CivArchive
    Taiwan Doll Likeness - v1.0
    NSFW
    Preview 86917
    Preview 86928
    Preview 86927
    Preview 86926
    Preview 86925
    Preview 86924
    Preview 86923
    Preview 86922
    Preview 86921
    Preview 86920
    Preview 86919
    Preview 86918

    Greetings,

    This is 2nd LORA,

    If you're a Asian/Chinese/Taiwanese aesthetics enjoyer, this LORA might be just for you.

    All sample images are generated from ChilloutMix. Please check it out as all the setup process and important steps are listed there on how to produce such images.

    After contacting the maker of Chilloutmix, due to the many parties violating the various warnings and non-commercial usage of Chilloutmix, the author of Chilloutmix have decided to take down all his models from Civitai and has left Civitai. The future of me making/updating more LORAs to be used with Chilloutmix is current UNKNOWN, as I too have felt discouraged from this.
    _______________________________________________________________________________
    Please be responsible for your generations and creations! Do not use them for any malicious intent or harm in any form, and be respectful!

    请小心处理生成的图像,并自行承担风险。禁止对人工智能社区的发展产生不利影响的行为和侵犯个人权利和隐私的使用。

    生成された画像の取り扱いには、自己責任の上、十分注意してください。 AIコミュニティの発展に悪影響を与えるような行動や個人の権利やプライバシーを侵害するような利用は禁止します。

    - It is regretful as the author to say that this LORA is NOT OPEN for ANY commercial purposes to avoid any potential controversy, legal, impersonation, and defamation issues, and it would affect the development of the AI community negatively.

    _______________________________________________________________________________

    Please check out my other LORA as well!
    Korean Doll Likeness
    Japanese Doll Likeness
    ___________________________________________________________

    IMPORTANT
    - THIS IS A LORA SAFETENSORS FILE, NOT A CKPT/SAFETENSORS MODEL FILE, DO NOT USE THIS LORA AS A MODEL AS IT WILL PRODUCE WEIRD RESULTS.

    - PUT THE DOWNLOADED LORA SAFETENSORS FILE UNDER 'STABLE-DIFFUSION-WEBUI > MODELS > LORA FOLDER
    - MAKE SURE TO UPDATE YOUR STABLE DIFFUSION WEBUI TO NEWER VERSIONS, IT HAS A NEW FEATURE THAT LETS YOU ACCESS INSTALLED LORAS EASILY.

    IF THE FACE YOU'VE GENERATED IS NOT GOOD LOOKING, PLEASE USE HIRES FIX, HIGHER RESOLUTION GIVES BETTER RESULTS.


    ___________________________________________________________

    Things to note if you've used my other LORAs: -
    1. It has more variation, and less consistent in making the same types of faces, it is easier to have more facial expressions due to this.
    2. Different LORA weight control would easily change the face variation.

    Explanation:
    1. This is a LORA trained for sd1.5, the trigger words are 'girl' and 'woman', or you could control the weight with LORA taiwanDollLikeness_v10 prompt.

    2. This LORA works extremely well with ChilloutMix and potentially any other photorealistic models, including nsfw models.

    3. Works okay when used together with PureErosFace but you must know how to control the weights properly.
    https://civarchive.com/models/4514/pure-eros-face

    5. May also work together with 'ulzzang-6500' but weight control is a must.
    https://civarchive.com/models/8109/ulzzang-6500-korean-doll-aesthetic

    6. Try to use hires fix with around 10-20 hires steps, latent upscaling (bicubic antialiased), denoising strength between 0,4-0.7, its given the best results by far.

    7. Higher resolution also helps when creating further subjects, hires fix helps greatly with this too.|

    8. Best samplers are DPM++ SDE Karras, Euler a, DPM2 a Karras and DPM++ SDE

    Caution:
    1. Recommended weights for txt2img - 0.5-0.8
    2. Recommended weights for img2img - 0.4-0.8

    Pros:
    - Works greatly with pretty much any photorealistic models, including nsfw
    - Able to do img2img with anime pictures as a reference. (Recommended weight - 0.4-0.8)
    - Easier to generate face variations with the weight control
    - Easier to have teeth when smiling

    Cons:
    - Difficult to make consistent/ replicate faces, due to the sample of images when training.
    - Struggles to create side view faces.
    - May give bad results when using low resolution, all sample images are 512x768, and hires fixed with a scale of 1.75.

    Thank you for downloading!
    Please leave a review or comment, you can upload images with the review.

    Please give this a try while I'll make more LORA in the near future!

    _______________________________________________________________________________
    Due to using Dreamlike Diffusion 1.0, this model has the following license:

    License

    This model is licesed under a modified CreativeML OpenRAIL-M license.

