This LoRA is not official or authorized.

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

Activation token is TogaHimiko

Her outfit wasn't the main focus with this LoCon so the dataset doesn't have a ton of it in it. that being said, you can get the outfit with the tags:

miniskirt, yellow cardigan, long sleeves, sailor collar, red neckerchief, black kneehighs

her features are baked into the activation tag, but her hair color might be a bit weak, so you might have to tag blonde hair

2023/4/17 update

Assign different weights to each tag for training. Adjust linear dim to 64. The preview images of the new model are generated using 'Hires Fix' and upscaled with 'SD Upscale'.

Introduction

A Lycoris model trained with ~180 images by LAM. The trigger tag is ‘lam-style’. The 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) are recommended. 0.6 ~ 0.8 weights are good. The preview images are generated using 1.6 version (original) YesMix.

Cautions (Must Read)

• Some preview images are generated using hires fix. Thus, if you wanna generate the preview images, pls download the original preview images and analyze them using 'PNG INFO' in webui to get all generating information.

• The maximum of generating resolution should not exceed 1024 theoretically, otherwise extra/deformed head/limbs will be waiting for you :). Try to use 'hires fix' to prevent this problem and help to generate higher-resolution images.

• For 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) users, make sure add '--no-half-vae' in command line of webui to prevent generating black images. See voldy's log for the details of edit method. For high-performance GPU, make sure 'medvram' and 'lowvram' mode are disabled.

An extraction of AOMv2_hard using the scripts included in sd-scripts and LyCORIS.

Since I knew the base model, it was easy to get the desired output without needing to extract with multiple base models

the dim sizes are 16 and conv dim 16 as that is what I extracted at. Very useful for adding a bit of AOMv2_hard to any other model you want to use

Is obviously very good at generating NSFW outputs, but it works very well for SFW as well.

This LoRA is not official or authorized.

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

这只是一个性暗示手势。使用时可以加点表情tag，例如"blush,evil smile,tongue out,licking lips"之类的

It's just a sexually suggestive gesture. When using it, you can add some emoticon tags, such as "blush, evil smile, tongue out, licking lips" and the like

成功率并不算非常高（所有的手部姿势成功率都不会多高，这是事实，除非过拟合），而且测试中有的模型手部成功率会更低，你要多一些耐心

The success rate is not very high (the success rate of all hand poses is not very high, this is a fact, unless overfitting), and some models in the test have a lower success rate, you need to be more patient

Weight 1

I check Twitter every day, so I will respond when I receive a DM.
ツイッターは毎日チェックしているのでDMが来たら対応致します。

This is a LoHa LoRA! You need the extension!
Why? LoHa have virtually no effect in output art style and they do a lot better with multi-concepts.
I want to do a v2, but not sure when! There's a few things I'd like to fix, but it works well enough.

Trigger words for V1.20:

<lora:AnalToysv120:0.8>(anal-tail, ass focus, from behind)

You can try "anal-tail", "round-buttplug", "heart-buttplug" and "anal-beads". You can combine some tail types along with "anal-tail" like "anal-tail, cat tail" or "anal-tail, wolf tail" and it works. Its limited but it works.
You should probably prompt for scenes where the ass is visible.

If you want to use it for male characters, go for a very low LoRA weight like 0.3-0.4! I did some very, very limited testing for it since its not my thing.

Follow the steps in the Model Version notes to reproduce my images.

Sample images generated from Hassaku.

All negative embeds should be easy to find. and Prompts should be included in the sample images! Click the little icon in the bottom right to see it.

You might want to add some negatives, for example:

It was trained on 400 512x512 images scraped from the internet and 13 epochs using animefull_final.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger and fine tuned for specific tags.

LoHa of Mejiro Mejiro Bright (umamusume). Please install extension on WebUI.

