Models

Introducing my versatile photorealistic model - the result of a rigorous testing process that blends various models to achieve the desired output. While I cannot recall all of the individual components used in its creation, I am immensely satisfied with the end result. This model incorporates several custom elements, adding an extra layer of uniqueness to its output.

One of the model's key strengths lies in its ability to effectively process textual inversions and LORA, providing accurate and detailed outputs. Additionally, the model requires minimal prompts, making it incredibly user-friendly and accessible.

VAE recommended: sd-vae-ft-mse-original.

It's a merge of some popular checkpoint based on Basil Korea to create 2D aesthetic

• Use hires fix

• Use DPM ++ SDE Karras

• Use ERSGAN 4X+Anime 6B

• Use 4 CFG

• Use Use 0.4 denoise

• Use 10 hires steps

• Clip Skip:2

• Use VAE

You can experiment it all you want, the guide above is basically my template to use the checkpoint

I have recently tried some STABLE DIFFUSION custom large model training for installation art. I found some random training materials from printerest, partly from UCL Bartlett RIBA Part2 unit8, partly from behance, and envisioned the result to be something like a deconstructionist and multi-material mix of conceptual models.

The training is just a purposeless and illogical experience.

Training model based on SD1.5

It can be used as it is, or it can be the basis for merging.

There is no problem if you use the usual option.

I recommend this much DPM++ SDE Karras, R-ESRGAN 4x+ Anime6B. There's no problem using anything else.

Basically, it's a non nsfw model, so I think sexual drawing is the worst to be honest.(Nude comes out well)

If you like the model, help me with some of the electricity bills -> https://yuribun500.fanbox.cc/

If you have any questions, please contact us via Pixiv or Twitter -> https://twitter.com/nzxt_cam

Thank you so much for the feedback and examples of your work! It's very motivating.

Do you like what I do? Feel free to buy me a coffee ☕

My models:

• Colorful

• Animatrix

new model merged.

nsfw possible.

sample prompt

positive -

RAW, 1girl, colorful, (masterpiece, best quality), {1$$cowboy shot|upper body},

(detailed skin:1.3, detailed face:1.3), dslr, realistic, {waving| arms behind head}, looking at viewer,

sharp focus, delicate, (korean beauty, korean mixed), (mature female:1.2),

(Glossy skin + Shiny skin:1.4), (sweaty:1.1),

(big breasts:1.2, thick thighs:1.1, narrow :1.1),

(black_hair, long hair, straight hair:1.3), (big_eyes, brown_eyes), red_eyeshadow, long_eyelashes, red lips,

eyes_smile, stud earrings,

(holographic iridescent metallic black latex swimsuit:1.4),

beach, day, sunlight,

negative -

(worst quality:1.4, low quality:1.4), (normal_quiality:1.4), (loli, child, infant, baby:1.3), ((muscular)), (mouth mask:1.3), wrong_face, topless, bottomless, facepaint, mouth mask, sunglasses, glasses,

hat, tattoo, extra arms, extra digits, multiple girls, multiple views, indoors,

With the tightening of the law, it may be prohibited to use non copyrighted images to train models in the future. Therefore, I found a method to train models using only one original image, which is derived from only one original image.

随着法律的收紧，未来可能会禁止使用非版权图像来训练模特。因此，我找到了一种只使用一张原始图像来训练模型的方法，这个模型是从一张原始图片中训练出来的。

2023/4/20 更新了 yden-V3

由于V3版本和前两个版本的融合配方完全不同，因此对构图的改变是很大的，但相比前两个版本，V3版本对于手部有良好的表现，尤其是在高steps的条件下。

所有示例图像均使用txt2img功能生成，所使用的VAE为原生VAE。

Due to my poor English proficiency, I used ChatGPT to translate Chinese into English. Please bear with me.

The following is the translated description:

Due to the completely different fusion formula between V3 version and the previous two versions, there are significant changes in composition. However, compared to the previous two versions, V3 version has good performance for hands, especially under high steps conditions.

All example images were generated using txt2img function and native VAE was used.

