CivArchive
    Yae Miko | Realistic Genshin LORA - YaeMiko-Mixed
    Preview 173273

    Realistic Yae Miko from Genshin Impact. All samples generate with Ulzzang-6500" embeddings, "vae-ft-mse-840000" VAE.

    For everyone: I am not using the official lora loader, so you should use the additional-network webui plugin to use my lora setting.
    Recommended weights for txt2img - 0.45-0.75, higher weights may cause bad face generate.

    • Update 23/3/1: I believe it's illegal if you made any works/Loras/Embeddings that using actual person‘s face without admission.

    The Standard version working better only 1 lora model, will not missing characteristics, but sometimes will overfitting.

    The Full version use bigger dataset and working better with other face lora model, but missing some of characteristics.

    Thank you for downloading!
    Please leave a image with comment if you like it, I would love to see.

    If you see anyone selling images that generated by this model on pixiv or other platform please message me on twitter.

    (Perfroms better when prompt includes 'yae_sakura' :] )

    Description

    (Recommended weights for txt2img in this version - 0.65-0.8)

    FAQ

    Comments (70)

    StarsetFeb 17, 2023
    CivitAI

    can you make more models of other characters?love it so much!!!!!!!

    floatroarFeb 17, 2023· 1 reaction
    CivitAI

    老哥,我也试着自己炼了一下胡桃的lora但是炼出来后用chil模型跑出来的图里装扮和胡桃关系很小,所以想要问问你是用cos图还是二次元图炼的。

    hi sir,I'm from civital and have some trouble in making a lora for generating pictures.I make Hutao lora in 15 picture s from game and fanwork,but when i use it in Chilloutmix,the clothes and hat is wrong .So I want ask that what type picture do you use when you make this lora.

    tingyueFeb 19, 2023

    肯定cos炼的,如果是二次元脸部和服饰不会这么统一。

    Cronus2436111975Feb 21, 2023

    我也觉得是,二刺猿的图很难保证是同一个面相

    laofang2333687Feb 25, 2023· 1 reaction

    100% coser的图,二次元的不用想了,连鼻子都没有,怎么生成真人图

    kozueMar 19, 2023

    @laofang2333687 说真的这位的几个lora下载给我看懵了?其他的那些将军和神子的二次元lora明明都可以直接生成三次元图片,而且脸更方便调,这位的分辨率和训练数量有点少,我试了下泛用性也不太够……哥们你这鼻子都没有我都不知道咋说……

    duggieFeb 20, 2023
    CivitAI

    These are so amazing!
    Did you use the images from danbooru and other websites for training this LoRA, and how many images did you use?

    Cy1zu_
    Author
    Feb 20, 2023· 2 reactions

    15~50 imgs is enough to train a character lora, and I dont recommand use mirrored imgs to increase your dataset because they will made character features generate in wrong side

    duggieFeb 23, 2023

    @Cy1zu_ Thank you for answering!
    Do you mean like I shouldn't duplicate imgs to increase dataset?

    Cy1zu_
    Author
    Feb 23, 2023· 2 reactions

    @duggie horizontal flipped imgs is what I saying

    duggieFeb 23, 2023

    @Cy1zu_ Okay, thank you for explaining to me!
    If possible, can I ask you one more. If the dataset is too small, maybe about 20 imgs, from your point of view, do you think duplicate imgs can make the dataset better?

    Cy1zu_
    Author
    Feb 23, 2023· 2 reactions

    @duggie I believe the effect will be same as training more epoch, and too many epoch will only cause overfitting

    duggieFeb 23, 2023

    @Cy1zu_ Thank you for answering my question!

    UMR_Feb 21, 2023
    CivitAI

    都用了哪些coser的图呢

    frantisitMar 15, 2023

    显然有河里滑

    a379896403908Feb 21, 2023
    CivitAI

    Who can tell where the korean/japanese/taiwan doll likeness LORA are? I can't find them.

    a379896403908Feb 21, 2023

    @Username010101 Thx thx

    zwtsinFeb 28, 2023

    I can't find them either

    zhezhiiMar 7, 2023

    @zwtsin have been taken down

    124RFQAWFAFeb 22, 2023
    CivitAI

    Ulzzang-6500" embeddings,WHY I cant find it in my UI? I have down it in embeddings

    Nevermore857Apr 28, 2023

    So do I. God DAMN it

    944932079381Feb 24, 2023
    CivitAI

    1

    944932079381Feb 24, 2023
    CivitAI

    biust

    465812220720Feb 27, 2023· 3 reactions
    CivitAI

    1

    xiaobai0852144Feb 28, 2023
    CivitAI

    qiankun521Feb 28, 2023
    CivitAI

    真人那几个doll模型怎么都不见了

    caihuiwenFeb 28, 2023

    都没了,今晚发生了啥

    SOLO1900Mar 1, 2023· 16 reactions

    主模型chillmix跑路了,做LORA的几个肉身在国内也要被抓进去了,还不是跑涩图的惹的事儿

    CoblyuMar 1, 2023

    @SOLO1900 有更多相關資訊嗎?