    - You can't host or use the model or its derivatives on websites/apps/etc., from which you earn, will earn, or plan to earn revenue or donations. If you want to, please email us at [email protected]

    - You are free to host the model card and files (Without any actual inference or finetuning) on both commercial and non-commercial websites/apps/etc. Please state the full model name (Dreamlike Diffusion 1.0) and include a link to the model card (https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0)

    - You are free to host the model or its derivatives on completely non-commercial websites/apps/etc (Meaning you are not getting ANY revenue or donations). Please state the full model name (Dreamlike Diffusion 1.0) and include a link to the model card (https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0)

    - You are free to use the outputs of the model or the outputs of the model's derivatives for commercial purposes in teams of 10 or less
    - It is regretful as the author to say that this LORA is NOT OPEN for ANY commercial purposes to avoid any potential controversy, legal, impersonation, and defamation issues, and it would affect the development of the AI community negatively.

    - You can't use the model to deliberately produce nor share illegal or harmful outputs or content

    - The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license

    - You may re-distribute the weights. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the modified CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully) Please read the full license here: https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0/blob/main/LICENSE.md

    Description

    FAQ

    Comments (20)

    ming1Feb 10, 2023
    CivitAI

    Couldn't find Lora with name TaiwanDollLikeness

    yomanusernameFeb 10, 2023

    You're looking at it.

    aeternusFeb 10, 2023

    when you're downloading it, it come with other name than provided in those prompts. Just copy the name and replace it in <lora:.........> and it will be ok

    ming1Feb 10, 2023· 1 reaction

    thaknks all,i fix it

    ming1Feb 10, 2023· 1 reaction

    @aeternus thanks!

    Flower2788Feb 10, 2023

    @aeternus 
    "taiwanDollLikeness_v10.safetensors" rename to "taiwanDollLikeness.safetensors" ?

    thank fixed.

    138557Feb 10, 2023· 2 reactions

    @Flower2788 You may rename it, the safetensor file name must match with the LORA prompt

    KorialstraszFeb 10, 2023· 1 reaction
    CivitAI

    I struggle to get a face that is not completely washed up by beauty filters. Do you have any advice?

    138557Feb 10, 2023

    May I ask what model are u using with the LORA?

    KorialstraszFeb 10, 2023

    @Kbr sure! Chilloutmix, as advised

    138557Feb 10, 2023

    @Korialstrasz Make sure you're also using the sd vae from https://huggingface.co/stabilityai/sd-vae-ft-mse-original/tree/main
    Download this >vae-ft-mse-840000-ema-pruned.safetensors

    Go to settings, Stable Diffusion, vae and change to it, and click apply settings

    KorialstraszFeb 10, 2023· 1 reaction

    @Kbr tymv!

    138557Feb 11, 2023

    @Korialstrasz Do leave a review if it gave you good results!

    mistletoe9Feb 11, 2023· 2 reactions
    CivitAI

    Very nice. Even works well with western faces, when the strength is lowered. Between the Korean and the Taiwan one, the Korean one gives a more softer look while the Taiwan gives a more western/aggressive look

    Lrain233Feb 12, 2023· 3 reactions
    CivitAI

    it`s beautiful,but why can`t i get the same result?

    MLYEFeb 14, 2023
    CivitAI

    How to start this? I really dont get it

    verXDFeb 15, 2023· 1 reaction
    CivitAI

    Thank you so much!!! This is the best present for Asian lovers.

    We rarely have actually good Asian faces before. And even after the amazing ChilloutMix came to life, we still mostly have Korean faces only.
    Wanted those pretty Instagram Taiwanese faces for a long time

    This is like having your own Taiwanese Instagram babes to take photos with. Thanks for bringing a dream come true!

    Hope to see more from you may be more Taiwanese variants or Japanese version! :)

    TokisakiKFeb 21, 2023
    CivitAI

    she like vivian

    HanYunxinFeb 26, 2023
    CivitAI

    why my picture,looking so weird?I use chilloutimx and 840000vae

    NiConfuciusFeb 26, 2023
    CivitAI

    I found that each generated image has a necklace, even if the HD is fixed, it doesn't work, I also added ((jewelry, accessory, necklace)) in the Negative prompt, but it still doesn't work, can the author solve this problem in the next version?

    LORA
    SD 1.5

    Details

    Downloads
    45,432
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/9/2023
    Updated
    5/14/2026
    Deleted
    -

    Files

    TaiwanDollLikeness.safetensors

    Mirrors

    HuggingFace (184 mirrors)
    TensorFiles (1 mirrors)
    CivitasBay (1 mirrors)

    Available On (1 platform)

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