LoadImageTextSetFromFolderNode - ComfyUI Built-in Node Documentation

Loads a batch of images and their corresponding text captions from a specified directory for training purposes. The node automatically searches for image files and their associated caption text files, processes the images according to specified resize settings, and encodes the captions using the provided CLIP model.

Inputs

Parameter	Description	Data Type	Required	Range
`folder`	The folder to load images from.	STRING	Yes	-
`clip`	The CLIP model used for encoding the text.	CLIP	Yes	-
`resize_method`	The method used to resize images (default: “None”).	COMBO	No	”None" "Stretch" "Crop" "Pad”
`width`	The width to resize the images to. -1 means use the original width (default: -1).	INT	No	-1 to 10000
`height`	The height to resize the images to. -1 means use the original height (default: -1).	INT	No	-1 to 10000

Note: The CLIP input must be valid and cannot be None. If the CLIP model comes from a checkpoint loader node, ensure the checkpoint contains a valid CLIP or text encoder model. Note on folder structure: The node supports the kohya-ss/sd-scripts folder structure. If a subfolder’s name starts with a number followed by an underscore (e.g., 5_myclass), that number is used as a repeat count, and the images inside that subfolder will be loaded that many times.

Outputs

Output Name	Description	Data Type
`IMAGE`	The batch of loaded and processed images.	IMAGE
`CONDITIONING`	The encoded conditioning data from the text captions.	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): ffd6399783fc281a58bae811112d9ecacb51ab8ea3b512befa9b9fab2c6860de

LoadImageTextDataSetFromFolder - ComfyUI Built-in Node Documentation

LoraLoader - ComfyUI Built-in Node Documentation

⌘I

​Inputs

​Outputs

Inputs

Outputs