————————————————————————————————

2023/3/25 更新了 yden-V2

作为yden的新版本，它和V1一样都是为了生成动漫风格的图片而被创造的模型，且同样已被压缩至fp16，它可以在保持原来的效果和速度的同时减少模型大小。

相较于V1版本，V2版本生成的图片画面会更加精细，且构图有比较大的改变。

在参数设置方面和V1版本没有任何区别，另外在此强烈推荐使用Hires.fix功能，以便更好的发挥这个模型的潜力。

————————————————————————————————

以下是yden-V1的介绍

这是一个用来生成动漫风格的模型，我已将该模型压缩成fp16，它可以在保持原来的效果和速度的同时减少模型大小。

示例图像均由novelai原版VAE生成。

以下是推荐参数：

Sampler：DPM++ 2M Karras

steps：22

CFG Scale：7

Negative prompt：(worst quality, low quality:1.4)

Hires.fix：R-ESRGAN 4x+ Anime6B

Hires steps：10

Denoising strength：0.35-0.6

Upscale by：1.5-2

fly, eat mushrooms, travel to space, cure, help

☕You can buy me coffee

If you support me, I will make many AI male models and try to improve them more beautifully.

***

Color Boys 2D model is a simple and cute style, suitable for drawing male and female characters together.

It might be a little hard to paint only men.

Ddetailer(Inpaint face) was used in this sample image.

This model is mix of Kenshi by Luna and AnythingV3 by Yuno779.

It was my first model merge, and I was quite suprised. It does consistenly great and works pretty well with NAI-based LoRAs.

Also you can check out AIBooru, a website on which you can share your AI-generated works with others.

Best VAE: vae-ft-mse-840000-ema-pruned (highres fix or sd upscaling is desirable to use)
Best upscaler: 4x-UltraSharp

※このモデルにはいくつかのバリエーションがあることを見逃さないでください

　Note:Don't miss that there are several variations of this model!

caulkinumV1_S series for Realistic characters.

Note: If you like it, please give it a STAR MARKS !

,

The caulkinumV1_S series is a model devised to depict realistic and beautiful people and landscapes.

The primary goal of the "S8" series is to depict beautiful people.

On the other hand, "S4FM" selected the ones that depict people who are not necessarily beautiful, such as celebrities.

You can choose either according to your preference, or of course you can try both.

,

The caulkinum series was born from the hierarchical merge of the underlying Silicon28 and Elysium series.

The light expression is especially beautifully rendered and will surely bring you satisfactory results.

I will be very happy if someone likes these models.

,

Below are some samples to compare with other models.

I am generally satisfied with the results, although I do not always get it right.

Note: ) Underscores at the beginning of the checkpoint names are given in AUTOMATIC1111 for sorting purposes and have no particular meaning.

,

,

caulkinumV1_S シリーズ（写実的なキャラクター用）

※ 気に入っていただけたならぜひ★をつけてください！

,

caulkinumV1_S シリーズは写実的で美しい人物や風景を描くために工夫したモデルです。

「S8」は美麗な人物を描き出すことを第一の目標としています。

かたや「S4FM」はあえてセレブのような美人とは限らない人物が描かれる物を選別しました。

あなたは好みに応じてどちらでも選ぶことが出来ますし、もちろん両方試してみることもできます。

,

caulkinumシリーズは、根底にSilicon28とElysiumシリーズの階層マージを元に産まれました。

特に光の表現が美しく描写され、きっと満足のいく結果をもたらすでしょう。

これらのモデルを誰かが気に入ってくれたなら私はとても幸せです。

,

下記に他のモデルとの比較サンプルを掲載しておきます。

いつも必ずうまく錬成できるとは限りませんが、私はおおむね満足しています。

注釈 ) チェックポイント名先頭のアンダースコアはAUTOMATIC1111で並べ替えのために付与したもので特に意味はありません

,

SamplePrompt (サンプルプロンプト)

cute girl photo realistic

Negative prompt: 
title text, signature, watermark, username, artist name, an489 tongue out sharp_teeth micro mini skirt out of focus cropped skirt lift holding skirt curtsey EasyNegative

Steps: 15, Sampler: DPM++ 2M Karras, CFG scale: 5, Seed: 3777540191, Size: 384x512, Model hash: 38cc90a8b7, Model: caulkinumV1_S8, Clip skip: 2, ENSD: 31337

,

,

caulkinum Series Checkpoint (caulkinumシリーズ チェックポイント)

,

caulkinumV2 series for Anime Style

• https://civitai.com/models/32537/caulkinumv2ar-for-anime-style

,

caulkinumV1_RL series for semi Realistic Style

• https://civitai.com/models/32958/caulkinumv1rl-for-25d-style

,

caulkinumV1_S series for Realistic Style

• https://civitai.com/models/45689/caulkinumv1s-series-for-realistic-style

,

,

Donation (寄付)

Coffee will increase work efficiency and motivation.

コーヒーは作業効率を高めモチベーションをアップさせるでしょう。

https://ko-fi.com/489489

,

Thank you for reading to the end.

