CivArchive
    Tool that shuffles and picks x amount of prompts from input prompt file(s) - v1.0
    Preview 278342
    Preview 278344
    Preview 278343

    Updated 21.3.:

    • Support for multiple input files added

    • Extended sample range to 10 000 by default


    Tool that helps with selecting a random amount of prompts from a file that contains prompts. I am using it when testing the different prompt packages I am uploading. I'll take a big enough sample to generate a few images. Remove and fix obvious maligned prompts, rinse and repeat.

    Requirements

    • pip install gradio

    Usage

    • gradio guitoolkit.py

    • or use python guitoolkit.py

    How to guide

    1. Download this file / copy the code below into a file called guitoolkit.py (or whatever you want to call it)

    2. Make/use a virtual environment python -m venv venv

    3. Activate environment venv\Scripts\activate

    4. Run the command pip install gradio to install the gradio library which is required to use this

    5. When you have installed that, run either gradio guitoolkit.py or python guitoolkit.py

    6. You should now have the tool ready to use if you get the following: gradio .\guitoolkit.py
      launching in reload mode on: http://127.0.0.1:7861 (Press CTRL+C to quit)

    7. You can now visit http://127.0.0.1:7861 where the tool is ready to use

    8. Input the file(s) you want to shuffle, select how many you want, copy the output, insert it into e.g. Automatic1111


    Source code

    import gradio as gr
    import random
    
    def shuffle_file(file_obj, no_prompts):
        prompts = []
        for file in file_obj:
            with open(file.name) as infile:
                in_prompts = infile.readlines()
            prompts.extend(list(set(in_prompts)))
        
        prompts = random.sample(prompts, no_prompts)
        random.shuffle(prompts)
        print(type(prompts))
        return "".join(prompts)
    
    demo = gr.Interface(
        fn=shuffle_file,
        inputs=["files",  gr.Slider(5, 10000)],
        outputs=["code"],
    )
    
    if __name__ == "__main__":
        demo.launch(server_port=9800)

    Description

    Initial release

    FAQ

    Comments (6)

    PolygonMar 19, 2023· 1 reaction
    CivitAI

    Can't one use wildcards and other already existing tools for this purpose?

    Automaticism
    Author
    Mar 19, 2023· 1 reaction

    Sure you can, this is mostly for supporting quick prompting via the default Script "Prompts from file or textbox" which enables you to copy paste all the prompts into the input box.

    I found it easier to handle prompt generation outside of wildcards plugin, but that might be because I ended up building different CLI based tooling outside of automatic1111.

    Athos252Mar 22, 2023· 1 reaction

    @Automaticism Please continue to work on this, I do not like the way Wildcards is setup I find it difficult to use and manage. I think alternatives are much needed.

    Automaticism
    Author
    Mar 22, 2023

    @Athos252 is there something specifically you are looking for in a tool?

    Athos252Mar 28, 2023

    @Automaticism Yeah, I kinda think something that will pull a random embedding, preferably from a specific folder so say you have a folder of all your female people (celebs) and you want to make like 4 images of 4 different random celebs to test out who would fit into this image. You don't then need to make and maintain a text doc with all your embeddings. Add this to alternate rendering (the [subject|subject] format) and I think it would be a cool tool for making different faces using random people and character creation.

    Automaticism
    Author
    Mar 29, 2023

    @Athos252 but you can do this with sd dynamic prompts plugin? https://github.com/adieyal/sd-dynamic-prompts

    Other
    Other

    Details

    Downloads
    175
    Platform
    CivitAI
    Platform Status
    Available
    Created
    3/19/2023
    Updated
    5/13/2026
    Deleted
    -

    Files

    toolThatShufflesAndPicksX_v10.zip

    Mirrors

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.