2023/4/15 update

Update dataset and upload lycoris version. Adjust trigger tag to 'soyincapazz', username of this artist. The preview images of the new model are generated using 'Hires Fix' and upscaled with 'SD Upscale'.

2023/4/1 upate

Fix the compatibility problem of non-NAI-based checkpoints. Now arbitrary anime model with NAI's VAE or kl-f8-anime2 VAE can also generate good results using this LoRA, theoretically.

2023/3/1 update

Reduce loss to 0.09. 0.6 and 0.7 weight should be enough.

Introduction

A LORA model of Tsuki no i-min style. The 784MB VAEs are recommended. 0.6~0.8 weights are good.

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.

Wilykit - Thundercats 2011

This is a Lora based off of (you guessed it), Wilykit from the Thundercats 2011 reboot. This is my first character Lora, so I figured I'd start off with something obscure without a lot of source material to go off of. Great plan, I know.

Works just fine at 0.8 strength.

There are a few quirks with the Lora.

• Clothing likes to stay pretty close to her color theme and needs some weights added to the color if you want to change it and a negative of what it's generating.

• Styles also tend to vary from seed to seed, mostly from style bleed from the source images. It's a mix of fan art and official art, so sometimes those can show themselves a bit.

• The tail showing up is luck of the draw if unprompted. If that's important to you, add it somewhere into the prompt to make it show up a bit better.

• The face markings and countershading are a similar deal. Luck of the draw. Prompting for "facial mark" might help it to show up better, but maybe not.

I trained it using the Anylora model, and it seems to work fine in different models, but if people are into this Lora, I might retrain it using NAI instead. As I'll be using that as the base to train from now on.

As with my other Lora, it's created using a merge of 2 separately trained Lora. One on the face and one on the body.

A Character LoCon for W | Arknights (明日方舟)

萨卡兹佣兵，W，真是好久不见，虽然这个“好久”可能含义不同。但没关系，既然现在的你不是我要找的你，那你暂时是安全的，暂时哦。

• Recommeneded prompts:

  • Default outfit: black gloves, black scarf, blouse, breasts, fingerless gloves,jacket, red eyes, scarf, skirt, open clothes, black jacket, open jacket, black skirt, short hair, black pantyhose, white shirt, pantyhose, bangs,miniskirt

• Trigger word: red eyes/yellow eyes/orange eyes, antennae

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

• Some sample figures were generated with my Vivid Impactful Style - LoRA

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

Weight : 0.4～0.6? (Weight of the sample image is about 5.5)

Recommended hair length prompt "(pixie cut: 1.2) or (pixie cut: 1.4)"

In v5.0, the quality of the generated pictures is less likely to deteriorate when changing to costumes other than school uniforms.

The model for the illustration for the sample is based on 7th_anime_v3_A and has been merged several times, including AOM3.

7th_anime_v3_A will be a close illustration.

Please use it in an environment where LoCon is available.

AnyLoRA is used as the model for the study.

LoHa of Yamanin Zephyr (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

    • yamanin zephyr \(umamusume\)

  • Racing competition costume :

    • hair ornament, red scrunchie, choker, necklace, blue gloves, detached sleeves, long sleeves, bare shoulders, blue sleeves, white dress, strapless dress, frilled dress, flower, frills, blue thighhighs, black footwear

  • Casual :

    • casual, hair ornament, red scrunchie, white shirt, brown bow, collared shirt, puffy long sleeves, open clothes, open jacket, brown jacket, layered sleeves, center frills, short over long sleeves, dress shirt, belt buckle, brown belt, short shorts, blue shorts

  • 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

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

• 68 images consist of

  • 4x low quality 22 images

  • 8x medium quality 19 images

  • 16x high quality 27 images

  • => 1 epoch = 672 images

• replaced character feature tags with yamanin zephyr \(umamusume\)

  • 1girl, grey hair, ear ornament, multicolored hair, simple background, solo, streaked hair, white hair, brown hair, black hair, purple hair, orange eyes, animal ears, two-tone hair, horse girl, red eyes, hair scrunchie, horse tail, very long hair, horse ears, bangs, yellow eyes, brown eyes, long hair, hair between eyes, low twintails, hair bobbles

  • I think these tags are the essentials of Yamanin Zephyr 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.

Onko Akino / 温故 昭乃 / アキノ from wixoss Series.

来自wixoss系列的昭乃。

***LOCON file***

I also use the LoRA Block Weight, picture above shows the results of different parameters. If you don't want to install the plugin, you can just read the 'ALL' part.

我还使用了 LoRA Block Weight 插件，上图表示出不同参数下的结果。如果不打算安装这个插件，只看“ALL”的部分即可。

Use Hires.fix please.

请使用Hires.fix。

Triger: onko_akino, akino_diva

触发词：onko_akino, akino_diva

problems/已知问题

Not a good locon lora. I do not know how that I can make it better. Anyway, it is here.

作为locon质量不好。我没有优化的思路了。总之先发布了。

The model is trained at clip skip 1 on ACertainty

You can however use the extracted locon on a lot of different models as I show in the examples. Otherwise you can perform (block) merge and (block) extraction yourself using the full model.

Other versions can be found on the associated hugging face repository https://huggingface.co/alea31415/oshinoko

Screenshots and fan arts are respectively tagged with aniscreen and fanart. Using aniscreen gives both stronger anime style and better resemblance. Using aniscreen on style models may cause weird effects.

You may want to use ; to separate characters (see the prompts of the example images)

Characters

For now, we have

• HoshinoAi (add hair ornament if you want)

• AmemiyaGoro

• TendojiSarina

• Aquamarine baby

• Aquamarine child

• Aquamarine

• Ruby baby

• Ruby child

• Ruby (add red ribbon if you want)

• ArimaKana child

• ArimaKana (I suppose you should add some hat)

• SaitoIchigo

• Miyako

• GotandaTaishi

• KurokawaAkane

According to my estimate, the following characters will be added in this season

• NarushimaMelt

• KaburagiMasaya

• KichijoujiYoriko

• ShiranuiFrill

• KotobukiMinami

• MemCho

• KumanoNobuyuki

• SumiYuki

• MorimotoKengo

Please also see the version note of each version for more details

To avoid spoiler, an image that is related to the plot is put only in the gallery of the FullCkpt version

The models can do eye right but for better quality I would always suggest inpainting

Dataset composition

fanart 168

screenshots

• EP 01 1268

regularization ~20K

LoHa for kofji sister (with kawashiro_nitori skin)

Recommanded prompts:

`blue shirt, white skin, black pants, best_buy, blue hair`

It's recommanded to use https://github.com/hnmr293/sd-webui-cutoff extension to get best character style and the target tokens is `blue,white,black`

LoHa for udk sister.

To enhance the character features, you can add these tags to prompt:

`single braid, purple bowtie, witch hat, black headwear`

A chaotic brush artstyle

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。

Legal & Risk

It is prohibited to use this model for commercial purposes and any scenarios of illegal acts and purposes.

In principle, I do not support the use of this model to generate any adult content. When using this model, users must abide by laws and regulations, and do not infringe other people's reputation, privacy or portrait rights.

Changelogs

• v1 : LoRA version, first version

• v2 : updated to LyCORIS, still can be use as LoRA

• v3 : updated datasets training to give better generated image

Usage

ATTENTION, YOU NEED TO INSTALL THIS EXTENSION TO USE THE LyCORIS VERSION IN SD-WEBUI EXTRA NETWORKS

https://github.com/KohakuBlueleaf/a1111-sd-webui-lycoris

WARNING!!! Due to the datasets nature, it's really hard to generate SFW image.

Any Sd1.5 based model will produce good results

• 512,768 image

• DPM++ SDE Karras

• Use 20-30 sampling steps

• Weight: 0.5-0.8

• CFG: 4.5-7

If using Hires:

• Hires step: <20

• R-ESRGAN 4x+ or Latent variants

• Denoising: 0.3-0.7

After download this model, try to share with us about your findings.

Please give star reviews with your picture creation if you like it.

But, don't use this for commercial use and she is not an existing person!

p.s. It's not necessary but If you like my LoRAs please consider to support me to create more projects by donate in here:

https://ko-fi.com/gorillamonsooniii

I'm also accept commisioning for LoRA

Made using the Kohya LoRA Dreambooth method https://github.com/linaqruf/kohya-trainer

Trained on AnyLoRA, haven't tested too much yet (I'll edit with more samples later)

0.7~1.0 weight depending on style you want
Hires fix

Descriptive tags help, black eyes, spiked hair, barefoot, etc
Negative tags, border, letterbox, multiple boys

This is a costume LoHa. Not character LoHa.

weight: 0.6-

1boy, solo, orange hair, short hair, mask on head, red mask, grey coat, mask, half gloves, stand, look at you, red scarf, single red earrings, tartaglia \(genshin impact\), grey pants, close mouth, blue eyes, ruby on scarf, masterpiece, best quality, <lora:childeloha:0.75>

权重0.6-0.85之间都可以的，如果想融合别的lora权重可调低到0.3-0.45

训练依赖的大模型：dalcefoV3Painting

这个版本比较内向，有过拟合，随缘更新

This one was very tricky since I didn't have a lot of data to work with. Outputs may be unstable, do not use "multiple tails" as it will try to give her 9 tails instead of 3!

As always, check the preview pictures for example prompts!

2023/4/15 update

Upload lycoris version. The preview images of the new model are generated using 'hires fix' and upscaled with 'sd upscale'.

Introduction

A LoRA model trained with ~180 images by Dishwasher1910. The 784mb VAEs (NovelAI, Anythingv3, Orangemix, Counterfeit) are recommended. 0.6 ~ 0.8 weights are good. The preview images are generated using latest version of YesMix and EasyNegative embedding. It's also recommended to use this and any Azur Lane LoRA (e.g. Musashi, Taihou, Shinano) to generate high-quality masterpiece.

Decided to create a different vaporwave aesthetic compared to the ones already uploaded on the site.

Seris from Paladins

Add holding, lantern tags to generate her weapon

This LORA is Yuzurizaki Nero from Detective Opera Milky Holmes.

An obnoxious girl with a hefty appetite. She speaks with the masculine term 'boku' and is very selfish, often talking down on others while comforting her own greed. Her Toys allows her to control machines using a small piece of metal.

inital tags:

"yuzurizaki nero, orange hair, short hair"

main costume:

"hat, deerstalker, capelet, detective uniform"

"school uniform, necktie"

"serafuku, sailor hat"

You may want to use 7thLayerAbyss or AOM3A.

凸(☎︎ω☎︎)凸 凸(☎︎ω☎︎)凸

Kuso Model.

Trained on Candy, Gems and other goodies.

Thanks to Barrett for the suggestion and most of the data.

We used SOME of our own data we picked up and some Niji crystals data we found on hugging face.

The outputs are JUST FOR NOW from the kohya SS train, and if you got better ones show us!

We used BLIP to caption it, and these are the custom tags!

LyCORIS of Helen Hunt

LyCORIS of Felicity Jones.

LyCORIS of Emily Hartridge

LyCORIS of Dichen Lachman

🌟Want to buy me a coffee or commission me a lora ?🌟

Nothing much to say, liked the style, made the lora

LyCORIS of Chase Masterson

LyCORIS of Audrey Marie Anderson

LyCORIS of Ana Kasparian

A LoHa trained with 12 images. 1 batch size. Uses my own method.

LyCORIS of Adrianne Palicki

A LoHa made using a 30 image dataset and my own methods. 15 batch size.

V1 - a lot more repeats, 5 epochs

V2 - 30 epochs, a lot less repeats

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。

https://twitter.com/asanoruri/status/1135778784899026944
朝ノ姉妹ぷろじぇくとは同人マークを応援しております🙆🏻✨もちろんおっけー🙆🏻✨
引用ツイート
いかやきくん！
@Ikayaki_tail
·
2019年6月4日
朝ノ姉妹の二次創作ガイドラインってあるんだっけ？
ガイドライン的に大丈夫そうなら朝ノ瑠璃AIスピーカーをイベントで自分のサークルスペースに展示したいんだけど。布教活動の一環として。

Although there are no secondary creation guidelines, we have decided that the Asano Sisters Project allows doujin activities.