I hope one of you will use it.

,

最後まで読んでくれてありがとうございます。

使用してくれるか方が一人でもいることを祈っています。

リアル系マージモデルです。日本人を始めとするアジア系の再現ができるように調整しています。特にjapanese doll likenessとの親和性を意識しています。

このマージモデルを公開するにあたり、使用したモデルの製作者の皆様に感謝申し上げます。

This is a realistic merge model. It is adjusted to reproduce Japanese and other Asian descent. We are especially conscious of the affinity with japanese doll likeness.

We would like to thank the creators of the models we used for making this merge model available to the public.

マージ使用モデル(Merge use model)

MUSE_v1

https://civitai.com/models/13564/musev1

Colorful

https://civitai.com/models/7279/colorful

majicMIX realistic

https://civitai.com/models/43331/majicmix-realistic

※このモデルにはいくつかのバリエーションがあることを見逃さないでください

　Note:Don't miss that there are several variations of this model!

,

caulkinumV2 series for Anime characters.

Note: If you like it, please give it a STAR MARKS !

,

The caulkinumV2_AR series is a model designed to beautifully portray anime characters and light.

This checkpoint was spawned accidentally through multiple iterations of hierarchical merging.

Therefore, the detailed recipe is already lost.

,

However, there is no doubt that Silicon28 is at the root of it and that it has inherited the influence of ElysiumAnime_V2.

If someone likes it I will be very happy.

,

The caulkinumV2 series has several variations.

"AR" is the first model, which is a modern anime/game style.

• https://civitai.com/models/32537?modelVersionId=38959

"AR2" is the normal evolution of the first model, and it can create better illustrations.

• https://civitai.com/models/32537?modelVersionId=48531

"A4FS" has a dramatic style, which is more in line with recent RPG games.

• https://civitai.com/models/32537?modelVersionId=48630

"ARCT" is an anime-style painting that makes the characters look young.

• https://civitai.com/models/32537?modelVersionId=48791

"ARc" has the same anime style, but is a bit more difficult to handle and a bit outlandish.

• https://civitai.com/models/32537?modelVersionId=48898

"NCA" is a little bit special and unique, but it was created to depict a slime girl.

• https://civitai.com/models/32537?modelVersionId=50350

,

A sample comparison is provided below.

It does not always work, but I am generally satisfied with the results.

Note: Underscores at the beginning of checkpoint names are assigned by AUTOMATIC1111 for sorting purposes and have no particular meaning.

,

If the samples are too hard to see and judge, you can dl and try them out.

Then post and show us your great results!

Or you can use ctrl+wheel.

,

,

caulkinumV2（アニメキャラクター用）

※ 気に入っていただけたならぜひ★をつけてください！

,

caulkinumV2_ARシリーズは、アニメのキャラクターと光の表現を美しく描くために考案したモデルです。

このチェックポイントは複数回の階層マージを繰り返して偶発的に産まれました。

そのため細かいレシピはもう失われています。

,

しかしSilicon28が根底にあること、ElysiumAnime_V2の影響を受け継いでいることは間違いありません。

誰かが気に入ってくれたなら私はとても幸せでしょう。

,

caulkinumV2シリーズはいくつかのバリエーションがあります。

「AR」が最初のモデルでモダンなアニメやゲームの画風です

• https://civitai.com/models/32537?modelVersionId=48531

「AR2」は正常進化版でより良いイラストを錬成できます

• https://civitai.com/models/32537?modelVersionId=48531

「A4FS」は劇画調で近年のRPGゲームらしさが表現できます

• https://civitai.com/models/32537?modelVersionId=48630

「ARCT」はアニメ風の塗りでキャラクターは幼くなるでしょう

• https://civitai.com/models/32537?modelVersionId=48791

「ARc」も同様にアニメの画風ですが、やや扱いづらく突飛なところがあります

• https://civitai.com/models/32537?modelVersionId=48898

「NCA」は少し特殊で独特の画風ですが、スライム娘を描くために作りました。

• https://civitai.com/models/32537?modelVersionId=50350

,

下記に比較サンプルを掲載しておきます。

必ずしもうまくいくとは限りませんが、私はおおむね満足しています。

注釈 ) チェックポイント名先頭のアンダースコアはAUTOMATIC1111で並べ替えのために付与したもので特に意味はありません

,

サンプルが見えづらく判断に困る場合、あなたはDLして試用することが可能です。

そして投稿してあなたの素晴らしい成果を見せてください！

もしくは ctrl+wheel を使う方法もあります。

,

※ You can download the original image at HuggingFace

　(HuggingFaceで元画像をダウンロードできます)

