CivArchive
    ← All articles
    Published July 26, 2024by Decentre

    Decentre Studio Features

    71 views0 reactions0 comments on CivitAI0 collected
    announcement

    Is a combination of a Windows based application and an extension for WebUI Automatic 1111

    The WebUI Automatic1111 extension has the following functionality:-

    Object detection

    Via OS Vision model Yolo V3

    (other models TBC)

    Image to text Caption Generation 

    Via OS Llava ver 7

    (Other Models TBC)

    User will be able to select detection confidence threshold 0 to 1.00

    Select the isolated object image size 

    Regularised (Square 512x512 or 1024 x 1024)
    
    Non regularised ( any size and any resolution)

    Functionality to combine multiple Image to text models on the same set of detected assets, generating captions of different types (natural language  and/or deepbooru style tags) and have all these captions associated with assets

    When Complete the extension will store all the asset information  (image size, generation date, etc) and caption information (comma separated text) in a local database on the users PC.

    Windows Application has the following  functionality:-

    The windows application will read and database and provide the user with the following 

    Add / remove / edit captionsasset from the database Record

    Make a list of frequently used captions.

    The application will maintain a list of frequently used words/captions

    Bulk addition / removal of captions

    Exporting image text pair dataset in a user specified folder

    With associated settings 
    
    Generate a log of the export session.

    View / Merge / Edit  existing Image text pair datasets.

    Settings page to control various functions, application and storage settings.

    Auto Update checker

    This list is correct as of the 22nd of July 2024.