二次創作ガイドラインは存在しないが、朝ノ姉妹プロジェクトは同人活動を許していると判断しました。

I have created a Locon model for Fujiwara Chika from Kaguya-sama: Love Is War. While there is already a Lora model available for her, it struggles with drawing the hair bow on her bangs.

My Locon model is designed to be easy to use, requiring fewer trigger words and making it easier to draw the hair bow. For generating original character designs, I suggest using a weight of 0.8-1.

If you want to change her clothing and use keywords to describe her features, I recommend using a weight of 0.5-0.6.

Here are my training configurations:

• Training data: 59 images

• Based model: anything-v3-fp32-pruned

• network_dim: 40

• network_alpha: 40

• conv_dim: 40

• conv_alpha:40

• optimizer_type: DAdaptation

• mixed_precision: fp16

• clip_skip: 2

• learning_rate: 1

• unet_lr: 1

• text_encoder_lr: 0.5

• max_train_steps: 600

• lr_warmup_steps: 50

• batch size: 4

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。

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。

BigrBear | Stylized「LoCon」

Training models take alot of time and effort! Support me by leaving Feed backs and Reviews! ☕

This is LyCORIS Network as such you will need an extension to utilize them.

(I recommend using it natively instead of loading the network via KohyaSS Extension as the Kohya extension likes to break at times.)

My models are made to be used at [ Weight 1 ] Unless it is being combined with other Networks.

There are 2 Versions to this model.

「Normal」and「Hard」

「Normal」is very flexible and you're able to prompt just about anything with it.

「Hard」is Fixated to the dataset and will create backgrounds very similar to the dataset with the artists stone like texture on every background image. Its not recommended to use but its there for the sake of being there.

TOS! Terms of Service!

• You are not allowed to sell generated images

• You are not allowed to host this model on generation services

Exceptions : Unless the style of this model is merged enough with different styles making it not easily distinguishable then the TOS does not apply to such derivative models.

Here's a google colab to run this LoCon along with other models If you don't meet the requirements to run Stable Diffusion Locally.

We will comply with the VERSE secondary creation guidelines.
https://verse-n.com/guideline/
Please read the secondary creation guidelines.
二次創作ガイドラインをご一読お願いいたします。

just for fun.I capture from low res video.just feel this man is very funny.so don't care.

John Doe - A Style Lora

This is more or less an experiment in creating Lora/Locon, and not meant to be a replacement or a strict recreation of John Doe's work. Please support the original artist!
https://www.patreon.com/JDArt

This is a style Lora (a Locon, specifically. So you'll need the extension to properly use it) meant to emulate the style of John Doe (dimensional forms, dramatic lighting, "realistic-er" facial features). That's pretty much it.

Seems to work best for me at around 0.9 strength, but feel free to play around with it. Lower settings will flatten things out and bring out more "anime" features (though certain tags like "smirk" tend to have a strong effect of bringing those out as well).
It was trained on explicit images as well, so it should be pretty fine generating actual intercourse, and most subjects therein. Though you might have the occasional floating dick here and there.

While the Lora also does well enough as you can expect at 512x512, I recommend using a slightly higher resolution to bring out facial features a bit more. Either by using a higher base res, or just turning on hires fix. It also seems to work well with character Lora as well as Lora depicting real people. I've given a few celebrity Loras a try, and while not a perfect recreation, their traits do show up, even if they were trained on different models and tagging systems. A surprise, but a welcome one!

Since this was kind of a "lazy" Lora that I didn't meticulously curate, things might get a little off from time to time. Your mileage may vary, but the age-old adage reigns true, the better your prompts, the better your result. Still, it's absolutely possible to get decent results using basic prompting.

Feel free to leave any comments you've got as well as any images you have made!

The Lora is made using my usual workflow of training 2 separate Lora. One on full bodies, and another with head crops, and merging them together.

Nobunaga's Ambition is a series of turn-based grand strategy role-playing simulation video games. The original game was one of the first in its genre, being released in March 1983 by the Japanese video game developer Koei. Nobunaga's Ambition takes place during the Sengoku period of feudal Japan. The player is tasked with achieving the ultimate goal of warlord Oda Nobunaga: the conquest and unification of Japan.bla bla bla,hey man just take it!

A Character LoCon for Reki Kyan (喜屋武暦) 【sk8 the infinity】

Weight:0.4～0.6

Hires. fix

trigger words: kyan reki\(sk8 the infinity\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

【注意】スケボが題材なのにスケボに乗った画像生成はレアです。スケボに対して暦くんがいい位置に来ない、スケボが2枚出現する、スケボが異様に小さい、ローラーが６つ生成、靴からローラーが生えてくる、スケボが曲がっているなど…t2iで200生成近くテストしましたが指ガチャも合わさり根気がいります。体感512*768よりも上のサイズ640*960などだとまともになりやすい気がします（高くしすぎるとSD特有の奇形体型になる）。

どうしてもスケボ暦を作らせたい場合は大人しくcontrolnetかi2iで修正を加えつつが大正義かもです…。また学習の過程でグラフィティアートとスケボ柄をユーザーIDやサインとご認識している感じがありますので気になる方はネガティブにuseridなどお書き下さい

共通呪文 : kyan reki\(sk8 the infinity\), red hair, short hair, orange eyes, headband

※orange eyesはyellow eyesとお好みで使い分けて下さい。第7話のあのシーンを再現したい場合は下記にも書きますがheadbandを消しておいて下さい

デフォ衣装

出現率アップ呪文:black gloves, hoodie, wristband, yellow hoodie, black jacket, black pants, school uniform

※靴が欲しい場合はshoes, sneakers, 追記。基本青色が出現すると思いますが出ない場合はblueを足して下さい。

私服

出現率アップ呪文: 〇〇 hoodie,

※基本パーカー属性の彼なので〇〇に色指定。

7話のアレ（髪下ろし）

出現率アップ呪文: hair down, hair between eyes

※プロンプトのheadbandを消しておいて下さい。ネガティブにheadbandを書かなくても成功率100％だったので多分大丈夫かなと思います。

以上です！！

最後に、暦くん以外のSK∞ エスケーエイト他のキャラLORAもしくはLOCONを作るつもりはないです。

We are all love ALIEN right?

Add at 1.0 weight, prompt sarashi or fundoshi. Base anime models couldn't do them well enough so I trained this.

Sarashi are the chest wrap, fundoshi are the underwear.

True Pilyeon

Model Made on the Art Generated from the previous model.

This model shouldn't violate any kind of copyright or delusional/Ethical/Self-made/Unethical laws.
Please, I'll appreciate your support on my Patreon and kofi.

Here is a Form you can request me Lora there (for Free too)

https://forms.gle/h9iuLCCnaEs6W9ZB8

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

The Rito species as depicted in Breath of the Wild!

Use tags "rito" and "winged-arms".

This is intended as a Rito species lora, but it can do a few specific Rito to varying degrees of success. It's not perfect for that though, so don't leave a bad review if you can't get a perfect Kass!

Characters trained in this include:
Kass, Revali, Teba, Saki, Amali, Cecili

If going for one of those 6 characters, adding their body color and gender to the prompt can help. for exable, (Revali male dark_blue_body) will be more consistent than just Revali.
Not including "rito" in the prompt and just the character's name can be more consistent as well.

Works well with most models I have tested it with. Have fun!

This is a model of Jesus Christ from the show Black Jesus.

Use jesus-christ as your tag and it should work. Seems fine at a power of 1 but I'll do some testing. This one is fresh out just for Easter.

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。

I would suggest using "multicolored hair, streaked hair, abstract, colorful" for the full effect.

「この姿は……ああ、そういう事なのですね。最早こうなれば、我が身に流れる鬼の血を厭うのは、今日この時を以てやめましょう。巴は、持てる力の全てを以て戦います」

• CFG Scale: 5.5 - 7

• Weight: 0.8 - 1

• Recommended prompts: To Be filled. (See sample images for now)

✨✨Want your personalized LoRA or illustrations? Click here✨✨

あなただけのLoRAやイラストが欲しいですか？ ここをクリック

🎨 Pixiv https://www.pixiv.net/en/users/22936156

🐦 Twitter https://twitter.com/tsumasaky

📷 Instagram http://instagram.com/tsumasaky

Note #1: Please read the "how to use" section for this Locon. While it does a lot of the heavy lifting to get a good picture, it definitely still needs a little help from generic tags.

Note #2: This is a Locon. Install the extension to get it to work. I've found them to do better when working on ideas that I want easily added to one of my LORA characters. Also, they're WAY less damaging to images, even at high weights.

After many, many version of this, I think it's in an okay enough state to release. So, here it is, my brainwashing Locon. It's still not perfect and could use a v2 (and v3, and v4, etc.) but I've made 9 versions so far and I just want to release something lol. It was trained on animefull_final with 101 images with the usual self-pruning and tagging and a wide range of different images. I like keeping my prompts simple for the first few pictures so that you can see it without a whole bunch of mumbo jumbo... I would say. This one is a teeny different. Make your pictures better than mine though!

What does it do you ask? A Locon that will create a "brainwashed" waifu who has the general blank stare and lustful. Sex optionally can be included.

How to use it: I've found that with these Locon's I've been submitting, weight's can be wide and varied. This one specifically can range from 0.6 to 1.4 in terms of how much you want it to take over the image.

Ok. Now trigger words. First, baked into the Locon we have:

BrnWsh - Main word if you just want to get the thing to be recognized well. Use it hwen you don't care which of the next option you get AND when you want to force the look even more-so.

NFucked - To lean the Locon to create images where the character is not having sex.

BotFucked (recommended 512x768 width x height) - To lean the Locon to create images where the character is being fucked while on top of someone.

SFucked (recommended 768x512 width x height) - To lean the Locon to create images where the character is being fucked from the side.

ADDITIONALLY USE THESE:

1boy, hetero, penis (or dildo), sex, vaginal (or anal or oral) - I like pairing my images with these to increase the chance of success of the image.

OPTIONALLY USE THESE:

glowing eyes, rolling eyes, swinging pendulum, fucked silly, clothes lift - Or anything else you want to be emphasized. It should adapt to what you add.

IMPORTANTLY, EVEN WITH ALL THIS SOME PICTURES WILL JUST BE BAD :(.

Extra, Extra!: Should be well trained and usable in most models. Images shown for example.

Picture Descriptions:

1-3: AnythingV3 using just this Lora with simpler prompts

4: AnythingV3 combo'd with my in progress V2Corruption Locon.

5-6: AnythingV3 combo'd with my released Seraphina (Disgaea 5) Lora.

7: AnythingV3 combo'd with my release Ghislaine (Mushoku Tensei) Lora.

8: AnythingV3 combo'd with my in progress v2Ameri (Iruma Kun) Lora.

8: AnythingV3 combo'd with my in progress "Symbioted Taking Over" Locon (basically a symbiote merging with the user).

9: AbyssOrange with just this Lora.

10: PerfectWorld using just this Lora.

What's next for this? Find ways to make this better. Also, bring back the BoobFucked and SPendulum tags that I had to cut since they weren't helping much.

What's next that isn't this?

More character Loras. (Ivy from Soul Calibur 5, Heine from Combatants Will Be Dispatched, Conqueror Lilias from E7, Rozalin from Disgaea 2, Tharja from Fire Emblem, Catherine from Catherine Full Body, Miku from Idaten Deities, Kuribayashi Shino from GATE, Rin Sohma from Fruits Basket, Kasumi from P5, Ann from P5, Ikumi Mito from Food Wars, Alice Nakiri from Food Wars, Chika Fujiwara from Kaguya-sama, Lulu from FF10, Momo from My Hero, Vivy from Vivy, Dominique de Sade from Vanitas, and Dos from One Punch Man.) And maybe Wilhelm van Astrea, my favorite Re:Zero husbando or a Prinny from Disgaea.

