CivArchive
    Face and hand restorer - fast test workflow - v1.0
    Preview 2322333Preview 2322364Preview 2322363Preview 2322358Preview 2322360

    Extra fast workflow for setting and testing of face restorer and hand restorer.

    Version 1.1 is exactly same as previous 1.0, but you can change the mood.

    Check the attached example images.

    This workflow much faster than general workflows, because no need to resample images for several settings.

    How to use:

    1; Generate and save images on your own working environment, but don't restore faces and hands, the best if just save the raw image output without post-processing, restoring, and upscaling.

    2; Copy seed and prompt of original generation process to the right input fields of input block. Set the same models, VAE, etc. Download SAM and detector models if required by the saved name. This is very important to be exactly same as your original workflow. Don't use the example images.

    3; Open your own generated images in image loader.

    4; Set several details to same image in face restorer and hand restorer group, and compare the result. Modify the additional prompts, maybe you can merge with original prompt. Set the text encoders and all what you want.

    5: Copy settings to your production environment and test with new generated image.

    6: If the new image wrong, back to 2, and set the environment again.

    7: Good luck :)

    Check the workflow with tested attached example:

    1: Select attached workflow_example.png

    2: Original prompts, seed, size, etc... already set to workflow json.

    3: Run and check the result.

    4: Try with your own image.

    PS: Don't forget to download missing nodes :)

    Description

    First version. Try refine if you like.

    Workflows
    Other

    Details

    Downloads
    78
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    9/2/2023
    Updated
    8/9/2025
    Deleted
    8/7/2025

    Files

    faceAndHandRestorer_v10.ZIP

    Mirrors

    CivitAI (1 mirrors)