Minimal tool to extract the generation metadata of an image produced by Automatic1111 or Fooocus, such as prompt, negative prompt and model name.
You will need have Python and Anaconda downloaded before installing Prompt Retreiver.
Open a terminal where you want to install Prompt Retreiver and clone this repository:
git clone https://github.com/brayevalerien/Prompt-retreiverThen cd into the cloned repository and create a new Anaconda environment and activate it:
cd Prompt-retreiver
conda create -n promptretreiver -y
conda activate promptretreiverInstall the required libraries:
pip install -r requirements.txtEverything is now setup to start using Prompt Retreiver.
Run start.bat to start the Gradio server of Prompt Retreiver. It uses local port 6006 so make sure no other application is using it.
Simple drop or upload an image you have generated and it will display the image generation informations in the textboxes bellow, if they were saved in the image. You can then copy the informations and reuse them elsewhere.
Note that vanilla Fooocus requires you to manually check the option for saving metadata to images. If your image does not contain metadata, nothing will be output in the textboxes.
Prompt Retreiver should remain a minimal tool that is fast to start and to use, but feel free to contribute to the project if you feel like it could improve. There are two main ways to do so:
- Opening an issue if you have a suggestion of if you have found a bug
- Opening a pull request if you have forked the project and added things to the source code
Both types of contributions are very helpfull!
Mandatory features:
- build fonctional gradio UI with callbacks
- implement the function to extract metadata
Optional features:
- rearange the UI to make it easier to work with
- support for both Fooocus and Automatic1111 metadata schemes
- cleaner parsing code
- integrate into Fooocus Lite?
- add one-click launching script