TextEncodeQwenImageEdit - ComfyUI Built-in Node Documentation

The TextEncodeQwenImageEdit node converts text prompts and optional images into conditioning data for image generation or editing. It uses a CLIP model to tokenize the input and can optionally encode reference images with a VAE to create reference latents. When an image is provided, it is automatically resized to keep a consistent processing scale.

Inputs

Parameter	Description	Data Type	Required	Range
`clip`	The CLIP model used for text and image tokenization	CLIP	Yes	-
`prompt`	Text prompt for conditioning generation, supports multiline input and dynamic prompts	STRING	Yes	-
`vae`	Optional VAE model for encoding reference images into latents	VAE	No	-
`image`	Optional input image for reference or editing purposes	IMAGE	No	-

Note: When an image is provided, it is resized so its total pixel count stays close to 1,048,576 (1024 × 1024), and only its RGB channels are used. The resized image is passed to the CLIP tokenizer together with the prompt. When both image and vae are provided, the node also encodes the image into reference latents and attaches them to the conditioning output.

Outputs

Output Name	Description	Data Type
`CONDITIONING`	Conditioning data containing text tokens and optional reference latents for image generation	CONDITIONING

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): ec6980a63eab0d6c95be3abea00b2bf3018d30a1267f0b39a21be29a3e9228fe

​Inputs

​Outputs

Inputs

Outputs