V2's of some of the previous Loras. Specifically make them more diversely acceptable with other models and LORA's.

I also have some more quite NSFW Locon's. I decided to list them as well, so if people have preferences on which to fix up first let me know. Specifically, CorruptV2, Symbiote Taking Over (imagine a venom-like symbiote taking over your waifu), being stepped on, Robotification, Petrified (I'm unsure if I should bother with this one anymore), Infection, Dominatrix (waifu becoming a dominatrix), Brainwashing Machine, Tentacles, Pillory, Masked Evil (A mask that possess' the waifu), Latex Corruption, Hourglass Figure, Combat Drone (hard to describe but basically making them a mindless sci-fi combat warrior), JAV Supervillain, Jiangshi, and Holding a Leash.

And, uh, I want to make a few "normal" Locon's so I have an "knight-ify" your character one coming up. And a forest fairy one as well. And a officer one. And a barbarian one.

Or just give me some random suggestion or priority opinions. That works too.

A lycoris model of bed invitation pose.

Trained on 55 anime pictures on NAI model.

Trigger: bed_invitation.

Samples are generated on Anything4.5 with LoRA Block Weight.

Here is the Lora for handjob from behind !

The trigger words is reach-around

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Introduction

A Lycoris model trained with ~20 images of Medusa from Battle Through the Heavens. The trigger tag is ‘c-medusa’. 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. This model is a bit overfitted due to few training images, which will be fixed when dataset is updated.

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.

Mellow swimsuit worn by Ayane from the Dead or Alive franchise.

• Trained with Anime (full-final pruned) model.

• Recommended using with inpainting by masking the desired areas for the outfit to appear.

• Recommended making a image of a character wearing a green two-piece swimsuit (with short included) before applying this LoRA with inpainting.

実写向けのアニメ調のアジア人風にするLoRAです。

他の顔LoRAと併用して調整に使用する方が良いかも知れません。

層別LoRAは適用していません。

~~~Translated by DeepL~~~

This LoRA can resemble an Asian person in a anime style.

It may be better used in conjunction with other facial LoRAs for adjustment.

No lora block weight are used.

~~~~~~~~~~~~~~~~~~~

weights: 0.6~0.8.

trigger words: none. (1girl)

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。

Making loras takes up a part of my free time, any help is welcomed: https://ko-fi.com/truefalse

Sometimes in rear view it wants to make a different vintage car, don't know why
Also it's a shame i didn't manage to get the ai to generate the big beautiful multiple exhausts this car has...
image grid: https://files.catbox.moe/tks9q0.png

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。

2023/4/9 update

Update dataset. Upload lycoris version. v1.0 version is conventional LoRA model. The preview images are generated using 1.6 version (original) YesMix.

Introduction

A conventional LORA model of musashi (azur lane) trained using 160 highres images, including nsfw. The NAI's VAE is recommended. 0.6~0.8 weights are good.

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.

We will comply with the with the secondary creation guidelines.
https://vspo.jp/guide/
Please read the secondary creation guidelines.

Since it is a lokr model, please use the latest version of webui and the extended function locon.

Retrained with LoCon using Hollowstrawberry Colab

• Smaller size 28.9mb

• Better costume

<lora:human_anatomy_v1-9:0.7>,human_anatomy_style

the character should expose some skin without clothes.

A Character LoCon for Will.Sprout (ウィル・スプラウト) 【Helios Rising Heroes】

Weight:0.4～0.6

Hires. fix

trigger words: will sprout\(helios rising heroes\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

共通呪文:will sprout\(helios rising heroes\), blonde hair, short hair, brown eyes

制服

出現率アップ呪文:collared shirt, belt, black shirt, black pants, belt buckle, yellow necktie, police uniform, helios uniform,

ヒーロースーツ

出現率アップ呪文: black gloves, gloves, black shirt, yellow gloves, goggles, goggles on head, two-tone gloves, yellow jumpsuit, headset, yellow coverall, the kindhearted green thumb

※the kindhearted green thumbはお守り呪文（greenの文字があるので謎緑が生成されるので無いほうが吉かも）

水着（真夏のサバイバルアイランド）

出現率アップ呪文:shorts, sleeveless, hood, fingerless gloves, goggles, goggles on head, swim trunks, sleeveless jacket, yellow jacket, hoodie, diving_mask, diving mask on head, Survival Island

※Survival Islandはお守り。ジャケットを開けたがる癖あり。ガチャをするか呪文で制御お願いします。原作にあるような腹チラはmidriff peekで…（制御むずい）

Kaho from Blue Archive

• Trained on YesMix 1.6

• Weights between 0.5-0.8 seem good. Use higher if you want to use with original clothes.

• Careful with colours, likes to bleed into clothes.

A model trained on _himehajime and their style of artwork. (おののいもこ

魚介 @_himehajime). https://www.pixiv.net/en/users/165192

Note:

This is a LYCORIS/LoCon model, and you need to install a1111-sd-webui-locon extension to use this model.

This is a LoRa with a vector graphic art style, which tries to showcase simplicity and flatness in colors as much as possible.

There are two versions available here, namely the trim version and the original version. For specific differences, please refer to the version introduction. Generally, downloading the trim version is sufficient.

Trigger word:flat color

Suggest weight:around 1 or lower

I really like the color scheme of this LoRa, and by the way, you could add chibi tag to make the character generated even cuter and more distinctive.

Based on novel-anime-full model.

I made it for another purpose, it's pretty easy to accidentally call a bear
Steps: 50, Sampler: DPM++ 2M Karras, CFG scale: 4
don't call dog 😂

WARNING: This is not meant to be used to deepfake this brilliant man. IF anything this file is a mere tribute. (Queue Jack black and Arin Hanson singing) - Yes, i'm sure you could produce your own only fans, but let's be honest - Why would you?

This is a FAN created piece, and yes sure Mark I've literally taken screenshots off your videos - but y'know I don't intend on doing anything dirty with this (and I would sincerely hope nobody else does LOL).

Best tips: Use an anime model at about .7 strength! Comes up with GREAT AI fan art ;)

NO THIS IS NOT MEANT TO REPLACE ART OR FAN ART :3 I just ... I wanted to see if it could do it LOL.

THIS DEPICTS MARK FISCHBACH OR MARKIPLIER.

What you do with the pictures is up to you, I am not advocating or anything to do with the final product of what you make.

However, PLEASE LET SOMEONE TEACH MARKIPLIER WHEN HE HAS TIME TO DO STABLE DIFFUSION ON SCREEN - I really want him to use this. (Or Jacksepticeye, and then have him like laugh maybe... i dunno just don't sue me pls XD)

Urethral Blowjob!! Peehole Blowjob!!👍

LoHA model for TRS sister. It is seems like the background is overbaked so you may add more prompt to background, idk.

We will comply with the with the secondary creation guidelines.
https://vspo.jp/guide/
Please read the secondary creation guidelines.

Anzu Mazaki/ Téa Gardner from Yu-Gi-Oh! Duel Monsters

• Trained with Anime (final-full pruned) model.

• Comes in 3 versions; 6 epochs, 8 epochs, and 10 epochs.

• You can use the following activation tags; anzu_yugioh: for general info, anzu_classic: to get her classic hairstyle, anzu_dsod: to get her Dark Side of Dimensions hairstyle (make sure to type short ponytail, high ponytail to get her ponytail in the movie.)

圣诞之吻/アマガミ/amagami style

没有为角色设置关键词

キャラクターのトリガーワードを設定していない

No trigger words are set for the characters.

可以使用以下角色名来生成对应角色，不保证和角色lora有同等效果

以下のキーワードで対応のキャラクターを生成できるが、キャラクターLoraと同じ効果を保証できない

It can be generated using the following character names, which are not guaranteed to have the same effect as the character Lora.

ayatsuji_tsukasa, morishima_haruka, tanamachi_kaoru, tachibana_miya, nanasaki_ai, sakurai_rihoko, kamizaki_risa, nakata_sae, tsukahara_hibiki, takahashi_maya

例图使用t2i -> i2i -> Tiled Diffusion+Tiled VAE生成，在最后一步中删除了除画质之外的prompt

サンプル図はt2i -> i2i -> Tiled Diffusion+Tiled VAEの順番で生成、Tiled Diffusion+Tiled VAEの時一部のpromptsを消した

Samples are generated by t2i -> i2i -> Tiled Diffusion+Tiled VAE. Some prompts were deleted by Tiled Diffusion+Tiled VAE process, so the prompt of the samples is incomplete.

Samples full pormpt:

1. (masterpiece, best quality, extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer,(1girl, solo, close-up:1.3), dynamic angle,bedroom, (smile:1.3), sitting on bed,
   (kibito high school uniform:1.3),black jacket, pleated_skirt,school_uniform, blazer, large breasts, collarbone,thighhighs, no shoes, white hair, ponytail, long hair, yellow eyes,
   <lora:amagami_lycoris_bf16:0.6>

2. (masterpiece, best quality, extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer,(1girl, solo, close-up:1.3), dynamic angle,bedroom, (smile:1.3), sitting on bed,
   (kibito high school uniform:1.3),black jacket, pleated_skirt,school_uniform, blazer, large breasts, collarbone,(nanasaki ai:1.3)
   <lora:amagami_lycoris_bf16:0.6>

3. (masterpiece, best quality, extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer,(1girl, solo, close-up:1.3), dynamic angle,bedroom, (smile:1.3), sitting on bed,
   (kibito high school uniform:1.3),black jacket, pleated_skirt,school_uniform, blazer, large breasts, collarbone,
   <lora:amagami_lycoris_bf16:0.6> , <lora:ariane_lycoris_bf16:0.35>,arianeglenysmaple,white hair, pointy ears, ponytail, hair ornament, yellow eyes,

4. (masterpiece, best quality, extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer,(1girl, solo, close-up:1.3), dynamic angle,bedroom, (smile:1.3), sitting on bed,
   (kibito high school uniform:1.3),black jacket, pleated_skirt,school_uniform, blazer, large breasts, collarbone,
   <lora:amagami_lycoris_bf16:0.65> , <lora:sasha_lycoris_bf16:0.4>,sashanecron,twintails, blonde hair, blue eyes,

5. (masterpiece, best quality, extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer,(1girl, solo, close-up:1.3), dynamic angle,bedroom, (smile:1.3), sitting on bed,
   (kibito high school uniform:1.3),black jacket, pleated_skirt,school_uniform, blazer, large breasts, collarbone,
   <lora:amagami_lycoris_bf16:0.5> , (white hair:1.4), sidelocks, ponytail, open jacket, open clothes,

Hikage from the Senran Kagura franchise.

• Trained with Anime (full-final pruned) model

• Comes in 2 versions; 3 epochs and 4 epochs, use which ever fits you the most. I also advise you to make combinations of them to get interesting results.

• Try them with any LoRA block weight.

We will comply with the with the secondary creation guidelines.
https://vspo.jp/guide/
Please read the secondary creation guidelines.

This was me experimenting with a plain 1boy style dataset to get an idea of how fast the AI should learn those without overfitting. Uses my own methods. Dataset was provided by somebody else, I hadn't done anything to it except remove 16:9 aspect ratio photos.

27 image dataset.

Just a little extended lora trained with a small dataset of ~130 images.

Dataset is 100% SFW, so it will not help you produce NSFW imagery, but it shouldn't tamper with that ability either, as it's trained with enough classification images and shouldn't bleed in any meaningful way to all your lewd tags.

It's triggered with javi token. Without it effect should be negligible, or not appear at all.

Dataset was gathered from MJ with specific style prompt, though, v5 models are quite rigid(especialli niji), so, im not sure it made something unique enough. But one thing is for sure - it's not relevant to any artist.

Basically, whole thing of this one is i needed to waste my MJ subscription hours, okay? If you would enjoy using it - good xD

If you create any good images using this LyCORIS, please consider sharing them by leaving a review. Thank you!

