Automatic Captioning Workflow - FLUX/Stable Diffusion 3.5 - ComfyUI — v1.1

Automatic Captioning Workflow - FLUX/Stable Diffusion 3.5 - ComfyUI is for captioning a folder of images: it uses Florence2 to generate text captions and saves matching text files in the same folder.

The workflow reads images from a folder, generates captions with Florence2, and writes the caption text files back to that folder.

It includes steps for loading images from a directory, running Florence2, previewing images, and saving text files.

Input: a folder full of images. Output: caption text files saved in the same folder with the images' correct names.

Before running, use the ComfyUI Manager to install any missing nodes.

Suggestion · not verified

Running the workflow is described as automatically downloading the Florence2 model.

Check your ComfyUI installation for Florence2ModelLoader, Florence2Run, DF_Text_Box, LoadImageListFromDir //Inspire, and Fast Groups Bypasser (rgthree).

Suggestion · not verified

Also check for CR Text Replace, JWStringSplit, String Switch // RvTools, Text Concatenate, Save Text File, PreviewImage, ShowText|pysssss, and String.

Suggestion · not verified

Version v1.1 adds pre- and post-text fields for captions.

Flux.1 D is the listed base model.

License terms are not specified, so verify the applicable terms before use.

Suggestion