CivArchive
    Preview 22006514Preview 22006519Preview 22006532Preview 22006515

    Do not use these styles with other SDXL models, you just won't see any effect, These styles are for Pony Models only!

    LoRA have been trained on the Pony Diffusion V6 XL (start with this one) and I do not guarantee out good quality on other Pony Models

    Preview generated without using ADetailer & Hires. fix

    Recommended LoRA Strength (Weight): 0.5-0.8

    About Styles

    The OIS in the name stands for One Image Style, these models are literally trained on a single image

    ALT (Alternative) These versions of styles differ in that they are of higher quality, more sharpness and fewer JPEG artifacts (These styles are not a substitute!)

    About Smooth Styles

    Smooth Boys - Smooth Boys is a more honest name for Smooth/Smooth 2, because it was the original name of Smooth 1 before the renaming and because there are only muscular guys with smooth skin in the dataset. This version can be considered a replacement for Smooth/Smooth 2 (The dataset is similar to Smooth 2 but better not so sharp)

    Smooth Boys Colored Skin - Improved the creation of colored skin in this style, I selected several images and edited them manually to change the skin color. The style may be slightly different from Smooth Boys, This style is not perfect but it looks good (In About this version you can find a list of skin colors)

    Smooth Boys Low Light - The images were edited by me manually to make them as dark as possible without losing the style, the style may be slightly different from Smooth Boys

    These styles* don't have Trigger Words, but you can use realistic to change the style.

    This style is trained on images not created by me, as far as I know the images were created by NAI3, But I do not know whether the names of the artists were used or not. Therefore, this is the most controversial style of Not Artists Styles

    *Smooth Boys/Smooth Boys Colored Skin

    First Seven Styles

    These styles were created by the Pony Diffusion V6 XL model itself, LoRA were trained on images created by Me using img2img and abstract textures

    Blue Water, this style creates a watercolor effect with a slight cool effect

    Green Patina, this style creates a whiter textured watercolor effect with a slight bluish-greenish effect

    Cosmic Void, this style creates a dark and mysterious watercolor effect, and can create nighttime images

    Turquoise, this style creates a cool turquoise watercolor effect as if everything is underwater

    Pink Dreams, this style creates a soft pink hue on the image, the style is most similar to digital painting. The tokens pinkdreamsstyle, digital art, have less effect than concept art, realistic, but it is better to use these tokens together

    Red Lion, this style creates a very nice warm effect in the images and a slight watercolor effect and imperceptible paper texture effect

    Gold Fish, this style creates a vintage watercolor effect and a very strong paper texture effect

    These Styles won't work without Trigger Words!

    Styles are based partially or entirely on images created by the Neural Network (Artificial Intelligence)

    Description

    Repeats: 10

    Epoch: 4

    Image Count: 72

    Network Rank (Dimension): 16

    Network Alpha: 8

    Batch size: 1

    Clip skip: 1

    Training steps: 2880

    Resolution: 576x832

    Train Resolution: 1024,1024

    LR Scheduler: cosine_with_restarts

    Optimizer: AdamW8bit

    Learning rate: 0,0005

    Text Encoder learning rate: 0,00005

    Unet learning rate: 0,0005

    Captioning: WD14 Captioning (SmilingWolf/wd-v1-4-moat-tagger-v2) + Edit

    Recommended LoRA Strength (Weight): 0.7-0.8

    List of skin colors - red skin, orange skin, yellow skin, green skin, dark green skin, blue skin, cyan skin/teal skin, purple skin, pink skin, pale skin/white skin, dark skin/brown skin,