• This LyCORIS can help generate specific stewardess uniforms Including red and dark blue uniforms.

• This is a LyCORIS, please make sure you have installed the related extension.

• Stewardess, Red uniform, and Dark blue uniform to trigger.

• This model cannot accurately reproduce the appearance of Russian people; it mainly provides the styling of the uniforms.

• If you want to remove the hat, put "garrison cap" and "hat" in the negative prompts.

• The scarf is considered part of the uniform and is difficult to remove separately.

Introduction

A Lycoris model trained with ~260 images by Jeneral. The trigger tag is ‘jeneral’. 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.

This is a locon, please make sure you download the extension

recommended Weight: 0.6-1

recommended model: Anything4.5

Refer to https://civitai.com/models/23476/the-idolmster-cinderella-girls-starlight-stage-style-90-characters, thanks to the author

The main purpose of making this collection is to let the characters wear clothes that they don't usually wear, just like some characters in the example picture, you can try to replace the promt of the example picture with the character you want

Therefore, some characters may not be highly restored or have insufficient details, and have not learned specific clothes, so it is recommended to download or make lora of individual characters

I didn't do too many tests on each character because of the large amount of work, so I hope to get your review and feedback

You can also use this lora as a Artstyle lora, I used genshin character in the example

Please do not use this lora for any commercial purpose

参考了https://civitai.com/models/23476/the-idolmster-cinderella-girls-starlight-stage-style-90-characters， 感谢作者

做这个合集主要目的是想实现让角色们穿上他们平常不会穿的衣服，就像例图中的一些角色一样，你可以试试看将例图的promt替换成你想要的角色

因此有些人物可能还原度不高或细节不够，也没有学习特定的衣服，所以还是建议下载或制作单独人物的lora

因为工程量太大我没有对每个角色做太多测试，所以希望得到你们的返图

你也可以把这个locon用作画风locon，例图中我用了原神的角色

请不要将此locon用于任何商业用途

Recommeneded promt:

sparkle

（if you want the kirakira effect like game cards）

Characters promt:

Subaru Akehoshi, 1boy, male focus, blue eyes, solo, orange hair

Hokuto Hidaka, 1boy, male focus, blue eyes, solo, black hair

Makoto Yuuki, 1boy, male focus, green eyes, glasses, blonde hair

Mao Isara, 1boy, male focus, green eyes, red hair

Chiaki Morisawa, 1boy, male focus, solo, brown hair, brown eyes, orange eyes

Tetora Nagumo, 1boy, male focus, solo, multicolored hair, black hair, streaked hair, red hair, solo, smile, red eyes, teeth,

Kanata Shinkai, male focus, 1boy, green eyes, aqua hair, blue hair

Midori Takamine, 1boy, male focus, brown hair, blue eyes

Shinobu Sengoku, 1boy, male focus, hair over one eye, solo, purple hair, yellow eyes, multicolored hair, streaked hair

Rei Sakuma, 1boy, male focus, solo, red eyes, black hair

Kaoru Hakaze, 1boy, male focus, blonde hair, brown eyes

Koga Oogami, 1boy, male focus, solo, yellow eyes, grey hair

Adonis Otogari, 1boy, male focus, solo, purple hair, dark-skinned male, yellow eyes, dark skin

Eichi Tenshouin, 1boy, male focus, blonde hair,blue eyes

Wataru Hibiki, male focus, long hair, braid, blue hair, 1boy, purple eyes,

Tori Himemiya, 1boy, green eyes, male focus, solo, pink hair, sidelocks

Yuzuru Fushimi, 1boy, male focus, solo, blue hair, purple eyes

Nazuna Nito, male focus, 1boy, blonde hair, solo, red eyes

Tomoya Mashiro, 1boy, male focus, solo, blonde hair, brown eyes

Hajime Shino, male focus, 1boy, purple eyes, solo, blue hair, aqua hair

Mitsuru Tenma, 1boy, male focus, brown hair, brown eyes

Yuta Aoi, 1boy, male focus, green eyes, orange hair

Hinata Aoi, 1boy, male focus, green eyes, orange hair

Hiiro Amagi, 1boy, male focus, red hair, hat, blue eyes

Aira Shiratori, male focus, 1boy, blonde hair, solo, green eyes

Mayoi Ayase, 1boy, male focus, purple hair, solo, braid,long hair, ahoge

Tatsumi Kazehaya, 1boy, male focus, purple eyes, green hair

Tsukasa Suou, 1boy, male focus, purple eyes, solo, red hair

Izumi Sena, 1boy, male focus, blue eyes, grey hair

Leo Tsukinaga, 1boy, male focus, green eyes, solo, orange hair

Ritsu Sakuma, 1boy, male focus, red eyes, solo, black hair

Arashi Narukami, 1boy, male focus, blonde hair, blue eyes

Nagisa Ran, 1boy, male focus, solo, red eyes, long hair, white hair

Hiyori Tomoe, 1boy, green hair, purple eyes

Jun Sazanami, 1boy, male focus, blue hair, solo, yellow eyes

Ibara Saegusa, 1boy, male focus, red hair, blue eyes, solo, glasses

Rinne Amagi, 1boy, male focus, blue eyes, red hair, solo

HiMERU, 1boy, male focus, blue hair, yellow eyes

Niki Shiina, 1boy, male focus, blue eyes, blue hair, low ponytail

Kohaku Oukawa, 1boy, male focus, purple eyes, solo, pink hair

Kuro Kiryu, 1boy, male focus, solo, green eyes, multicolored hair, two-tone hair, black hair

Keito Hasumi, 1boy, male focus, glasses, green eyes, green hair

Souma Kanzaki, male focus, 1boy, solo, purple hair, purple eyes, long hair, ponytail

Mika Kagehira, 1boy, male focus, heterochromia, green hair, black hair, blue eyes, smile, jewelry, yellow eyes

Shu Itsuki, 1boy, male focus, pink hair, solo, blue eyes

Natsume Sakasaki, 1boy, red hair, solo, yellow eyes, multicolored hair, streaked hair

Tsumugi Aoba, 1boy, male focus, glasses, blue hair, yellow eyes, solo

Sora Harukawa, blonde hair, male focus, blue eyes, 1boy, green hair, multicolored hair, streaked hair

Madara Mikejima, 1boy, male focus, solo, brown hair, blue eyes

A Character LoCon for Otori Akira (鳳アキラ) 【Helios Rising Heroes】

Weight:0.4～0.6

Hires. fix

trigger words: otori akira\(heliosrising_heroes\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

共通呪文:otoriakira\(heliosrising_heroes\), red hair, short hair, hair between eyes, bangs, green eyes, ear piercing, earrings, piercing, earclip,

制服

出現率アップ呪文:red shirt, black pants, white belt, police uniform, helios uniform,

ヒーロースーツ

出現率アップ呪文: open clothes, pants, black shirt, black pants, gloves, fingerless gloves, red gloves, hood, jacket, long sleeves, black gloves, open jacket, hood down, hooded jacket, red jacket, two-tone gloves, breastplate, the unyielding wild phoenix

※the unyielding wild phoenixはお守り呪文（効果薄）

水着（真夏のサバイバルアイランド）

出現率アップ呪文:sleeveless hoodie, blue hood, survival iland, red shorts, belt, sandals, bracelet, necklace, white hoodie

※再現を目指す場合、色指定箇所が多が多く混同しやすいのでガチャ覚悟必須。red shortsは丈が短くなる傾向のため期待薄ですがmale swimwear, swim trunksかpantsにして膝辺りから映さないようにするか…

IF世界線のアキラ（イクリプスアキラ）※原作にないので苦手な方はご注意を！

(white hair:1.7), (white jacket, black shirt, white pants:1.5), red eyes

ネガティブ:(red hair:1.6)

※数値は参考程度にお願いします。かなり頑固に赤髪と赤ジャケットを着たがる傾向にあるので数値は高く設定しました

If you want to support me now you can: https://ko-fi.com/truefalse

Trained 10 epochs, car looks great. lowered some settings to make the files smaller, think the quality is still there while the files are smaller. Great!

image grid: https://files.catbox.moe/k5ocq7.png

This was a test after being requested, it uses somebody else's dataset that was left untagged save for the triggerword. Uses my own method. 824 images 15 batch size.

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

Urban Lingerie

We all like hoodies and lingerie.

Why not combine urban wear and lingerie then?

Get creative and mix prompts from my Urban Gal and KPG models!

This is a LoHa model trained from 123 various sources images. The tests found that portraits performed better,which could create a very dynamic expression,and that's where the name comes from.

You can use the LoHa to design eyes & face.

Weight 0.9-1.0 will perform better

Of course the tirgger word is still 光头强lora

Support my video here:https://www.bilibili.com/video/av823626310

My QQ group:https://jq.qq.com/?_wv=1027&k=o89HEhaS

It's a loha model trained wtih Pixiv painter opinew6.

The cute and vivid style always hits my heart straight.

So I'd like to share it with you guys all.

It works well when mixed with other Loras, so you can just create different stylized images of your own Lora.

Weight from 0.6 to 1.0 works ok.

目前使用权重在0.6到1.0均可。

You should refer it in prompt words instead of Additional Network.

注意loha模型不能够在Additional Network 中使用，你应当在prompt中引用。

这是一个画风LoCon，

This is an artstyle LoCon,

目前看来不是很稳定，当我学习了新的训练方法，我会更新这个LoCon。

It doesn't seem very stable at the moment, I will update this LoCon once I learn new training methods.

注意：触发词为：chibi，

Note: The trigger phrase is: chibi,

推荐配合角色lora使用。

I recommend using it with a character Lora.

推荐把这个LoCon的强度设定为1，角色lora的强度设定为0.8，把CFG设置为7，使用 anything-v3进行图片生成。

Setting the weight of this LoCon to 1 and the weight of the character Lora to 0.8，set CFG to 7,use https://civitai.com/models/66/anything-v3 。

根据实际情况进行调整以获得更好的画面效果。

Make adjustments according to the actual situation to achieve better visual effects.

如果有画风训练相关的心得，欢迎分享给我。

If you have any insights on artstyle training, please feel free to share them with me,

以便我能修改这个LoCon.

so that I can modify this LoCon.

• Weight: 1.0 - 1.1

• CFG Scale: 5.5 - 7

• Recommended Prompts: manaka nemu, brown hair, purple eyes, blunt bangs, school uniform, neckerchief, hair bun, sidelocks, long hair

Buy us a cup of coffee and we’ll spill the beans.

https://www.buymeacoffee.com/CoSubcreation

Or Fanbox

https://co-subcreation.fanbox.cc/

Or爱发电

https://afdian.net/a/Co_Sub_creation

Your support is the fuel that powers our world-building. Your ideas are the bricks that shape our world-making!