After installing extension, put LoHa file in your sd/models/lora folder and write LoRA notation in prompt to applying LoHa. <lora:loha_name:weight>

All uploaded images are generated by AOM3A1

Recommended options

• LoRA weight 0.7 ~ 0.8

• Trigger words

  • Base

    • mejiro bright \(umamusume\)

  • Racing competition costume :

    • hair bow, collared shirt, white shirt, puffy short sleeves, center frills, frilled dress, green dress, striped, belt, green skirt, frilled skirt, black thighhighs, mismatched footwear, black footwear, white footwear, asymmetrical footwear, knee boots

  • Casual :

    • hair bow, casual, frills, green dress, puffy short sleeves

  • Tracen summer uniform :

    • hair ornament, red scrunchie, tracen school uniform, summer uniform, serafuku, puffy short sleeves, purple bowtie, horseshoe ornament, sailor collar, sailor shirt, purple shirt, white skirt, pleated skirt, frilled skirt, miniskirt, zettai ryouiki, white thighhighs, brown footwear, loafers

  • Tracen winter uniform :

    • tracen school uniform, winter uniform, puffy long sleeves, sailor collar, sailor shirt, purple shirt, white bowtie, horseshoe ornament, purple skirt, frills, brown footwear, loafers

Use a model derived from or mixed with animefull model.

• Settings

  • Stable Diffusion WebUI by Automatic1111

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • I recommend kl-f8-anime2.vae

  • CLIP skip = 1

  • Use Hires.fix to get higher quality image

    • Upscaler

      • Latent, Latent (nearest-exact), Latent (bicubic antialiased) or other Latent series.

    • Denoising strength 0.50~0.65

LoRA training info

• trained on sd-scripts by kohya_ss.

  • LyCORIS by KohakuBlueleaf

  • Thank you all a lot!

• based on Animefull-pruned model

• 110 images consist of

  • 4x low quality 38 images

  • 8x medium quality 42 images

  • 16x high quality 30 images

  • => 1 epoch = 968 images

• replaced character feature tags with mejiro bright \(umamusume\)

  • 1girl, brown hair, brown eyes, wavy hair, ahoge, side braids, twin braids, horse girl, grey eyes, horse ears, solo, side braid, orange hair, very long hair, hair between eyes, yellow eyes, long hair, animal ears, horse tail, bangs, multicolored hair, green eyes, purple eyes, red eyes, tail

  • I think these tags are the essentials of Mejiro Bright or

  • unnecessary for training because of duplication.

• resolution : 768x768 with aspect ratio bucketing.

• other training settings is included in training_data.zip or metadata of LoRA file.

And, special thanks to my best friend. DH.

LyCORIS of Odette Annable

LyCORIS of Natascha McElhone

This is a model that generates airscape which trained by 100 img.

weight 1.0 maybe generates better img,trigger is cityx.

Recommend prompts such as scenery, no humans, cityscape and so on

It's a experimental work about dataset,lol

LyCORIS of Olga Bołądź

LyCORIS of Ruby Rose

LyCORIS of Snoop Dogg

LyCORIS of my V from Cyberpunk 2077

Ib from the RPG Maker horror game Ib. Really liked her when the game was still popular.

Recommended weight at 0.7

For her usual attire, use prompt: ib, shirt, skirt, ascot

导言

这是一个《Umamusume Pretty Deby》《闪耀优俊少女》的目白阿尔丹角色LyCORIS，其与其他同角色lora的区别在于，添加了新的Neige Émeraude卡面立绘。在使用中使用cut off插件（https://github.com/hnmr293/sd-webui-cutoff）防止元素污染会获得更好的效果。

示例图使用了Viewermix模型进行生成。

使用方法

建议以0.75左右的权重使用。

在打标过程中，为隔离元素污染，采用了每个卡面分开设关键词的方法，以下是各个立绘的关键词：

私服：mejiro_ardan_CA

Neige Émeraude：mejiro_ardan_NE

夜風に舞うは祈りの花：mejiro_ardan_PFN

特雷森制服：mejiro_ardan_UN

胜负服：mejiro_ardan_SF

其他非特殊立绘：mejiro_ardan

训练细节

以NAI Final作为底模训练，使用了约500张图片作为训练集，将约一半图片作为回归训练集，每张图片重复训练10次，训练10个epoch，最终步数为5150步。

以下为主要设置参数：

network_dim=32

network_alpha=32

keep_tokens=8

conv_dim=4

conv_alpha=4

lr="1e-4"

unet_lr="1e-4"

text_encoder_lr="1e-5"

batch_size = 2

局限性

1. 人物立绘中的不对称设计的还原难度非常大，基本上只能依靠随机抽卡和手动辅助达到一定的还原效果。

2. 在出示例图的过程中没注意到Additional Network插件失效，一些LyCORIS的层没有输出，可能对最终结果有所影响。

Introduction

This is a LyCORIS of Mejiro Ardan from "Umamusum Pretty Deby". The difference between other similar characters Lora is the addition of Neige Émeraude. Using the cut off plugin（ https://github.com/hnmr293/sd-webui-cutoff ）to prevent elemental pollution will achieve better results.