    FoxAhriMar 2, 2023

    @SOLO1900 没有在网上看到任何相关消息,不知道你是从哪看来的

    2481696199310Mar 12, 2023

    @SOLO1900 不会吧

    SOLO1900Feb 28, 2023
    CivitAI

    蹲一个看看这个活不活得到明天

    caihuiwenFeb 28, 2023
    CivitAI

    never settle !

    hahaeeeMar 1, 2023
    CivitAI

    不懂就问,怎么了?为啥要看能活多久

    zw1998Mar 1, 2023

    因为3个doll已经无了

    zdhvrinfy8811Mar 22, 2023

    哥们我想问一下你们用的哪个chiloutmix是在哪里搞到的

    nickster076Mar 2, 2023
    CivitAI

    One thing i noticed is with the results here is the PNG-info shows different models used from the one here, i guess using a base model is a requirement. Im obviously new to stable-diffusion, but can someone guide me into the right direction? I'm using Automatic1111, do i need to install something like an addon?

    gojiebMar 2, 2023

    that's right, you can look for said models in this site as well, or you can just look it up in a search engine, most of the time it'll lead to a download page

    nanoaiMar 8, 2023

    Install this inside your extensions folder.

    https://github.com/kohya-ss/sd-webui-additional-networks

    Now you can import LoRA into stable diffusion using "Additional Networks" in the GUI

    nickster076Mar 9, 2023

    @nanoai  Thanks !

    ztan060406Mar 11, 2023
    CivitAI

    Where should I place this document?yaeMikoRealistic_yaemikoMixed

    viper360Mar 12, 2023

    paste in to folder <your drive C:,D:,E:>\stable-diffusion-webui\models\Lora

    1531569617355Mar 13, 2023

    @viper360 Then, hou to use

    viper360Mar 13, 2023

    @1531569617355 in txt2img prompt add this message <lora:yaeMikoRealistic_yaemikoMixed:0.5> into your prompt
    recommended weights 0.45-0.75
    *be careful in "_yaemikoMixed" this lora have 3 version choose the one that matches the version you downloaded.

    ztan060406Mar 17, 2023

    @viper360 Thanks!

    Mori_PMar 12, 2023· 13 reactions
    CivitAI

    i think someone is selling your work without you knowing

    this guy on twitter: @soshu00

    and this guy

    https://www.pixiv.net/en/artworks/105869294

    Cy1zu_
    Author
    Mar 15, 2023

    trying to ask pixiv fanbox to require these guys stop selling images generated by this lora.

    cropbehel795Mar 29, 2023

    report him

    anonyvpn001962Mar 13, 2023
    CivitAI

    why many people use SD1.5 instead of 2.0?

    _Kapot_Apr 1, 2023

    There's no porn or nudity in 2.0.

    fuhuaasdf361Mar 18, 2023
    CivitAI

    兄弟,你真厉害,怎么炼的这个呀,你这个和别的lora的兼容性太好了,可以说是现在最好的一个

    songjiawen118Mar 22, 2023· 5 reactions
    CivitAI

    Dear Developers, I am very interested in your project/product/research, I saw your post on the forum, but I would like to know some information in more depth. If possible, I would like to provide my contact information so that you can communicate with me further. My contact information is 【[email protected]】. I would be very grateful if you would like to reply to me. At the same time, I understand the risks of publishing personal contact information publicly, so I also welcome you to reply to me in a private message or otherwise. Thank you again for your time and effort, and I look forward to further exchanges with you.

    mgur45758249Mar 30, 2023· 2 reactions
    CivitAI

    ValueError: Query/Key/Value should all have the same dtype query.dtype: torch.float32 key.dtype : torch.float32 value.dtype: torch.float16

    partho909das106Apr 5, 2023· 7 reactions
    CivitAI

    Anyone please tell me how to install or setup or use Lora model in stable diffusion colab ( not Google notebook)

    jonaalvarado028581Apr 8, 2023· 2 reactions

    just put the model in the stablediffusion/models/lora/ section, and search the lora in the right section on the web-ui

    mastkomApr 10, 2023

    isnt notebook use on colab? lol

    SuddenStriker111May 17, 2023

    i think u can post a link in the code/prompt to point to the lora file, not 100% sure.

    xingbinApr 23, 2023
    CivitAI

    yes

    Bryan0953Apr 26, 2023
    CivitAI

    What Model use? Plz

    UnknownUser943Apr 28, 2023
    CivitAI

    How exactly do you use a Lora on a WebUI, is it with an existing model or is it in itself a model?

    pymess508May 2, 2023· 1 reaction

    If you have lora plugin installed, just throw it into model/lora folder and you are ready to use.

    UnknownUser943May 6, 2023· 1 reaction

    @pymess508 Thank you!

    liuyunzhixingMay 23, 2023
    CivitAI

    第一次看见一个造型不变的模型

    andrew0120Feb 22, 2024
    CivitAI

    This is a personal question.

    How do I make the uploaded model possible to create within civitai?

    I keep trying, but it's hard to find related information, so I asked.

    lion123654Apr 30, 2024· 1 reaction
    CivitAI

    good

    yoskyfreeJul 23, 2024· 1 reaction
    CivitAI

    During the generation process, the face may fail and require face repair or reducing model weights. It may be better to use it in combination with other models

    mursal69Aug 19, 2024· 1 reaction
    CivitAI

    sir what use checkpoint for this ?

    madnessdevilpurg8570May 2, 2025· 1 reaction
    CivitAI

    非常好用的lora,可塑性很高,玩著玩著就沈迷了,包含nsfw也很听话,对于真人cosplay的相性很好,很推荐!!

    LORA
    SD 1.5

    Details

    Downloads
    115,721
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/17/2023
    Updated
    5/11/2026
    Deleted
    -

    Files

    YaeMiko_Test.safetensors

    Mirrors

    HuggingFace (96 mirrors)
    CivitAI (1 mirrors)

    Available On (1 platform)

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