Find what you need for your workflow.

Sae Hinata - Ijirare: Fukushuu Saimin

sae hinata, blue eyes, purple hair, dark skin, dark-skinned female,

[Illustrious] 浠Mizuki - vtuber (5 outfits)

For PixAI users : Don't Import this , instead use this -> This model was trained by azkiisverycute. If you see this model reprinted at PixAI other than the address above, please let azkiisverycute know.

Kagami Tsurugi, Ladybug, Miraculous. Different models and attires.

🔴 Hi everyone, I wanted to take a moment to talk about the future of my work and this channel. Dedicating so many hours to creating and refining LoRas is becoming increasingly difficult to sustain every single day....

Nanahime Style

If the original artist requests a takedown, I will immediately remove this model off of CivitAI. Thank you. Please support their work here. Meanwhile, please do not monetize your generations.

Apone3D - Style - Illustrious XL / Anima

3d and realistic helps to achieve the style, despite it not being trained in the captions

Jeanne - Mistress Kanan Is Devilishly Easy / ジャンヌ , Jannu / カナン様はあくまでチョロい , Kanan-sama wa Akumade Choroi

Wan2.2-Fun-Camera-LowNoise-14B-Q5_K_S.gguf

diffusion_modelWan 2.2

Otherxlr8harder/talkie-1930-13b-yarn-32k-tf

ModelHugging Face

Talkie 1930 13b Yarn32k Tf GGUF

This repo contains the GGUF and quantized version of talkie-1930-13b-yarn-32k-tf It is based off of the base version of Talkie-1930, and therefore is not meant for actual chat conversations.

Female Miqo'te (Final Fantasy 14 | FFXIV) | LoCon

I have opened up a request form

Ibuki Mioda (澪田唯吹) - Danganronpa (ダンガンロンパ)

ibuki mioda, black hair, blue hair, hair horns, ear piercing, lip piercing, long hair, multicolored hair, (pink eyes:1.5), pink hair, white hair,

ModelExternal source

flux2-vae.safetensors

Apollo Intensa Emozione

TRAINED ON LOCON KOHYA SS GUI

anima qingwei style

更新:这次是从base开了一个新分支,使用的是懂得都懂的训练集和懂得都懂的美学集,反正能画懂的都懂的东西,特长和v1.0不一样,应该是封面图那种和可以联想到的那种,但是我自己不喜欢和civitai审核斗智斗勇所以发的都清水图,你们自己开发吧。更新:前面一个版本的结果使我能够确信我实验的方法论有一个好的结果所以在这个正式版本,我认为我可以使用我那个2.5w的图数量的正式训练集达到一个不错画风微调这个版本使用2.5w的非生成数据训练,并且依旧沿用上个模型的1200图的美学参与比较anima base我觉得拥有一些我训练后的更偏向我喜好的画风质感,并且手指和脚趾比原版好稍微那么一丢丢,至少我通过xy图表看是这么一个效果,而且其实我自己测试下来,并没有失去对base model的Lora的兼容,至少我训练的Lora通常rank比较低,依然是有效的,而且画出来还行旧版本:微微调的anima base,改善了一些美学,使用了ai生成数据,我人工筛选了所有的图片以确保训练集的质量,都是我生成造出来的图片并且,我把自己的画风,训练过一个tag了,可以使用@qingwei画师串仅训练dit的方法下训练的,冻结一些模块,不会出现如同融模一样的anima本体的知识被损失的问题,仅仅是如我说的那样微微调了美学

cursive script (行草/書法/中文/東亞文字)

The model trained with calligraphic pictures (mainly cursive)

diffusion_modelWan 2.1 I2V 14B

ModelExternal source

Wan2.1-I2V-14B-480p-Q6_K.gguf