QwenImageEditApi - ComfyUI Built-in Node Documentation

This node uses the Qwen-Image 3.0 models to edit or combine up to 3 reference images guided by a text prompt. You select a model, provide the prompt and reference images, and the node returns one or more generated images.

Inputs

Common Inputs

Parameter	Description	Data Type	Required	Range
`model`	Model to use. This selection also includes the text prompt, up to 3 reference image inputs, and an optional negative prompt.	DYNAMIC_COMBO	Yes	”qwen-image-3.0-pro" "qwen-image-3.0”
`size`	Output resolution. “match input” reuses the first reference image’s size, “auto” lets the model pick a size with the same aspect ratio, “custom” sets an explicit width and height.	DYNAMIC_COMBO	Yes	”match input" "auto" "custom”
`n`	Number of images to generate, returned as a batch. (default: 1)	INT	No	1 to 6
`seed`	Seed to use for generation. (default: 42)	INT	No	0 to 2147483647
`prompt_extend`	Whether to enhance the prompt with AI assistance. (default: True)	BOOLEAN	No	True False
`watermark`	Whether to add an AI-generated watermark to the result. (default: False)	BOOLEAN	No	True False

qwen-image-3.0-pro and qwen-image-3.0 Inputs

Both models share the same sub-parameters.

Parameter	Description	Data Type	Required	Range
`prompt`	Editing instructions. Supports English and Chinese, and @Image1-style references to the input images. (default: "")	STRING	Yes	-
`negative_prompt`	Negative prompt describing what to avoid. (default: "")	STRING	No	-

Reference Inputs

Parameter	Description	Data Type	Required	Range
`images`	Growable slot: connect 1 to 3 reference images (`image_1`, `image_2`, `image_3`). Refer to them in the prompt as @Image1, @Image2, @Image3, numbered in input order; a batched input counts once per image.	IMAGE	Yes	1 to 3

Custom Size Inputs

Shown when size is set to “custom”.

Parameter	Description	Data Type	Required	Range
`width`	Output width. The total pixel area must be between 512x512 and 2560x2560; the aspect ratio must be between 1:8 and 8:1. (default: 1024)	INT	Yes (when `size` is “custom”)	256 to 2560, step 16
`height`	Output height. The total pixel area must be between 512x512 and 2560x2560; the aspect ratio must be between 1:8 and 8:1. (default: 1024)	INT	Yes (when `size` is “custom”)	256 to 2560, step 16

Constraints

The text prompt is required and must contain at least one character.
A maximum of 3 reference images is supported; an error is raised if more are provided (a batched input counts once per image).
When size is set to “custom”, explicit width and height values must be provided and are validated: the total pixel area must be between 262,144 (512x512) and 6,553,600 (2560x2560) pixels, and the aspect ratio must be between 1:8 and 8:1.
When size is set to “match input”, at least one reference image is required because the first reference image’s dimensions are used; the dimensions are scaled to fit the supported area and aspect ratio range.
When size is set to “auto”, the model picks the output size (1.9-4.2 megapixels) while preserving the input aspect ratio.
Prompt references use @Image1, @Image2, @Image3, numbered in input order; a reference to an index higher than the number of connected images raises an error. Tags are recognized only at word boundaries, so addresses like user@image1.com are left unchanged.
Input reference images are downscaled to at most 2048x2048 pixels before being sent to the API. If the PNG encoding exceeds the API size limit, a JPEG encoding is used instead.

Outputs

Output Name	Description	Data Type
`IMAGE`	The generated image or images returned as a batch. Up to `n` images are returned.	IMAGE

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): efa8d2b1a039a7b91789c0332b751a5f90ab8dad755ef0e25124d7d1c44d9abb

​Inputs

​Common Inputs

​qwen-image-3.0-pro and qwen-image-3.0 Inputs

​Reference Inputs

​Custom Size Inputs

​Constraints

​Outputs