CivArchive
    Texture Enhancer/chosen_style/质感增强 - anima V1
    NSFW
    Preview 138612159
    Preview 138612375
    Preview 138612164
    Preview 138612120
    Preview 138612409
    Preview 138612146
    Preview 138612150
    Preview 138612158
    Preview 138612163
    Preview 138612169
    Preview 138612165
    Preview 138612173
    Preview 138612179
    Preview 138612181
    Preview 138612188
    Preview 138612190
    Preview 138612257
    Preview 138612372
    Preview 138612403
    Preview 138612459

    anima版本说明:

    大家好,我是chosen鲷。这个系列模型致力于将nai4.5的图片风格训练成lora,这样就让大家能在anima模型上复现这类风格,所以做了这个,喜欢这种风格的话可以下载一个玩玩。

    anima V1致力于描绘肌肤的柔软触感,以及面部的红晕和眼睛的细节,在nsfw方面表现非常好,对整体人物画风的提升也非常大。

    触发词:chosen_style_3

    建议增加质量tag:masterpiece, score 7, highres, absurdres, best quality, explicit, chosen style 3, sfw, safe, @akizero1510, @ye_jji,

    大模型推荐用我miaomiao anima 1.4(下方推荐),或者其他的anima模型也可以,不过强烈建议使用miaomiao1.4的模型以达到最佳效果。

    nsfw内容也是可以的,具体的tag可以看我例图,值得注意的是,例图中的cfg和step信息是不正确的,我常用的是cfg1 step18(有加速lora的情况下),cfg4 step50(无加速lora的情况下),采样器是eular simple

    感谢大家的下载使用,如果喜欢的话欢迎关注我,我的bilibili pixiv ID均为chosen鲷,欢迎大家关注我的频道https://space.bilibili.com/3867754?spm_id_from=333.337.0.0 https://www.pixiv.net/users/102907778 谢谢。

    IL(光辉)版本说明:

    大家好,我是chosen鲷。

    这个是我最近突发奇想,我想要把我常用的nai4风格训练成lora,这样我就能够在sd上面复现出来这类风格,所以做了这个,感兴趣的话可以下一个玩玩。

    请注意,这里的两个模型并不是两个版本,而是两个不同风格的模型,chosen_style_1注重于哑光质感的图片风格,而chosen_style_2专注于描绘肌肤的闪亮质感以及柔软的感觉,这两个模型是不同的,大家根据喜好自己选择,具体的风格可以看我例图

    触发词:chosen_style_1/chosen_style_2

    大模型推荐用我的chosen-mix_xl(下方推荐),其他的常见光辉系模型也都可以。

    R18内容也是可以的,具体的tag可以看我例图。

    感谢大家的下载使用,如果喜欢的话欢迎关注我,我的bilibili pixiv ID均为chosen鲷,欢迎大家关注我的频道https://space.bilibili.com/3867754?spm_id_from=333.337.0.0 https://www.pixiv.net/users/102907778 谢谢。

    The following content is from machine translation

    Anima Version Description:

    Hello everyone, I'm chosen鲷 (Chosen Tiao). This series of models aims to convert the image style of NAI 4.5 into a LoRA, so that everyone can reproduce this style on the Anima model. That's why I made this — if you like this style, feel free to download it and have a try.

    Anima V1 focuses on depicting the soft texture of the skin, as well as facial blush and eye details. It performs exceptionally well in NSFW content and significantly improves the overall character art style.

    Trigger word: chosen_style_3

    Recommended quality tags to add:

    masterpiece, score 7, highres, absurdres, best quality, explicit, chosen style 3, sfw, safe, @akizero1510, @ye_jji

    For the base model, I recommend using my miaomiao anima 1.4 (see recommendation below), or other Anima models can also be used. However, I strongly suggest using the miaomiao 1.4 model to achieve the best results.

    NSFW content is also supported. For specific tags, you can refer to my example images.It is worth noting that the CFG and step information in the example images is incorrect. What I commonly use are: CFG 1 with step 18 (when using an acceleration LoRA), and CFG 4 with step 50 (without an acceleration LoRA). The sampler is Euler simple.

    Thank you for downloading and using. If you like it, feel free to follow me. My Bilibili and Pixiv IDs are both chosen鲷. Welcome to follow my channels:

    https://space.bilibili.com/3867754?spm_id_from=333.337.0.0

    https://www.pixiv.net/users/102907778

    Thank you.

    IL Version Description:

    Hello everyone, I'm chosen鲷 (chosen Diao).

    Recently, I had a sudden idea—I wanted to train the nai4 style I often use into a LoRA, so that I could reproduce this style on SD. That's why I made this. If you're interested, feel free to download and give it a try.

    Please note that the two models here are not two versions, but two different style models. chosen_style_1 focuses on a matte texture image style, while chosen_style_2 is dedicated to depicting shiny, sparkling skin textures and a soft feel. These two models are distinct—choose according to your preference. You can refer to my example images for the specific styles.

    Trigger words: chosen_style_1 / chosen_style_2

    For the base model, I recommend using my chosen-mix_xl (see the recommendation below). Other common "Glory" series models also work well.

    R18 content is also supported. For specific tags, please check my example images.

    Thank you for downloading and using my models. If you like them, feel free to follow me. My Bilibili and Pixiv IDs are both chosen鲷. Welcome to my channels:

    Thanks!

    Description

    FAQ

    Comments (6)

    a24716574Aug 4, 2026
    CivitAI

    是因為我用neo生圖的關係嗎,相同的tag和尺寸出來的效果差異很大(沒用hires fix)

    chosen
    Author
    Aug 4, 2026

    大模型用的那个?cfg采样器等是否一致

    a24716574Aug 4, 2026

    @chosen 也是用miaomiao 1.4,CFG推薦的5.5以及例圖的7都試過,但采樣器的話例圖看不到XD,我嘗試過Euler、Euler A、ER SDE、RES4LYF res系列,同seed的情況下看起來都跟原版不太一樣
    最大的差異,我覺得例圖看起來比較"乾淨"
    順便問一下大佬有考慮更新mature body的lora嗎,我在光輝上都是用這個~

    chosen
    Author
    Aug 5, 2026

    @a24716574 mature body那个是必然要更新的,等过几天就炼,我现在还在摸索阶段,我想等我积累足够的经验之后再训练他.至于neo的话,我今天去neo试试好了,我很多图都是挂了加速的lora,用的0.2版本,然后采样器一般是用Euler+simple,我cfg从来没用到过7啊,是不是图片信息出错了,一般用加速lora就设置成1,不用加速lora就设置成4

    a24716574Aug 5, 2026

    @chosen 破案了!是CFG的關係,改成1就可以了,不知道為啥例圖上顯示的是7哈哈

    期待您的mature body更新~

    chosen
    Author
    Aug 5, 2026

    @a24716574 实在不好意思,这个是我的原因,因为我用了weilin工具箱插件写tag,导致出的图comfy读取不了,所以我只能又借助了一个插件去把图片信息保存到图上,但是他无法正确的保存cfg 和step信息,所以就乌龙了。不过多亏你提醒了我,我以后就可以设置一下让他显示cfg1了

    LORA
    Anima

    Details

    Downloads
    1,091
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/3/2026
    Updated
    8/25/2026
    Deleted
    -
    Trigger Words:
    chosen style 3

    Files

    chosen style 3-000040.safetensors

    Mirrors