- https://huggingface.co/Shiyaku/caulkinumCheckPoint/blob/main/README.md

,

,

,

caulkinum Series Checkpoint (caulkinumシリーズ チェックポイント)

caulkinumV2 for Anime Style

• https://civitai.com/models/32537/caulkinumv2ar-for-anime-style

caulkinumV1_RL for semi Realistic Style

• https://civitai.com/models/32958/caulkinumv1rl-for-25d-style

caulkinumV1_BR3 for Realistic Style

• https://civitai.com/models/45689/caulkinumv1s-series-for-realistic-style

,

,

Donation (寄付)

Coffee will increase work efficiency and motivation.

コーヒーは作業効率を高めモチベーションをアップさせるでしょう。

https://ko-fi.com/489489

,

Thank you for reading to the end.

I hope one of you will use it.

,

最後まで読んでくれてありがとうございます。

使用してくれるか方が一人でもいることを祈っています。

Moving to Civitai from hugging face you can find the rest of my models on my hugging face as i move them to Loras to Civitai as Loras didn't exist when I made those

https://huggingface.co/scriche

AIOMonsterGirlsV3 is very good at Lamias and decent at everything else I over trained Lamias a lot and under trained the others one (especially Scylla). but still generates good monster girls i refined the dataset alot so i wont be able to generate all MGE girls nearly as well sorry but its for the quality

I know i said i would make LORAs for all the models but.... I suck. At making loras that is, so im just going to stick to what i know and make checkpoints and extract them :). Ill upload V3 Loras soon

Ive changed up the style of AIO for V3 as i really like the realistic shadow but still 2d look and i like my monster girls 2D :). This version is also way better at NSFW generation than the last two so go crazy. For the generated images im using a combinations of 2 extensions: Dynamic Threshholding and Detection Detailer the former is used for high detail without weird artifacts (usefull for the Lamis scales) and the ladder is great for monster girls as most of the detail go into the monster girl and DD automatically in-paints the face at high detail and fixes that almost good generation

The LoRA/LoCON version is a LoRA not a checkpoint

Feel free to comment what other monster girls you want in for the next version as most monster girls from Monster Girl Encyclopedia are somewhat trained but would need more training to get the quality i want. Have fun generating

リアル系マージモデルです。日本人を始めとするアジア系の再現ができるように調整しています。特にjapanese doll likenessとの親和性を意識しています。

このマージモデルを公開するにあたり、使用したモデルの製作者の皆様に感謝申し上げます。

This is a realistic merge model. It is adjusted to reproduce Japanese and other Asian descent. We are especially conscious of the affinity with japanese doll likeness.

We would like to thank the creators of the models we used for making this merge model available to the public.

マージ使用モデル(Merge use model)

DreamShaper

https://civitai.com/models/4384/dreamshaper

NeverEnding Dream (NED)

https://civitai.com/models/10028/neverending-dream-ned

MUSE_v1

https://civitai.com/models/13564/musev1

Colorful

https://civitai.com/models/7279/colorful

The purpose of this model is to help you to create realistic images with photographic quality.

You just need to write to write "ABYR23 person" in order to call the model, try to use camera prompts for better results.

This model is free, please share and give some credit.

This model is for producing toon-like anime images, but it is not based on toon/anime models. This model is based on the photorealistic chilled-regeneric v2 model, and then transformed to toon-like one.

If you use my model "CityEdgeMix", you may notice that same prompt/seed as CityEdgeMix also produces very similar images as CityEdge_ToonMix. CityEdge_ToonMix is based on CityEdgeMix, which is based on chilled-regereric v2, merged some anime/toon models with low coefficients to expand desctiption variations, and then transoformed to a toon-like model.

Both SFW and NSFW is capable. VAE is not needed but you can try some.

アニメ調の出力を得るためのモデルですが、アニメ系のモデルからではなく実写系のモデル（chilled-regereric v2）から出発し、最後にアニメ調のタッチに変換して作成したモデルです。

公開している実写系（chilled-regereric v2ベース）のモデル「CityEdgeMix」に、アニメ系の出力のバリエーションを持たせるためにいくつかのアニメ系モデルを軽くマージし、その後アニメ系の出力モデルに変換しています。ですので、CityEdgeMixと同じプロンプトやシード値を使うと、よく似た出力を得ることができるはずです。

エロ・非エロどちらでも可。VAEは焼き込み済ですが彩度は浅めなので別のものも試してみていいと思います。

MIX CKPT:

Analog Diffusion+

vae-ft-mse-840000-ema-pruned

e other ckpt

This model is more likely to give a realistic impression and display a better shade than the previous version, besides that this model also provides a more detailed background and, The drawback may be that this version of the model lacks understanding of short prompts, but let's see what you can create using this one. (>.0)

Something that I've been throwing together over time and I think I've come to something stable enough to share with others. I can't remember everything I've put into it.

