Running Stable Diffusion web interfaces on my computer, I’d been wondering if there is a better way to have a script that can read the TXT information embedded in the PNG generated by different interfaces like ComfyUI, Automatic1111 and InvokeAi etc.

I use mostly ComfyUI and Automatic1111 myself and to get prompt info or other generation parameters I need ensure I save the accompanying TXT file or using PNG info tool in Automatic1111 but this means I need to run up the webUI each time.

So I researched a bit and found this wonderful standalone Diffusion Prompt Reader which does a wonderful job of reading and displaying prompt information and other metadata in a very nicely laid out User Interface.

Installation

There is none, simply download the application and create a shortcut to start using it.

It supports various different formats and images created from different Web UIs.

Usage

You can see below some screenshots from my PC using the app with two different images: Automatic1111 (A1111 WebUI) and ComfyUI.

In ComfyUI case if you had multiple node with prompts (positive + negative) as per your workflow you will see various tabs below the prompt to allow you to switch between them eg. Clip G – Clip L – Refiner etc.

The most common use case for me is to copy the prompt to use in another AI tool like Midjourney, Leonardo.AI or grab prompt from A1111 image for use in ComfyUI & vice versa. I use many different tools so its handy for me to use this standalone app to just grab the prompt, seed, model and LoRA info.

You can edit the prompt data if you need to (not sure why you would but you can), export the prompt data into TXT file, remove the prompt data from the PNG before sharing with someone.

I also did a review of the SD Prompt Reader in one of my videos on the channel. You can watch it below

All and all I’m very happy that somebody has made this application that let’s you do this, as a techie for me who is constantly experimenting with new tools, its very handy for me to go back and grab prompts from previous images without cluttering my PC with TXT files.