CivArchive
    tsubasu izumi art style(和泉つばす画风) | Anima - v1.0 lora
    NSFW
    Preview 132464616
    Preview 132464848
    Preview 132464895
    Preview 132464959
    Preview 132465017
    Preview 132465083
    Preview 132465118
    Preview 132465230
    Preview 132465296
    Preview 132465384
    Preview 132465603
    Preview 132465672
    Preview 132465786
    Preview 132465857
    Preview 132466016
    Preview 132466094
    Preview 132466184
    Preview 132466266
    Preview 132466343
    Preview 132466432

    这是和泉つばす的画风anima版。

    This is tsubasu izumi art style anima ver.

    一些说明:我重新上传了预览图,因为我发现WebUI Forge Neo之前的版本没有正确加载LoKr……实际表现比我之前想象的好得多。

    A quick note: I reuploaded the preview images because I realized that previous versions of WebUI Forge Neo weren’t loading LoKr properly… The actual results are much better than I originally thought.

    一些Anima的小提示:使用tag时保留", "之后的空格,同时不要使用"BREAK"和下划线"_"。

    A few tips for Anima: when using tags, keep the space after ", ", and don’t use BREAK and underscore "_".

    这是一个提醒:你需要关闭R、X和XXX可见才能看到我的部分LoRA。

    This is a tip: you need to disable R, X and XXX visible to see some of my LoRA.

    欢迎来我的Q群玩:962631038(新2群,原备用群)/514186900(人已满),答案是我炼过模型的一个角色的名字,她的英文名在群名中。

    My Discord server: https://discord.com/invite/gVtk8XYWwt ←我的Discord服务器。

    在爱发电为我买一份一美元炸鸡:https://afdian.com/a/YDDLJW

    Tip me a one-dollar fried chicken: https://patreon.com/YDDLJW

    ————————————

    V1.1:

    搭配其它角色LoRA时拥有更好表现的Lokr。

    Lokr with better performance when using with other character LoRA.

    V1.0:

    当你想要还原某些角色时拥有更好表现的LoRA。

    LoRA with better performance when you want to recreate certain game characters

    ————————————

    触发词 (trigger word)

    @tsubasu izumi, 

    ————————————

    推荐使用底模(Recommended base model)

    WAI-ANIMA

    https://civarchive.com/models/2544636/wai-anima

    Anima Base

    https://civarchive.com/models/2458426/anima

    AnimaYume

    https://civarchive.com/models/2385278/animayume

    ————————————

    推荐参数(Recommended setting):

    Sampler: Euler a

    Schedule Type: Flux2

    Steps: 32 (28~40 is fine)

    width: 768~1536

    height: 768~1536

    CFG Scale: 4 (3~6 is fine)

    Description

    Trained on Anima base-v1.0 without regularization.

    FAQ

    LoCon
    Anima

    Details

    Downloads
    316
    Platform
    CivitAI
    Platform Status
    Available
    Created
    6/1/2026
    Updated
    6/29/2026
    Deleted
    -
    Trigger Words:
    @tsubasu izumi,

    Files

    tsubasu_izumi_art_1a_anima-000048.safetensors