あなたの応援で、世界は生まれ変わる！あなたのアイデアが、世界を作り上げる原動力になるよ！

您的支持将成为创造世界的动力，您的想法将成为构建世界的基石！

If you have any ideas, suggestions, or questions during use, you can ask them on Discord ↓

ご利用中に何かアイデアや提案、質問がございましたら、Discordにてお問い合わせいただけますようお願いいたします↓

使用中有任何想法&建议&问题，可以去Discord提问↓

https://discord.gg/tVpJMhjUwx

If you’re new to LoHa, please read this:

This model was trained with LoHa, so you need to install the WebUI extension. Trust me, it’s worth it.😉

https://github.com/KohakuBlueleaf/LyCORIS

LoHa model path: models/Lora

Usage: Click on the 🎴 button under “Generate” in the WebUI, choose the Lora tab, select the model, and adjust the weights. It’s as easy as pie.🍰

第一次使用LoHa请看：

本模型训练使用了LoHa，需要安装WebUI的扩展。

https://github.com/KohakuBlueleaf/LyCORIS

LoHa模型路径: models/Lora

使用：点击WebUI“生成”下面的🎴，选择Lora 选项卡，选择模型，更改权重。

LoHaを初めて使う場合は、こちらをご覧ください：

このモデルトレーニングにはLoHaが使用されており、WebUIの拡張機能をインストールする必要があります。

https://github.com/KohakuBlueleaf/LyCORIS

LoHaモデルのパス：models/Lora

使い方：WebUIの「生成」の下の🎴をクリックし、Loraタブを選択して、モデルを選択して、重みを変更してください。

Recommended parameters:

推奨パラメータ：

推荐参数：

Model: blueboys2D_v10

Vae: kl-f8-anime2

Clip skip: 2

weight:0.7~0.9

Sampler: DPM++ 2M Karras

Steps: 25

Size: 512x512|384x640|640x384

CFG Scale: 9 or 14~17(see Instructions: need extension)

Extension setting↓

Enable Dynamic Thresholding (CFG Scale Fix)

Mimic CFG Scale:7

Advanced Options▼

Top percentile of latents to clamp:100

Mimic Scale Scheduler:Constant

Minimum value of the Mimic Scale Scheduler:0

CFG Scale Scheduler:Half Cosine Up

Minimum value of the CFG Scale Scheduler:3

Preferred Prompts:

推奨TAG：

推荐提示词：

Prompt:

1boy,solo,upper body,fata, <lora:Style_GlassyandGlaze (FATA)_v1.0:1:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1>(see Instructions)

Negative Prompt:

EasyNegative, badhandv4

EasyNegative | Stable Diffusion TextualInversion | Civitai

badhandv4 - AnimeIllustDiffusion | Stable Diffusion TextualInversion | Civitai

Instructions:

Pulling the CFG too high can lead to flower patterns on your artwork, but fear not! You can install the Dynamic Thresholding (CFG Scale Fix) plugin to fix the issue:

https://github.com/mcmonkeyprojects/sd-dynamic-thresholding

We recommend increasing the weight, but be warned, it may lead to overfitting. We suggest using the Lora block weight extension to adjust the layering (you can use the recommended values directly from the prompt):

https://github.com/hako-mikan/sd-webui-lora-block-weight

It works well with some Sairu and Hira-nuri base models, but since the model was trained on Blueboys2D, we recommend sticking with the original base model for the best results. It does not support pastel colors, thick paint styles, 2.5D, or 3D base models.

It's great for drawing upper bodies and faces, but drawing full-body images may cause distorted faces, so you'll need to import the image into the image-to-image local redraw tool.

The default hair color is black and the hairstyle is an M-shape. To change the hair color, hairstyle, or eye color, you'll need to add specific prompts (as shown in the example image).

I don't use high-definition restoration because running multiple HD images at once during the lottery stage can't guarantee high-quality results. Instead, I prefer to batch process several good-quality images using the txt2img tool (usually with good composition), then use img2img to increase the resolution of individual images and determine the level of detail to add by adjusting the denoising strength. Finally, for those who really need HD images, I'll use Extras to generate high-definition restoration images at the end.

説明：

CFGの値を高くしすぎると、花模様が発生することがあります。この問題を修正するために、Dynamic Thresholding (CFG Scale Fix)プラグインをインストールすることをお勧めします。

https://github.com/mcmonkeyprojects/sd-dynamic-thresholding

重みは少し高めに設定することをお勧めしますが、オーバーフィッティングが発生する可能性があります。分層を調整するには、lora block weight拡張機能を使用することをお勧めします（推奨ヒントの数値を直接使用できます）。

https://github.com/hako-mikan/sd-webui-lora-block-weight

セルルロース、平塗りタイプの背景をサポートしていますが、学習モデルの背景はblueboys2Dであり、より良い効果を得るためには、元の背景を使用することをお勧めします。また、粉彩系、厚塗り系、2.5D、3Dタイプの背景はサポートされていません。

顔や上半身の描写には慣れていますが、全身の描写は難しいため、一部分を描いた後、txt2imgで画像を再構築する必要があります。

デフォルトの髪色は黒色で、髪型はM字型です。髪色、髪型、瞳の色を変更したい場合は、適切な指示語を追加する必要があります（例えば、図のように）。

抽選フェーズで多数の画像を一度に生成すると、高品質の画像を保証することができません。通常、私は画像の品質が良好で構図が適切な画像を一括生成し、img2imgを使用して単一の画像の解像度を高く調整し、Denoising strengthを調整して詳細度を調整します。最後に、Extrasで高品質の修正画像を生成することができます。

说明：

CFG拉的过高会导致花图，可以安装插件Dynamic Thresholding (CFG Scale Fix)来修复：

https://github.com/mcmonkeyprojects/sd-dynamic-thresholding

权重建议高一点，但可能会导致过拟合，建议使用扩展lora block weight来调整分层（可以直接使用推荐提示词里的数值）。

https://github.com/hako-mikan/sd-webui-lora-block-weight

这里是大佬的一些关于这个插件的教程和预设@大江户战士

https://www.bilibili.com/video/BV1Wv4y157NH/

支持一些赛璐璐、平涂类底模，但由于训练的底模就是blueboys2D，所以如果您想得到较好的效果，还是建议使用原底模。不支持粉彩系、厚涂系、2.5d、三次元底模。

擅长画上半身和脸，画全身容易崩脸，需要导进图to图局部重绘。

默认发色是黑色，发型是m形，想要更改发色、发型和瞳色需要加提示词（如例图）。

我没有使用高清修复是因为，在抽卡阶段一次性跑多张无法保证画面质量的高清图比较浪费时间，通常我比较喜欢在txt2img批量跑出还不错的图后（通常是构图比较好），转img2img直接拉高分辨率跑单张，并通过调整Denoising strength来决定添加细节的程度，最终跑出符合预期的图。如果确实需要高清图，在最后才会拉到Extras生成高清修复图。

In fact, you can also use img2img to generate beautiful anime girls, for example: ↓

実は、img2imgを使って美しいアニメの女の子を生成することもできます。例えば、↓

其实它也可以用img2img来生成好看的妹子，比如↓

Walkure Romanze style

Can generate knight armor with pleated skirt and school uniform of Walkure Romanze.

No trigger words are set for the characters.

Samples are generated by t2i -> i2i -> Tiled Diffusion+Tiled VAE. Some prompts were deleted by Tiled Diffusion+Tiled VAE process, so the prompt of the samples is incomplete.

Samples full prompt:

1. (masterpiece, best quality,extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer, (1 girl, solo,street, upper body,dynamic angle),(victory pose:1.2), (smile, shiny_eyes,large breasts),sidelocks,long hair, armor, (white hair:1.5), (ponytail,golden eyes:1.4),breastplate, red skirt, thighhighs, greaves, gauntlets, lace_trim,<lora:WalkureRomanze_lycoris2_bf16-000025:0.6>

2. (masterpiece, best quality,extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer, (1 girl, solo,street, upper body,dynamic angle),(spread legs:1.2), (smile, shiny_eyes,large breasts),sidelocks,long hair, , (white hair:1.4),(ponytail:1.5), (golden eyes),white competition swimsuit, thighhighs,black greaves,black gauntlets, lace_trim,<lora:WalkureRomanze_lycoris2_bf16-000025:0.6>

3. (masterpiece, best quality,extreme detailed beautiful eyes), best illumination, perfect lighting,look at viewer, (1 girl, solo,bedroom, upper body,dynamic angle),(spread legs:1.2), (smile, shiny_eyes,large breasts),sidelocks,long hair, (white hair:1.4),(ponytail:1.5), (golden eyes:1.3),military_uniform, thighhighs, aiguillette, epaulettes,pleated_skirt, zettai_ryouiki,panties, <lora:WalkureRomanze_lycoris2_bf16-000025:0.4>

This is retrained from thesauceer dataset, just a version of his Tutu LoRA, please check out his original work! He was kind enough to share his dataset.

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.

Trigger words for V2.20:

positive: <lora:Tutuv220:0.5>(tutu)

You can also try "ribbon-trimmed skirt", "layered skirt" and "miniskirt".
If failing to make full shirt, try adding either "leotard" or "tank top" or "covered navel" to prompt.

Keep in mind you might have to adjust LoRA and/or tag weight for it to appear, defaults to "ribbon-trimmed skirt" most of the time! Dataset wasn't an exactly split between the different concepts.

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

Sample images generated with 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.

It was trained on 62 512x512 images provided by thesauceer, 20 epochs using Animefull-final-pruned for the Training Model.

Danbooru tags generated by Waifu Diffusion 1.4 Tagger.

LyCORIS of Jarosław Kaczyński

LyCORIS of Janusz Korwin-Mikke

Weights: 0.8-1.4

This is an experimental and somewhat unstable attempt at squeezing 9 different versions of Shinobu Oshino/Kiss-shot from the Monogatari series into one model. Getting desired results may be difficult, please check the preview images for the most reliable prompts for each version of the character!

Versions and tokens:

• Adult Kiss-shot | kiss-shot acerola-orion heart-under-blade

• Teen Kiss-shot (Kizumonogatari) | kizu_teen

• Teen Kiss-shot (TV) | tv_teen

• White dress (Kizumonogatari) | kizu_long

• Pink dress/short hair (Kizumonogatari) | kizu_short

• White dress (TV) | tv_white

• Pink dress (TV) | tv_pink

• School uniform (TV) | tv_student

• Gothic dress (TV) | tv_gothic

Again, check the preview images for example prompts!

LyCORIS of our boy Asmongold.

Suggested weights: 0.8 - 1.4

LyCORIS of Antony Starr as Homelander.

This was trained on purely Homelander pictures by design. Still, it is sometimes possible to get other characters.

Suggested weights: 0.8 - 1.4

LoCon trained on the artstyle of kojima ayami, who did the art for a lot of Castlevania games.

It is not quite her style but still looks good enough on it's own.

Trigger word is SotnStyle.

I recommend weight 0.7~0.8, or 0.9~1 for higher resolutions.

Trained on NAI.

LyCORIS of Thomas Jane

Suggested weights: 0.8 - 1.4

A Character LoCon for Niki Shiina (椎名ニキ) 【Ensemble Stars!】

Weight:0.4～0.7

Hires. fix

trigger words: shiina niki\(ensemble, stars!\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

共通呪文:shiina niki\(ensemble, stars!\), long hair, bangs, blue eyes, shirt, grey hair, hair between eyes, ponytail

Crazy:B衣装

出現率アップ呪文: jacket, choker, belt, necklace, black pants, open jacket, white shirt, white jacket, Crazy:B jacket, Crazy:B Uniform

ホットリミット衣装

出現率アップ呪文: navel, jacket, purple pants, necklace, ring, tank top, yellow jacket, open jacket, open clothes, hot limit_hot summer

ホットリミット衣装 (SCR)

出現率アップ呪文:navel, jacket, purple pants, necklace, ring, tank top, purple jacket, open jacket, open clothes, hot limit_hot summer

Please see https://huggingface.co/alea31415/bofuri-full for details

It seems that many people are not aware of my bofuri model so I extract a locon as usual for easier use.

The model is trained on top of ACertainty with clip skip 1.

Trigger words

• BoDrag

• BoDread

• BoFrederica

• BoIz

• BoKanade

• BoKasumi

• BoKasumi sarashi

• BoKuromu

• BoMaple

• BoMaple angel form

• BoMaple beginner outfit

• BoMaple black armor

• BoMaple machinegod form

• BoMaple pajama

• BoMaple sheep form

• BoMaple uniform

• BoMarcus

• BoMay

• BoMii

• BoMisery

• BoPain

• BoSally

• BoSally cape

• BoSally pajama

• BoSally turtleneck scarf

• BoSally uniform

• BoShin

• BoYui

• Dorazou

• Fox Oboro

• Monster Atrocity

• Turtle Syrup

• Hydra

Anything + dataset(FineTune)

I've been thinking about things lately, you know, like the things we think about when we're not really thinking about anything.

This is still an early version and will be improved in the future.

Enjoy!

Contact → My Twitter

Here is the Lora for buttjob !