Example images are generated by Viewermix model.

Usage

Suggest using a weight of around 0.75.

During the marking process, in order to isolate element contamination, a method of setting keywords separately for each card surface was adopted. The following are the keywords for each vertical drawing:

Casual clothes: mejiro_ ardan_ CA

Neige Émeraude：mejiro_ ardan_ NE

夜風に舞うは祈りの花: mejiro_ ardan_ PFN

Teresan uniform: mejiro_ ardan_ UN

Shobofuku：mejiro_ardan_SF

others：mejiro_ardan

Training Details

Using NAI Final as the base model for training, approximately 500 images were used as the training set, and about half of the images were used as the regression training set. Each image was repeated 10 times for 10 epochs, with a final step count of 5150.

The following are the main settings parameters:

network_ dim=32

network_ alpha=32

keep_ tokens=8

conv_ dim=4

conv_ alpha=4

lr="1e-4"

unet_ lr="1e-4"

text_ encoder_ lr="1e-5"

batch_ size = 2

Limitations

1. It is very hard to draw the asymmetrical designs in character portraits. Maybe a lot of tries and manual assistance is a certain way to achieve good results.

2. During the process of presenting the example image, I did not notice that the Additional Network plugin was invalid, and some LyCORIS layers did not output, which may have an impact on the final result.

简介

本模型希望尽可能还原星野 アイ眼睛细节，需要较大的图幅获得相对好的效果，手部表现和相对欠缺。

Introduction

This model aims to restore the eye details as much as possible, requiring a larger image size to achieve good results. However, the hand performance are relatively lacking.

A very healthy model

• 欢迎加入AI绘画模型博物馆QQ群

• ①群：791222249

• ②群：599962550

This LORA is Hacka Doll No.3 from Hacka Doll.

Hacka Doll No.3 is one of the three main "Hacka Dolls" computer programs that are used to advance their users.

inital tags:

"hacka doll 3, very long hair, blue hair, purple eyes", (with hair ornament, headphone, otoko no ko)

main costume:

"bare shoulders, white dress, strap slip, black camisole, bike shorts, white glove, boots"

"wedding dress, garter straps, bridal veil, elbow glove"

"school uniform, pleated skirt"

As an intelligent being, you will immediately recognize that he is a boy at a glance.

Therefore, the use of the " penis in shorts (or penis in panties)" tag is ideal.

You may want to use 7thLayerAbyss or AOM3A.

weight: 0.8-

option: jacket, head-mounted display, thigh strap, sandals(low quality), energy sword(low quality)

We will comply with the ANYCOLOR secondary creation guidelines.
https://www.anycolor.co.jp/guidelines/

It seems that no particular intention is indicated. I pledge to create and publish works that respect the spirit of ANYCOLOR content.

特に個別の意向が示されていないと思われます。ANYCOLORコンテンツの精神を尊重した作品を制作、公表していく事を誓います。

This is a LoHA clothing style for wet clothes. It will modify what your subject(s) are wearing with a variety of different wet see-through clothing options. This LoHA is capable of generating wet shirts, tied shirts, crop tops, gym uniforms, serafukus, and school uniforms.

To prompt this clothing, enter "<lora:WetClothesv1:strength>wecl, see-through,". Wet clothing tags include "shirt, tied shirt, crop top, gym uniform, serafuku, school uniform,". This clothing is not limited to shirts, and with extra prompting, works on untrained clothes. This LoHA is also trained on supporting tags to give the subject a wetter or more customized look; these tags include include "bra visible through clothes, bikini under clothes, wet, wet hair, rain, off shoulder, single bare shoulder".

This LORA is best at 0.7 strength in my opinion. It was trained on 55 various images of wet clothes on AnythingElsev4.5. This is a sfw model, but it can generate nsfw if prompted to.