CivArchive
    Andy Goldsworthy style nature art - portals, circles, spheres and spirals for SDXL - V1
    Preview 31543012Preview 31543091Preview 31542865Preview 31542996Preview 31542879Preview 31542991Preview 31542876Preview 31542856Preview 31542868Preview 31543011Preview 31542713Preview 31543555Preview 31542708Preview 31542711Preview 31543083Preview 31542709Preview 31543078Preview 31543079Preview 31543089Preview 31542712

    This LoRA was created to mimic the look of the artist Andy Goldsworthy's beautiful creations, which focus on constructions using natural materials. If you would like to see the kind of works intended to be created with this LoRA, search for his work online.

    Activation word: g0ldsw0rthy

    Reinforcement phrase: nature scene

    Recommended weight: 1.0

    Keyword concepts:

    portal - an upright construction of the kind you could step through

    circle - a circular pattern, most often laid on the ground, and often with a dark hole in the center

    sphere - a large constructed ball

    spiral - either a spiral of materials laid on the ground, or a material spiraling upright around other objects, most often trees

    zigzag, path - a crooked line (mostly an unsuccessful concept with too little training)

    gradient - applies a gradual color fade from the edge to the center

    close-up - typically gives you an overhead shot of the construction on the ground

    landscape - typically a wide shot of nature with the construction standing upright

    leaves, autumn leaves, sticks, rocks, stones, ice - the most common materials that the LoRA understands, however feel free to try many other types of materials, some are shown in the example images above

    spikes - most often associated with ice, will make it very spikey and jagged

    crack - most often associated with stones, will crack or otherwise draw a dark line through the materials

    stripes - likely a very undertrained concept that might still work regardless, makes the construction striped

    See the example images above for tips in prompting various concepts.

    I trained this LoRA with clip skip 2 in spite of common recommendations to use clip skip 1 for photography/realism, in the hope that this might help it be more compatible with Pony-based models. However, I have no idea what I'm doing or whether this was of any value.

    All images were generated with Juggernaut XL. I hope you find this LoRA interesting and fun to mess with!

    Please publish your generations so they get added to the gallery below this model, I would love to see them!

    Description

    First release

    LORA
    SDXL 1.0

    Details

    Downloads
    116
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/26/2024
    Updated
    9/27/2025
    Deleted
    -
    Trigger Words:
    g0ldsw0rthy

    Files

    Andy_Goldsworthy.safetensors

    Mirrors

    CivitAI (1 mirrors)