The trigger word is buttjob

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Note: This is a Locon. Install the extension to get it to work. I've found them to do better when working on ideas that I want easily added to one of my LORA characters. Also, they're WAY less damaging to images, even at high weights.

I had made a couple NSFW style Locons (and I have a few more in development with sneak peaks here), but I don't always want my waifu's to become horny and evil. So. Here I am combating the evil part with an Angel (warning, still can be horny though). It was trained on animefull_final with 29 images with the usual self-pruning and tagging and a wide range of different images. I like keeping my prompts simple for the first few pictures so that you can see it without a whole bunch of mumbo jumbo. Make your pictures better than mine though!

What does it do you ask? A Locon that will transform your waifu into an angelic form. Usually with wings. I leave these Locon's very general, so if you want to push a specific direction with your images (like a halo/wings), I'd tag it. Otherwise it'll be random on what form it takes.

How to use it: I've found that with these Locon's I've been submitting, weight's can be wide and varied. 0.5 is fine is you want only a small effect. But a lot of time you can go as high as 1.5 if you want to really push the LORA you're changing.

Also the trigger word is:

AngelT - Only trigger word currently added.

Optional - Of course you can add other words that describe angels if you want specific components and to make sure they get into the picture

Extra, Extra!: Should be well trained and usable in most models. Images shown for example.

Picture Descriptions:

1-3: AnythingV3 using just this Lora. Showing different types of outfits can be called when just using this.

4: AnythingV3 combo'd with my Seraphina (Disgaea 5) Lora.

5: AnythingV3 combo'd with my in progress Ghislaine (Mushoku Tensei) Lora.

6: AnythingV3 combo'd with my in progress V2Corruption Locon.

7: AnythingV3 combo'd with my in progress Brainwashing Locon.

8: AnythingV3 combo'd with my in progress "Symbioted Taking Over" Locon (basically a symbiote merging with the user).

9: HassakuHentaiModel just with the Lora.

10: AbyssOrange with just the Lora.

11: PerfectWorld using just this Lora.

What's next for this? Unsure. Feedback will be the decider if I do a V2.

What's next that isn't this?

More character Loras. (Ghislaine from Mushoku Tensei, Conqueror Lilias from E7, Rozalin from Disgaea 2, Tharja from Fire Emblem, Catherine from Catherine Full Body, Miku + Piscalat from Idaten Deities, Kuribayashi Shino from GATE, Rin Sohma from Fruits Basket, "Kasumi" from P5, Ann from P5, Ikumi Mito from Food Wars, Alice Nakiri from Food Wars, Chika Fujiwara from Kaguya-sama, Lulu from FF10, Momo from My Hero, Vivy from Vivy, Dominique de Sade from Vanitas, and Dos from One Punch Man.) And maybe Wilhelm van Astrea, my favorite Re:Zero husbando or a Prinny from Disgaea.

V2's of some of the previous Loras. Specifically make them more diversely acceptable with other models and LORA's.

I also have some more quite NSFW Locon's. Specifically, being pet play, being stepped on, brainwashing, corruptedv2, and one that might be too much for me to even upload here >.<.

And, uh, I want to make a few "normal" Locon's so I have an "knight-ify" your character one coming up. And a forest fairy one as well. And a soldier one. And a barbarian one.

Or just give me some random suggestion or priority opinions. That works too.

Two versions, soft and hard.

enjoy yourself!

• 欢迎加入AI绘画模型博物馆QQ群

• ①群：791222249

• ②群：599962550

Locon trained on 81 cum filled shoes.

Performed well when drawing shoes only.

Have a great tendency of making any shoes into high heels.

A Locon of ba-zerarudo from Frame Arm Girls.

Trained Trigger words:

ba-zerarudo | leotard | pantyhose | rabbit tail | high heels

Pictures below are generated on Any4.5 with Controlnet.

BOOOOBA Eldridge :D

Trained trigger words:

Eldridge

default outfit | Kitty idol | Holy Night's Embrace | The Inquisitive Recluse | New Year's Outing

The "New Year's Outing" is overfitted and may have poor performance.

loha training just for fun

Hires step 20

Denoising 0.6

This model was made trained on 160 images made by Chelodoy, most of theme were NSFW, but the model works flawlessly with any type of image, it is black magic.

Trigger word is chelodoystyle. But it is unnecessary to use it. Model should work just fine at 1 weight.

A Character LoCon for rinne amagi (天城燐音) 【Ensemble Stars!】

Weight:0.4～0.7

Hires. fix

trigger words: amagi rinne\(ensemble stars!\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

共通呪文:amagi rinne\(ensemble stars!\), red hair, short hair, blue eyes, hair between eyes, jewelry, earrings, ring

※hair between eyesは私服を再現しようと思う時に邪魔になります

Crazy:B衣装

出現率アップ呪文: jewelry, jacket, black pants, open jacket, chain, ring, white shirt, white jacket, Crazy:B jacket, Crazy:B Uniform

私服（ヘッドバンド）

出現率アップ呪文:headband, black shirt, pants

ラッキーチェイン（フィーチャースカウト燐音編衣装）

出現率アップ呪文:jacket, necklace, off shoulder, tank top, torn clothes, fur trim, black shirt, black pants, ring, fur-trimmed jacket, red belt, chain necklace

Flambé！衣装

出現率アップ呪文:gloves, hat, jewelry, jacket, earrings, black gloves, pants, fingerless gloves, collar, black jacket, torn pants, black pants, ring, peaked cap, single glove, torn pants, military cap

A dream style LoRA for Stable Diffusion webui.
一个梦幻风格的LoRA。

Some of my LoRA's stopped working on the latest Auto1111 version for some reason. Seems to be an issue with LoCon LoRA's. To fix it all you need to do is install this extension then you can use them as normal again!

Coco Bandicoot from Crash Bandicoot!

Use the tag "coco, furry". Optionally you can use "overalls, white shirt, goggles on head, flower in hair", for various parts of her outfit.

Works well with most models I have tested it with. Have fun!

::: 该页面为中国传统纹样loha的使用介绍，关于详细的训练过程，分层调试方法，可以到我们的wiki上去查看。https://aigc.ioclab.com/sd-showcase/chinese-ornament.html :::

中国传统纹样lora by IoC Lab

训练及数据：@shichen @ciaochaos

版本 介绍 2023.4.1 上传lora，介绍内容（本篇） 2023.4.4 更新训练方法（wiki待上传）

在图片中你可以看到有大约25个以上的版本，我挑选了其中最有特点的几个版本，你可以在version中挑选下载，或者前往huggingface下载全部的版本

https://huggingface.co/ioclab/ioc-lab-models/tree/main

使用方法：

在训练集的标注中，分为以下五个语段：

traditional chinese pattern

with circular symmetrical composition
with square symmetrical composition
with special-shaped symmetrical composition

red and blue color scheme

featuring horses and clouds
featuring flowers and branches

a flower in the center

所以你可以围绕着这五种固定句式进行扩充，如需要作为背景，可以调低权重，增加 1 girl

分层拆解：

在分层图片中可见，在高权重下，通过仅开启 midd，outd 等 lora 层的权重，可以做到相比于直接降低权重，更能保留纹样风格的人物背景图。

推荐模型版本介绍：

这里是一些效果比较特色的模型版本，但出图效果与参数，种子等都有关系，你可以尽情尝试，期待你的评论

特色模型版本

以下是一些特色模型版本：

• 4：特点是纹样更卡通化，色彩更缓和

• 7：特点是对关键词的还原更好，更符合数据集的倾向

• 10：特点是更对称，同时构图倾向会偏向于非中心对称构图

• 13：特点是相比10的颜色会更缓和，高权重下会出现第三种散布构图

• 15：特点是高饱和度和高横向对称构图

• 16：特点是高饱和度+散布构图

人物配合版本

以下是容易和人物配合的版本：

• loha01_old3: 更容易出卡通光影，构图和花纹装饰性强

• loha04_old3: 构图容易出全身，光影保留效果好

• loha06_old1: 对构图限制小，人物动作自然

• loha02_old2: 背景层次更多，对称性强

数据集标注及loha训练方法，见wiki链接

https://aigc.ioclab.com/sd-showcase/chinese-ornament.html

Chinese Ornament Wiki in English

::: This page is an introduction to the use of traditional Chinese pattern Lora, for detailed training processes and layer debugging methods, please refer to our wiki. https://aigc.ioclab.com/sd-showcase/chinese-ornament.html :::

Chinese traditional pattern Lora by IoC Lab

Training and Data: @shichen @ciaochaos

Version Description 2023.4.1 Uploaded Lora, Introduction Content (This Article) 2023.4.4 Updated Training Method (Wiki To Be Uploaded)

In the pictures, you can see more than 25 versions. I selected the most distinctive versions for you to download, or you can go to Huggingface to download all versions.

ioclab/IocLab_Models at main (huggingface.co)

How to use:

In the annotation of the training set, the following five language segments are divided:

traditional chinese pattern

with circular symmetrical composition
with square symmetrical composition
with special-shaped symmetrical composition

red and blue color scheme

featuring horses and clouds
featuring flowers and branches

a flower in the center

So you can expand around these five fixed sentence patterns. If you need to use it as a background, you can reduce the weight and add one girl.

Layered Decomposition:

As shown in the layered images, under high weight, by only turning on the weights of midd, outd, and other Lora layers, you can achieve a person-background image with a style more retained than directly reducing the weight.

Recommended Model Version Introduction:

Here are some model versions with distinctive effects, but the output effect is related to parameters, seeds, etc. You can try them out and we look forward to your comments.

Distinctive Model Versions

Here are some distinctive model versions:

• 4: The characteristic is that the pattern is more cartoonish and the color is more gentle.

• 7: The characteristic is that it better restores the keywords and is more in line with the tendency of the dataset.

• 10: The characteristic is that it is more symmetrical, and the composition tends to be non-centrally symmetrical.

• 13: The characteristic is that compared to 10, the color will be more gentle, and the third scattered composition will appear under high weight.

• 15: The characteristic is high saturation and high horizontal symmetry composition.

• 16: The characteristic is high saturation + scattered composition.

Character Matching Versions

The following are versions that are easy to match with characters:

• loha01_old3: More likely to produce cartoon light and shadow, with strong composition and decorative patterns.

• loha04_old3: The composition is easy to produce the whole body, and the effect of retaining light and shadow is good.

• loha06_old1: The restriction on composition is small, and the character's movements are natural.

• loha02_old2: More layers in the background, strong symmetry.

For dataset annotation and Lora training methods, please refer to the wiki link.

https://aigc.ioclab.com/sd-showcase/chinese-ornament.html

A Character LoCon for Ijichi Seika(伊地知 星歌) | Bocchi the Rock!

She may not be as popular as other characters in Bocchi the Rock, but I still like her.

• Recommeneded prompts: red eyes,long hair,ahoge,blonde hair,black choker,off shoulder,bangs (Check my sample pics for detail)

• Trigger word: red eyes, blonde hair

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

• This LoCon still have some small issues, I may fix it later.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

「貴方が今も、月の光を忘れぬように……」

• This is a lycoris model, it requires the locon extensions to be installed.

• The trigger words for alternative costumes are still under active testing and may not be stable

• Recommened prompts: 1girl, solo, chain necklace\(arcueid\), black pantyhose, white long sleeves turtleneck sweater, blue pleated skirt, looking at viewer, brown footwear, knees boots

• Weight 0.7-0.9 works best for me.

• CFG Scale: 5.5-7

Model based on Erkerut art style. Works best with MeinaHentai (v2 used for example images) at 0.9.

个人感觉八宝备老师画风前后期差异挺大，我选择的是偏后期的画风。

我使用anythingv4.5,权重0.5-0.9。最好使用Hires. Fix。

Tigger words使用Happoubi Jin，可以触发画风。

我顺便做了一下几个主要角色，你可以使用Happoubi Jin+角色名召唤她们。

可以召唤的角色有：蓮華Tsukioka Karen，燈露椎Dorothy，亜璃子Alice，篝ノ霧枝Kagarino Kirie，稲森はるInamori Haru，沢渡雫Sawatari Shizuku。其中Dorothy需要结合short hair使用，其他的使用名字即可。召唤角色时候推荐权重0.7-0.9。

I feel that there is a significant difference in the before and after style of Happoubi Jin's artwork, and I chose the processing style.