Has some good performance as low as 27 steps and I personally prefer DPM++ 2S a Karras and DPM++ 2M Karras, though it performs fairly well with all samplers.

The components in no particular order are: AnyHentai1.7, Bastard v4 Anime, Clockwork Oranges, Corneos, DHClassic, Digitals4wed0ff, DH SuperCute, ExpMix, Kazuki Merge, Moyu, Hassaku, DeGradeRev, Lazy Amateur Mix, URPM 1.5, SirenMix

Hello, the model was created as an artistic style, the model can do almost anything, the main thing is to follow the promt, hands and eyes looks good for the most cases

Model Information:

This model is generally designed for portraits and full-length anime style photos. Fantastic landscapes are quite decent. And it doesn't require kilometer-long queries to get a high-quality result.

Recommend: DPM++2M Karras, Clip skip 2 Sampler, Steps: 25-35+

This model would not have come out without XpucT's help, which made Deliberate

I hope you like it, thanks for the feedback

In this model, the main goal is to create the most realistic visuals possible. It has been trained with visuals selected from 100s of amateur photographs.

April 19 2023 Update:

First of all, thank you so much for 1,000 likes (and counting)! I had no idea this model would do so well. I just found myself using it all the time and decided to share it. It makes me happy that so many people enjoy it! And, I hope this encourages you to share your ai assets with the community as well!

Also, I added version 1.0 Baked Vae. It's the same as the original version, just includes the anything v4 vae. I was getting consistent results with this baked version, and decided to share it with the community! Can't wait to see your images!

A model I merged that I just keep coming back to, so I decided to share it.

A 50/50 blend of iCoMix v1 and Babes v1.1. Recommended you use a vae. I use anything_v4, which is available for download here.

EasyNegative also works very well with this merge, and minimal negative prompts tend to produce nice results.

I'm already working on some other versions that incorporate new versions of iCoMix and Babes.

Please share your images! I'd love to see what you create.

Thanks to lostdog and alexds9 for their terrific models.

お兄ちゃんはおしまい！の緒山まひろちゃんのボイスを学習したものです。

第一話・第二話のセリフでうめき声や叫び、残響の強いものを除去して学習させました。

誰でもまひろちゃんの声で喋ったり歌ったりできます。

サンプルのようにね！

"Clam" is named from using modes' name.

It has better performance in txt2img when using positive prompts to describe what you want without any quality tags, and you can use EasyNegative as negative prompt.

Ver.1.1 has more details. Compared to Ver.1.0, it is too similar to be a Ver.2.0.

hf:https://huggingface.co/Johnhex/Clam

How to Use

Here is the parameter recommended to use in stable diffusion.

VAE: animevae.pt [f458b5c6]

Sampler: DPM++ 2M Karras or DPM++ 2S a Karras

Step: 15~30

Hires Upscale: 2,Latent Denoising:0.5~0.8

CFG Scale: 7~15

Use Models

Sorry that I forgot to record using models' weight.

• Color Box Model [a80fee84]

• Ligne Claire Anime Diffusion V1 [ee5b91da]

• Anything-V3.0-pruned [2700c435]

• tmndMix-tmndMixPlus [65b8d0b2]

• momoko-e [a2a802b2]

V1: "This is a merged model of:

CrossKemono

Lawlas's Yiffymix

The original models are very good, SO I mixed them together!

Like the original models it can produce SFW and NSFW images"

V1 OJ added: "This is the same as V1 but with a bit of AM3O1 added; Hanse the name."

Recipes!

V1: Lawlas:55% CrossKemono:45%

OJ: V1

No baked VAE.
The recommended VAE is "vae-ft-mse-840000-ema-pruned.ckpt".
If you try it and make a good one, I would be happy to have it uploaded here!
The merging source is listed in each version detail.

VAEは焼いてません。
おすすめのVAEは"vae-ft-mse-840000-ema-pruned.ckpt"です。
使ってみて、いいのができたら、ぜひここにアップしてもらえると嬉しいです！

マージ元は各バージョン詳細に記載しています。