one square image] and give me a link to save it.

What to know

Output size must meet the total-pixel and aspect-ratio rules, and references require eligible public HTTPS links. Subject details, lettering and edits can differ from the instruction; actual count and dimensions must be checked. New model download links typically last about 24 hours.

Inputs & outputs

Inspect the current inputs, availability, and maximum charge before starting.

Input schema

Input schema
Field / Path	Type	Requirement	Meaning
body`body`	`object`	Required	Required fields `prompt, size`
Requested image count`body.n`	`integer`	Optional	Optional integer from 1 to 6, default 1. Sets the requested number of generated or edited images; check the actual successful count. Default `1` Minimum `1` Maximum `6`
Random seed`body.seed`	`integer`	Optional	Optional integer from 0 to 2147483647. Helps repeat the random setup, but using the same seed does not guarantee identical images. Minimum `0` Maximum `2147483647`
Output dimensions`body.size`	`string`	Required	Required widthheight string, for example 10241024; use an asterisk, not x or auto. For both tasks, total pixel area is 512×512 to 2048×2048 and aspect ratio 1:8 to 8:1. These are not independent side bounds, and actual dimensions may vary slightly. Pattern `^\d+\*\d+$`
Ordered reference images`body.images`	`array`	Optional	Optional array of up to three public HTTPS image links. Their order defines the reference order. For text-only generation, omit this whole field; do not assume an empty array selects that mode. Maximum items `3`
Reference image link`body.images[]`	`string`	—	A nonempty, publicly accessible HTTPS image URL. JPG, JPEG, PNG, BMP, TIFF, WEBP and GIF inputs are supported, each at most 10 MB. Sides of 384–2048 pixels are recommended rather than mandatory bounds. Minimum length `1`
Scene or editing instruction`body.prompt`	`string`	Required	Required nonempty text describing the new scene, or the change to make to supplied images. The recommended maximum is 4500 tokens, not a 4500-character limit. State any subject details you want to retain; adherence is not guaranteed. Minimum length `1`
Generated-image watermark`body.watermark`	`boolean`	Optional	Optional boolean, default false. Requests an AI Generated watermark when enabled. Default `false`
Prompt rewriting`body.prompt_extend`	`boolean`	Optional	Optional boolean, default true. Enables the model to rewrite the prompt before generation or editing; it does not guarantee a better result. Default `true`
Thinking mode`body.enable_thinking`	`boolean`	Optional	Optional boolean; the documented default is true. Requires prompt_extend=true and increases processing time, without guaranteeing image quality or fidelity.
Exclusion guidance`body.negative_prompt`	`string`	Optional	Optional text describing elements you would prefer to exclude. It guides the model but does not guarantee those elements will be absent.
Rewriting approach`body.prompt_extend_mode`	`string`	Optional	Optional direct or agent value; the documented default is direct. direct supports generation and editing. agent performs more extensive rewriting only for text-only generation without input images. Options `direct, agent`

View raw JSON

{
  "type": "object",
  "required": [
    "body"
  ],
  "properties": {
    "body": {
      "type": "object",
      "$schema": "https://json-schema.org/draft/2020-12/schema",
      "required": [
        "prompt",
        "size"
      ],
      "~standard": {
        "vendor": "zod",
        "version": 1,
        "jsonSchema": {}
      },
      "properties": {
        "n": {
          "type": "integer",
          "default": 1,
          "maximum": 6,
          "minimum": 1,
          "description": "Optional integer from 1 to 6, default 1. Sets the requested number of generated or edited images; check the actual successful count."
        },
        "seed": {
          "type": "integer",
          "maximum": 2147483647,
          "minimum": 0,
          "description": "Optional integer from 0 to 2147483647. Helps repeat the random setup, but using the same seed does not guarantee identical images."
        },
        "size": {
          "type": "string",
          "pattern": "^\\d+\\*\\d+$",
          "description": "Required width*height string, for example 1024*1024; use an asterisk, not x or auto. For both tasks, total pixel area is 512×512 to 2048×2048 and aspect ratio 1:8 to 8:1. These are not independent side bounds, and actual dimensions may vary slightly."
        },
        "images": {
          "type": "array",
          "items": {
            "type": "string",
            "minLength": 1,
            "description": "A nonempty, publicly accessible HTTPS image URL. JPG, JPEG, PNG, BMP, TIFF, WEBP and GIF inputs are supported, each at most 10 MB. Sides of 384–2048 pixels are recommended rather than mandatory bounds."
          },
          "maxItems": 3,
          "description": "Optional array of up to three public HTTPS image links. Their order defines the reference order. For text-only generation, omit this whole field; do not assume an empty array selects that mode."
        },
        "prompt": {
          "type": "string",
          "minLength": 1,
          "description": "Required nonempty text describing the new scene, or the change to make to supplied images. The recommended maximum is 4500 tokens, not a 4500-character limit. State any subject details you want to retain; adherence is not guaranteed."
        },
        "watermark": {
          "type": "boolean",
          "default": false,
          "description": "Optional boolean, default false. Requests an AI Generated watermark when enabled."
        },
        "prompt_extend": {
          "type": "boolean",
          "default": true,
          "description": "Optional boolean, default true. Enables the model to rewrite the prompt before generation or editing; it does not guarantee a better result."
        },
        "enable_thinking": {
          "type": "boolean",
          "description": "Optional boolean; the documented default is true. Requires prompt_extend=true and increases processing time, without guaranteeing image quality or fidelity."
        },
        "negative_prompt": {
          "type": "string",
          "description": "Optional text describing elements you would prefer to exclude. It guides the model but does not guarantee those elements will be absent."
        },
        "prompt_extend_mode": {
          "enum": [
            "direct",
            "agent"
          ],
          "type": "string",
          "description": "Optional direct or agent value; the documented default is direct. direct supports generation and editing. agent performs more extensive rewriting only for text-only generation without input images."
        }
      },
      "additionalProperties": false
    }
  },
  "additionalProperties": false
}

Result schema

Output fields have not been documented yet.

Additional notes

For a reference-photo edit, make the original image and the change easy to identify. Describe the background you want, the main subject and the details to retain. Reference links must be public HTTPS; supported JPG, JPEG, PNG, BMP, TIFF, WEBP or GIF inputs are at most 10 MB each. Sides from 384 to 2048 pixels are recommended for input quality rather than mandatory side bounds.

This Tool uses standard Qwen Image 3.0 for generation and editing. A square output of width 1024 and height 1024 is a useful starting point for the example background edit. Output dimensions must satisfy the total pixel area and aspect ratio rules. Generative editing can change subject details or lettering, and processing time varies; inspect the finished image before using it and save the temporary download promptly.

Alibaba Qwen image generation and editing documentation

Questions about this tool.

Inputs, results and details to know before running it.

Get connected

No. Supply a nonempty description and explicit output dimensions, and omit the images field entirely. A photo is needed for the reference-image editing task; do not rely on an empty array to select text-only generation.

Create an Image or Change a Reference Photo · Alibaba

Create an Image or Change a Reference Photo · Alibaba

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Share a photo link anyone can open

Change a photo background

Check and save the edited image

Questions about this tool.

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Share a photo link anyone can open

Change a photo background

Check and save the edited image

Questions about this tool.

Do I need a reference photo to create a new image?

Can an edit use the original photo dimensions automatically?

Will changing the background leave the subject exactly unchanged?