I used anythingv4.5, with a weight of 0.5-0.9. It is recommended to use Hires. Fix.

Trigger words: Happoubi Jin, which can trigger the art style.

I also made several main characters, and you can summon them using Happoubi Jin + character name. The characters that can be summoned are: Tsukioka Karen, Dorothy, Alice, Kagarino Kirie, Inamori Haru, and Sawatari Shizuku. Dorothy needs to be used with "short hair" to summon, while the others can be summoned using their names. It is recommended to use a weight of 0.7-0.9 when summoning characters.

Hitoimim Style

An Artstyle LoHa

V2b is out!

Important: I only recommend using this LoHa to generate upper body images for the best results.(我只推荐用这个LoHa生成上半身的图像以达到最好的效果.)

• Recommeneded prompts: simple background,upper body,cropped shoulders,hair ornament,ribbon,bow,jewelry (Please check my sample work for more details)

• Trigger word: you don't need any trigger word for this LoHa

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V4.5, Dan mix

• This LoHa may have some small issue, I will fix it later!

• How to use LoHa/LyCORIS? Please read instruction of a1111-sd-webui-locon.

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

When you don't just want bigger; you want more too. Example images generated from 1 set of 8 using AnythingV4.5

300+ Ayaka HQ images dataset

A Character LoCon for mao isara (衣更真緒) 【Ensemble Stars!】

Weight:0.4～0.7

Hires. fix

trigger words: isara mao\(ensemble stars!\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

Trickstar衣装（!時代）

出現率アップ呪文: plaid_red jacket, white shirt, red_plaid pants, black gloves, purple wristband, jewelry,

Trickstar衣装（!!時代）

出現率アップ呪文: plaid_red jacket, white shirt, black pants, black gloves, fingerless gloves, jewelry,

制服（冬）

出現率アップ呪文:hood, necktie, white shirt, open blazer, blue blazer, school uniform, plaid, pants,

髪下ろし（通常　hair ornament, hairclip, bangs pinned back,　と呪文指定しなくてもヘアピンと前髪をあげてしまいますのでおろしたい場合）

出現率アップ呪文:down hair, hair between eyes

ネガティブにhair ornament, hairclip, bangs pinned back,で多少は髪を下ろしている感じに出来ます

お世辞にも使い勝手が良いと言えないですが使っていただけたら幸いです。

weight: 0.6-0.8

recommend prompt: pauldrons

Benito Mussolini LoCon

IMPORTANT NOTES:

>This model is NOT meant to share or glorify any political view.

>Trigger word: "BenitoMussolini". Recommended weight: 0.7

>Trained on mostly b&w pictures, in order to get good results appropriate prompting and model is required.

>The best results achieved with realistic models (tested with Realistic Vision).

>Probably the easiest way is to just use inpaint.

>This time I’m testing LoCon, feel free to post your feedback.

Prompting

>Highly recommended to add this to your prompt:

Positive: in color

Negative: oversaturated, comics, fake, black and white, b&w, blurry, image noise, simple background, monochrome, (scar)

>Trained mostly on upper body pictures, might be hard to get full body.

>Negative prompts are powerful, use them to your advantage.

>I suggest using 768x768 resolution.

How to make a similar model

I know this model is far from being great, but if someone wants to know the workflow, here you go:

1. Find good images and crop them. 2. Upscale x2~x4 (in my case mostly R-ESRGAN). 3. Remove noise using any graphic software. 4. Scale down. 5. Train LoCon (using Google Colab, e.g. Kohya Trainer). 6. This time I’ve experimented with block weight adjusting (check out sd-webui-lora-block-weight and sd-webui-supermerger awesome tools for LoRa fine tuning)

I mostly use b&w photos, so following steps 2,3,4 leads to significant increase of quality and reduction of image noise.

Also:

>Why do you share such controversial models?

Mainly for fun, also because I can as it’s allowed and I’m not doing anything illegal.

>If you want to use my model anywhere feel free to do so, but it would be nice to put me in credits.

>Check out my models: Adolf Diffusion, Stalin Diffusion

>For any question you can contact me rafikxd123#7133

Note:

This is a LYCORIS/LoCon model, and you need to install a1111-sd-webui-locon extension to use this model.

Suggest weight:around 1

No trigger words,But suggest add chibi in prompt,When used to generate characters.

Based on novel-anime-full model.

(Not chi b i style)

Here is the Lora for against glass !

The trigger words are:

against_glass, breasts_on_glass

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Each trigger word is an independent concept and can be added as needed.

Weight 0.3-0.7

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.

Trigger words for V1.20:

positive: <lora:Mermaidv220:1>(mermaid, full body)

You can also try "monster girl", "pussy", "green scales" (or any other color WITH LIMITED results), "head fins"
LoHa is biased towards generating a "pussy" since the entire dataset had one! Use negative "pussy, nsfw" and "sfw" in positives for sfw gens!

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

Sample images generated with Grapefruit.

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.

It was trained on 75 512x512 images and 20 epochs using Animefull-final-pruned for the Training Model.

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

This is a LoHa model trained from 100 ノーコピーライトガール images and 100 flat color images. The tone is very close to Walden Blue,and that's where the name comes from.

You can use the LoHa to design hiphop girl,flat color design.

Weight 0.9-1.0 will perform better,also you can try new way to use,lol

2023/4/6 update

Upload Lycoris version. v1.0 version is conventional LoRA model.

Introduction

A conventional LoRA model of KDA Series, including 4 characters, Ahri, Akali, Evelynn and Kai'sa. 0.6~0.8 weights are good. The preview images are generated using 1.6 version (original) YesMix.

Usage

The universal trigger tag is 'k/da \(league of legends\)'.

• For Ahri, the trigger tag is 'ahri \(league of legends\)'. The key features are 'fox girl, fox ears, fox tail, blonde hair, choker, facial mark, whisker markings'.

• For Akali, the trigger tag is 'akali'. The key features are 'baseball cap, choker, cropped jacket, ponytail, mask (optional), kunai (optional)'.

• For Evelynn, the trigger tag is 'evelynn \(league of legends\)'. The key features are 'choker, crop top, feather trim, midriff, hair between eyes, purple hair, claws, sunglasses (optional)'.

• For Kai'sa, the trigger tag is 'kai'sa'. The key features are 'choker, double bun, hair bun'.

See the preview images for details. Btw, this model is trained with some images that contain the baddest outfits and prestige outfits. Therefore, images that contain the above outfits can be also generated using this model, theoretically.

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.

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.

Intro

This is a LoHa / LyCORIS model of a cute Japanese girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは可愛い日本人の女性の顔のLoHa / LyCORISモデルです。このモデルは、さまざまな実写の写真で訓練されましたが、実在の人物を描写しているわけではありません。仮想キャラクターと考えてください。

这是一个可爱的日本女孩脸部的LoHa / LyCORIS模型。该模型是基于各种实际照片进行训练的，但它并不描绘任何真实的人物。可以将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

-Keep only body-

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

-Keep only face-

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

-Keep only clothing (not tested)-

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

-Keep only pose (not tested)-

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

Here is the Lora for cowgirl position !

There are two types of cowgirl position, you can try the trigger words:

cowgirl_position, squatting_cowgirl_position

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Intro

This is a LoHa / LyCORIS model of a cute Asian girl face. Model was trained on various irl photos, however it does not depict any real person. Consider it a virtual character.

Trained on v1-5-pruned. Offset noise has been applied.

(Translation are all from chatGPT.)

これは、美しいアジアの女性の顔をモデルにしたLoHa / LyCORISモデルです。モデルはさまざまな実際の人物の写真で学習しましたが、実在の人物を描いているわけではありません。仮想キャラクターと考えてください。

这是一个可爱亚洲女孩面部的LoHa / LyCORIS模型。该模型通过训练各种真实人物照片而得，但并未描绘任何真实人物。因此请将其视为虚拟角色。

Settings

1 - Not necessary but highly recommended

To make this and any LoRA/LoHa more compatible, use this extension. I made some custom weight preset to seperate face from the rest of the contents. Copy and paste below lines into your LoRA block weight extension textbox:

このLoRA/LoHaと他のものをより互換性が高くするには、この拡張機能を使用してください。顔を他のコンテンツから分離するカスタムウェイトプリセットを作成しました。以下の行をコピーしてLoRAブロックウェイト拡張テキストボックスに貼り付けます。

为了使此模型及任何LoRA/LoHa更兼容，请使用此扩展。我制作了一些自定义权重预设，以将脸部与其他内容分开。将以下行复制并粘贴到您的LoRA块权重扩展文本框中：

----LyCORIS

-Keep only body-

LyCOBODY:1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1,1,1,1,1

LyCOBODY0.5:1,1,1,1,1,1,1,1,1,0.2,0.2,0.5,1,1,0,0,0,0.2,0,0,0.8,1,1,1,1

-Keep only face-

LyCOFACE:1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0

LyCOFACE0.5:1,0,0,0,0,0,0,0,0,0,0,0,0,0,0.2,0.5,0.8,1,1,1,0.2,0,0,0,0,0

-Keep only clothing (not tested)-

LyCOCLOTH:1,1,1,1,1,1,1,1,0,0.2,0.2,0.2,0,0,0,0,0.5,0.8,1,1,0.2,0,0,0,0,0

-Keep only pose (not tested)-

LyCOPOSE:1,0,0,0,0,0,0,0,0,0.2,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0

Like so. Remember to click Save Presets and Reload Tags. Then you use it like in the preview images, <lora:AG-yan:1:LyCOFACE>.

Reminder: As per the ReadMe of the extension, you cannot use the built-in presets (17 blocks) on a LyCORIS (26 blocks). Just keep in mind that you need to know clearly which it is that you apply onto.

If you don't want to use this extension, that's fine I guess, but things might got messed up pretty quick if you load more LoRA together.

このようにしてください。プリセットを保存してタグをリロードすることを忘れずに。そして、プレビュー画像のように使用します。 <lora:AG-yan:1:LyCOFACE>。

リマインダー：拡張機能のReadMeによると、ビルトインプリセット（17ブロック）をLyCORIS（26ブロック）で使用することはできません。適用するものがどちらかはっきりと把握しておく必要があることを念頭に置いてください。

この拡張機能を使用したくない場合は、それでも構いませんが、LoRAを一緒にロードするとすぐに混乱する可能性があります。

就像这样。记得点击保存预设并重新加载标签。然后您可以像在预览图像中一样使用它。

提醒：根据扩展的ReadMe，您不能在LyCORIS（26块）上使用内置预设（17块）。请牢记，在应用预设时需要清楚了解模型的类别。

如果您不想使用此扩展，我想这也没关系，但是如果同时加载更多LoRA，事情可能会很快变得混乱。

2 - Parameters

base res is 512*768

Sampler: DPM++ 2M Karras v2 (Basically the improved version of old 2M Karras)

Hires or img2img upscale is essential if you want full body portraits

upscale by 2 or higher.

Hires step: 36 or higher

Upscaler: R-ESRGAN 4x+ / Latent

Denoising strength: 0.5

Here is the Lora for after sex !

The trigger word are aftersex, cum_in_pussy, cumdrip

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Note: This is a Locon. Install the extension to get it to work. I've found them to do better when working on ideas that I want easily added to one of my LORA characters.

Another character changing Locon, this one specifically around taking your character and "bimbofying" them. It was trained on animefull_final with 63 images with the usual self-pruning and tagging. I like keeping my prompts simple for the first few pictures so that you can see it without a whole bunch of mumbo jumbo. Make your pictures better than mine though!

