CivArchive
    Kyoto Animation | Style LoRA | 京アニ スタイル - v2.0 (LoRA merge) [288mb]
    Preview 858843
    Preview 858844
    Preview 858845
    Preview 858852
    Preview 858854
    Preview 858855
    Preview 858858
    Preview 858860
    Preview 858867
    Preview 858875
    Preview 858886
    Preview 858887
    Preview 859065

    KYOTO ANIMATIONS

    Style LoRA

    (EN/日本語/中文)

    |___________________________________________________________________________________________|

    About this LoRA:

    1) This LoRA has been trained on images from the Anime Series >Violet Evergarden<.

    2) Works well with characters and landscapes.

    Checkpoints where this LoRA can really shine:

    powercolorV1 + pastel-waifu-diffusion VAE

    counterfeitV3.0

    vrwaifumix_v20 (for landscapes)

    >Trained with >CouterfeitV3.0<

    >Recommendations:

    -->LoRA strength: 0.8

    -->Highres fix: Latent

    -->Sampler: DPM++ 2M Karras

    -->CFG scale: 7

    __________________________________________________________________________________

    Prompt example:

    Prompt: 1girl, (masterpiece,best quality:1.6), scenery,night sky, open sky, stars, star light, milky way, starry sky, black_skirt, black shoes, full_body, green_eyes,necktie, pleated_skirt, dark blue hair, school_uniform, short_hair, short_sleeves, sweater_vest, white_shirt, short red hair, full body, standing,<lora:VioletEvergarden1.0:0.8>

    Negative: EasyNegative, bad anatomy, bad hands, text, error, extra leg, extra finger, deformed fingers,extra person, extra character,forest, trees, objects, covered sky,

    __________________________________________________________________________________

    このLoRAについて

    1) このLoRAは、アニメシリーズ >Violet Evergarden<の画像に学習させたものです。

    2)キャラクターや風景に適しています。

    このLoRAが活躍するチェックポイント

    powercolorV1パステル和風拡散VAE

    counterfeitV3.0

    vrwaifumix_v20(風景用)

    >CouterfeitV3.0<でトレーニング。

    >おすすめです:

    -->LoRA強度:0.8

    -->Highres fix: Latent

    -->サンプラー DPM++ 2M Karras

    -->CFGスケール:7

    __________________________________________________________________________________

    关于这个LoRA:

    1) 这个LoRA是在动漫系列>紫罗兰花园<的图像上训练的。

    2) 对人物和风景的效果很好。

    检查点,这个LoRA可以真正发挥其作用:

    powercolorV1 + pastel-waifu-diffusion VAE

    counterfeitV3.0

    vrwaifumix_v20 (用于风景画)

    >用>CouterfeitV3.0训练<

    >建议:

    -->LoRA强度: 0.8

    -->Highres fix: Latent

    -->采样器: DPM++ 2M Karras

    -->CFG规模:7

    Description

    A merge of two of my LoRAs.

    Changes:

    -->More versatile in landscape generation.

    -->Slightly more detailed.

    -->Performs better with characters.

    -->Works better with more checkpoints

    ------->Recommended: powercolor, anylora, cetusMix coda, daintyMixHentaiModel

    FAQ

    Comments (5)

    zeppelinofthe
    Author
    May 21, 2023
    CivitAI

    I have no idea how to edit a post afterwards but I forgot to metion that literally every city in this LoRA looks like Venice even though the training data was completely different xD

    peeledkotMay 21, 2023· 2 reactions
    CivitAI

    I thought 1mb meant 1 megabyte

    zeppelinofthe
    Author
    May 21, 2023

    i have corrected the title now

    victorc25744May 21, 2023
    CivitAI

    1MB + 287MB

    zeppelinofthe
    Author
    May 21, 2023

    Correct xD I have no idea but I merged a 1mb LoRA with a 80mb LoRA and the result was 288mb...makes sense right? ^^

    LORA
    SD 1.5

    Details

    Downloads
    979
    Platform
    CivitAI
    Platform Status
    Available
    Created
    5/21/2023
    Updated
    5/11/2026
    Deleted
    -

    Files

    violetmerge.safetensors

    Mirrors

    Available On (1 platform)

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