What does it do you ask? If you use this LORA without any other character LORA or descriptions, it'll create a someone with a thicc body, big breasts, and a relatively slutty outfit. I personally like combo'ing it with another LORA and at different variable to see different stages of transformation.

How to use it: The Locon strength best used is anywhere up to 1. Depends on how much effect you want it to have. Also the trigger word is:

BimboT - Only trigger word currently added.

Currently I have trained it to lean more towards bra/panty style outfits (though a usage of clothing trigger words should let it be swayed.) I've debated on making it lean slightly more towards skirts on a second version, so let me know what you think.

Extra, Extra!: Should be well trained and usable in most models. Images shown for example.

Picture Descriptions:

1-3: AnythingV3 using just this Locon.

4: AnythingV3 with my Shadow Sae Lora and this Locon.

5: AnythingV3 with my Sae Lora and this Locon.

6: AnythingV3 with my Hibana (Fire Force) Lora and this Locon.

7: AnythingV3 with my in progress Seraphina (Disgaea 5) Lora and this Locon.

8-9: AbyssOrange using just this Locon.

10: PerfectWorld using just this Locon.

What's next for this? Potentially a version that more balanced between the bra/panty combo and skirts when unprompted? Dunno if it's worth the effort though.

What's next that isn't this?

More character Loras. (Ghislaine from Mushoku Tensei, Conqueror Lilias from E7, Catherine from Catherine Full Body, Miku + Piscalat from Idaten Deities, Kuribayashi Shino from GATE, Rin Sohma from Fruits Basket, "Kasumi" from P5, Ann from P5, Ikumi Mito from Food Wars, Alice Nakiri from Food Wars, Chika Fujiwara from Kaguya-sama, Lulu from FF10, Momo from My Hero, Vivy from Vivy, Dominique de Sade from Vanitas, Dos from One Punch Man, and Seraphina from Disgaea 5.) And maybe Wilhelm van Astrea, my favorite Re:Zero husbando or a Prinny from Disgaea.

V2's of some of the previous Loras. Specifically make them more diversely acceptable with other models and LORA's.

I also have some more quite NSFW Locon's. Specifically, being pet play, being stepped on, brainwashing, corruptedv2, and one that might be too much for me to even upload here >.<.

And, uh, I want to make a few happy Locon's so I have an "angel-ify" your character one coming up. And a forest fairy one as well. And a soldier one. And a barbarian one.

Or just give me some random suggestion or priority opinions. That works too.

Tall dark matter choco cosmo caramel with dark choco sauce extra whip extra gravity

Tall dark matter choco plasma schwarzschild choco sauce with neutrino extra quasar

• CFG: 5.5 - 7

• Weight: 0.8 - 1.1

Features SN Chapayev from Azur Lane in her original costume & White Cavalier's Respite casual costume, trained on animefull-latest. See trigger words section for costume prompts.

Tested and works fine on high weight range up to 1.0, however finer details hadn't improved much from the earlier LoRA version.

Due to limited training data quality, it is highly recommended to apply with the following LoRA block weight:

1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0

to prevent style/image composition leakage.

A Character LoCon for Ijichi Nijika (伊地知 虹夏) | Bocchi the Rock! / 下北沢の大天使

[かき鳴らせ 交わるカルテット 革命を 成し遂げてみたいな]

• Recommeneded prompts: red eyes, side ponytail, sidelocks,school uniform, skirt,bangs

• Trigger word: red eyes, side ponytail

• Recommeneded Weight: 0.6-1 (You can adjust it by your perference.)

• CFG scale: 5.5-7

• Check point recommendation: Anything V5, DanMix

• How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.

• Some sample figures were generated with my Hitoimim Style - LoHa and Vivid Impactful Style - LoRA

If you like my work,please give me a like <3.

如果你喜欢我的模型请给我一个赞 <3.

もし、私のモデルが気に入ったら、「いいね！」をクリックしてください<3。

Another Loha trained to create women looking like Laura-Matsuda, it is not easy, but you can get some really near aproximations of this character.

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 honestly don't know if v1.20 or v2.20 is better, v2 have more tags for control tho.

Trigger words for V1.20:

<lora:CheekBulgev120:1>(cheek bulge, fellatio, penis, 1girl, 1boy)

You can try "deepthroat", "irrumatio", ":>=", "deepthroat", "cum in mouth", "saliva", "pov", "from above", "from side" (or no camera angle at all for some kind of front/side angle).
V2.20 also have on top of V1.20 tags: "head tilt", "finger to cheek:1.4", "penis grab", "lipstick on penis" and tagged for penis sizes but kinda just doesn't work.

I'm so happy with this that I couldn't stop generating with it!

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

Sample images generated from Grapefruit.

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 326 512x512 images scraped from the internet and 20 epochs using animefull_final_pruned.

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

Here is the Lora for female masturbation !

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Trained on anything v3.

Pictures are done with EasyNegative, kl-f8-anime2 VAE and various models.

Hikaru from (R)ape Gouhouka!!! (The name seems to avoid this LoCon to be published, so decided to tweak it a bit.)

• Trained with Anime (final-pruned) model.

• Comes in 4 versions; Anime-6 (Trained with 32 images), Extended-5 (Trained with 82 Images/5 epochs), Extended-6 (6 epochs), Manga style (6 epochs).

• Tip: You can make any combination of this LoCon's version to get interesting and high quality images; Put any "base" LoCon in the prompt with 0.5+ weights, and add another to Additional Networks with UNet weight of 0.2 and Text weight of 1.0 OR 0.2 weights in prompt with UNet weight of 0.5 and and Text weight of 0.7+. Also it works great with ALL, OUTD, and OUTALL LoRA weight blocks.

• Tip: Extended-5 and Extended-6 tend to get the character's arms up; You can avoid this by putting (arms up:1.21551) in the negative prompt.

Here is the Lora for bukkake !

The trigger word is bukkake, and you can control the cum by applying the following prompt:

cum_on_hair, cum_in_mouth, facial, cum_on_breasts, cum_on_body, cum_in_pussy

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

Here is the Lora for sexy pantyhose !

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

A little Lora without prentention able to do some Juri from street fighter. Trained On Never Ending Dreams, (NED)

Here is the Lora for peeing !

Below are the trigger words suggested,

pee, peeing, peeing_self

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

This LORA is Hakamada Hinata from RO-KYU-BU!

inital tags:

"nagatsuka saki, long_hair", "glasses, twin braids" or "goggles, forehead, hairband"

You can also mix and match.

main costume:

"gym uniform, bike shorts"

"basketball uniform, shorts"

You may want to use 7thLayerAbyss or AOM3A.

show you guts cool say what saikou daze.

Here is the Lora for POV from behind !

Hope you enjoy my work!

You are welcome to browser my other works if you are interested!

You are also welcome to leave your request for free and I will try to make it into a Lora if it is possible.

If you appreciate my works, you can support me via donation on ko-fi.com/dpp12

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

SDoom2.1

This LoRA was specifically designed with @fking 's model in mind.

Therefore I would advise you to please use fking_scifi_v2 | Stable Diffusion Checkpoint | Civitai

That should give you the best results!

This LORA is Hakamada Hinata from RO-KYU-BU!

inital tags:

"hakamada hinata, very long_hair, pink hair"

main costume:

"gym uniform, buruma"

"kimono"

You may want to use 7thLayerAbyss or AOM3A.

show you guts cool say what saikou daze.

This LORA is Tomoka Minato from RO-KYU-BU!

inital tags:

"minato tomoka, short hair, one side up, mole"

main costume:

"gym uniform, bike shorts"

"basketball uniform, shorts"

"school uniform"

You may want to use 7thLayerAbyss or AOM3A.

Thank you.

2023/4/6 update

Just update preview images.

2023/4/5 update

Upload Lycoris version (smaller file size). v1.0 version is conventional LoRA model.

Introduction

A conventional LoRA model trained with ~70 images by Hyulla. The trigger tag is ‘hyulla’. 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.

A Character LoCon for Hiiro Amagi (天城 一彩) 【Ensemble Stars!】

Weight:0.4～0.7

Hires. fix

trigger words: amagi hiiro\(ensemble stars!\)

※以下日本語で失礼します。

運要素が絡みますがぽい衣装を出現させることは確認しています。

ALKALOIDライブ衣装（出現率高め）

推奨呪文: uniform\(alkaloid_hiiro\), white pants, black jacket, fingerless gloves, hat

私服:秋〜冬（出現率高め）

推奨呪文: white jacket, black pants, open jacket, blue vest,

トランプ衣装（運要素強。ライブ衣装混同しがちですがそれっぽいのは数回確認）

推奨呪文: playingcards uniform, hat\(playingcards\), black jacket, black pants

ネガティブ:uniform\(alkaloid_hiiro\), white pants,

お世辞にも使い勝手が良いと言えないですが使っていただけたら幸いです。

Attention!

download Extension into sd-webui, and then use built-in lora system to run your model. LoHa Model supported

Not For Kohya-ss' Additional Network

• checkpoint: Anything v5

• trigger words: cat tail, cat ears

• recommond CFG Scale = 7.0

• recommond weight: 0.7~0.8

• recommond use hires.fix

training parameters

$network_module = "lycoris.kohya"
$resolution = "712,1024"
$batch_size = 3
$max_train_epoches = 8
$save_every_n_epochs = 1
$network_dim = 128 
$network_alpha = 32
$clip_skip = 2 
$lr = "1e-4"
$unet_lr = "1e-4"
$text_encoder_lr = "1e-5"
$lr_scheduler = "cosine_with_restarts" 
$lr_warmup_steps = 0 
$lr_restart_cycles = 1
$noise_offset = 0.05 
$keep_tokens = 3
$min_bucket_reso = 256
$max_bucket_reso = 1088
$use_lion = 1
$algo = "loha"
$conv_dim = 8
$conv_alpha = 1

This is the Vietnamese traditional dress for woman called Ao Dai.

Recommend Sampler is UniPC, CFG from 5 - 9.

Use hires fix for more than 800x800 image.

Please refer my prompt

This is the Vietnamese traditional dress for man called Ao Dai.

Recommend Sampler is UniPC, CFG from 5 - 7.

Use hires fix for more than 800x800 image.

Please refer my prompt

Trained by LoCON

Recommend Sampler is UniPC, CFG from 5 - 7.

Use hires fix for more than 800x800 image.

Please refer my prompt

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

LnF Aeros Lingerie

This is a spin-off of the LnF Aeros concept, but aimed towards Asian lingerie.

Have fun! Don't forget to like and review!

A simple Locon trained on "Never Ending Dream" (NED) to call Cammy-white, with a description of her.

Can have some trouble with the braids lenght or number. tested with strenght of 0.8

This is an EXTRACTION of 1.2 which isn't far off the OG one.

IT was done by Kappa_Neuro

Why don't you go TAKE A LOOK AT HIS AMAZING CONTENT?

THE MODEL THIS GOES WITH IS HERE: https://civitai.com/models/30363/cheesed-out-anime-backgrounds

KAPPA RECCOMENDS WITH LYCORIS EXTRACTIONS NOT TO USE A VAE.
THIS EXTRACTION LIKELY WORKS ON ANY MODEL AT WILL, AND WE FOR NOW HAVE YET TO TEST IT SO HAVE ONLY ACTUALLY USED AN IMAGE FROM 1.2 <3

HOW TO SUPPORT US:

We stream a lot of our testing on twitch: https://www.twitch.tv/duskfallcrew

Any chance you can spare a coffee or three? https://ko-fi.com/DUSKFALLcrew

Request image gens via our pixiv: https://www.pixiv.net/en/users/70748346

Hang with us on discord: https://discord.gg/Da7s8d3KJ7

If you'd like to support me and do more:
https://ko-fi.com/edgonkofi - commissions open!
If you're looking for a >>LoRA Making Tutorial<<

Oriental Lingerie

Very happy with my caftan and nuisette models, I use LoCon fusion design to bring you another KICK ASS lingerie model!

Download, like, and share!

This way to another great lingerie model!