sexy pose

defacta3thとcouterfeitV3を単純マージしたモデルです。

This model is a simple merge of defacta3th and counterfeitV3.

defacta3thの表現力とcounterfeitV3のプロンプトの忠実さと画風を組み合わせました。

It combines the expressive power of defacta3th with the prompt fidelity and painting style of counterfeitV3.

It's a realistic version of the fantasticmix

↓↓↓ fantasticmix2.5D ↓↓↓

https://civitai.com/models/20632/fantasticmix25d

An example is using dynamic thresholding. Reduce the CFG scale to 7 to 10 when not in use

VAE : vae-ft-mse-840000-ema-pruned.ckpt

chineseWeirdness + 526mixV145 + revAnimatedv122，

配合LoRA ConstructionyardAIV3很好玩

扩展了一下诡异风格，人脸还是容易崩

铭刻，漫步于樱花丛中的时光。这里是来自于galgame《樱之刻》的原画级大模型。

Time of walking among the cherry blossoms. Here is the original painting-level model from galgame Sakuranotoki.

特点是干净又卫生通透，眼眸与头发丝的高光非常细腻。

This model features clean, transparent, very accurate highlights on the eyes and hair.

使用时，以下配置可能比较有用（其他的用法待开发，欢迎尝试和分享）：

When using it, the following configuration may be more useful (other usage methods are to be developed, welcome to try and share):

DPM++ 2M Karras/DPM++ SDE Karras

Hires.fix/BSRGAN

CFG scale=6~9

ChikMix_v3 now release!

Blend with braBeautifulRealistic_v40

https://civitai.com/models/25494

Feature:

2.5D K-doll style focus

Super simple prompts focus

Wide nsfw content Support

No VAE need

Check more image I made: https://twitter.com/xxxholic6666

Recommended embeddings:

bad-artist: https://huggingface.co/NiXXerHATTER59/bad-artist

pure-ero-face: https://civitai.com/models/4514/pure-eros-face

For people who cant reach the sample images results:

Use Hires. fix with R-ESRGAN 4x+ Anime6B in 2x upscale, set Denoising strength as 0.3~0.4, like 640*960 to 1280*1920.

Before & After Hires. fix

マージモデル

HASSAKU

ANYHENTAI

fitCorder - 1.3c - fC

by fitCorder

This checkpoint is a fine tune focuses on female bodybuilding. Keywords are not required. Works well with the Lora's i've uploaded.

Anime+other realistic Anime

https://civitai.com/models/53862/meganimevae

https://civitai.com/models/52506/photo-and-anime-v5

https://civitai.com/models/53316/magazine-mixard

https://civitai.com/models/52441/megaanimerv5

https://civitai.com/models/48265/dream-real-comic

<lora:koreanDollLikeness:1>

这是我日常使用的融合模型。它有：

1. 非常好的光影

2. 比较好的手部

3. 简单的用色，在使用画风LoRA的时候不会冲突、产生杂乱的颜色

4. 非常少的污染，尽量减少提示词中没有的物品

我认为基础模型需要简单、少污染、会画各种概念。这样使用人物、画风LoRA的时候的可以更加好地还原。

建议负面：

(worst quality, low quality:1.4)

我的一些画风LoRA融合都是在这个模型上调试的，有兴趣也可以尝试一下：

https://civitai.com/models/53642/silvermoonmix01-or-delicate-skin-and-hair-texture

https://civitai.com/models/50232/cute-gothic-fashion

This is my everyday use model. it is aimed to:

1. Getting great light effects

2. Improving hands

3. Being simple in coloring, so that it won't conflicts with style LoRAs

4. Minimize objects that are not in prompts

I believe a good model should be simple, accurate, knowing many concepts. These are keys to make other LoRAs work well.

Suggested negative prompts:

(worst quality, low quality:1.4)

My style LoRA mixes are fine tuned on this model. Appreciate if you can give them a try:

https://civitai.com/models/53642/silvermoonmix01-or-delicate-skin-and-hair-texture

https://civitai.com/models/50232/cute-gothic-fashion

<lora:koreanDollLikeness_v15:1>

Model created from 4 other models

((masterpiece)),
 (best quality), (detailed), (1 girl),   
<lora:RanSemRikoIchinose_v20:0.8>,cowboy shot,red dress,long 
hair, black high heels,open cleavage

The model merge has many costs besides electricity. With your support, we can continue to develop them.

モデルマージは、電気代以外にも多くのコストがかかります。皆さんの応援があれば、これからも開発を続けることができます。

https://ko-fi.com/breakd

I have merged this model with a lot of background and detail.

This model series is expected to produce more realistic-looking output than BreakDomain.

背景や細部の緻密な描写にこだわってマージをしています。

このモデルシリーズではBreakDomainよりもリアルっぽい出力が期待できます。

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！

Future updates will be made to this model.

今後の更新はこのモデルを行っていきます。

Add a ❤️ to receive future updates.

The model has been carefully optimized for stability and reliability, and its checkpoint is regularly updated to ensure that it incorporates the latest advances in deep learning research. This means that you can trust the model to produce accurate results time and time again, making it an ideal tool for artists, animators, and anime enthusiasts alike.

So if you're looking for a powerful and dependable anime model to help you create stunning art and animations, look no further than the "Anime Model OtakuReliableEnable (AMORE)" anime model.

I have poured my soul into this materpiece. This is a mix of over 50 MODELS. If you appreciate my work, kindly consider supporting me:

• https://www.buymeacoffee.com/LyloGuumy

As a creator, I have always been passionate about creating digital art that blends realism with a touch of anime. That's why I'm excited to introduce you to my latest project: IceRealistic, a unique simulation model that generates stunningly realistic images of ice and snow, with a twist - my model features beautifully rendered anime girls that bring a new level of charm and elegance to the art.

IceRealistic is perfect for anyone looking to create game assets, architectural visualizations, or scientific research projects that require realistic renders. With my model, you can create stunningly detailed images that will capture the imagination of your audience, with the added bonus of a touch of anime charm that will make your work stand out from the crowd.

But to continue creating and improving my model, I need your help. Donating to my ko-fi page will allow me to invest more time and resources into my work, allowing me to create even more stunning and realistic images for you to enjoy. As a token of my appreciation, I will provide exclusive content and perks to all of my donors, including early access to new updates and personalized tutorials.

So if you're a fan of anime and love seeing digital art come to life, please consider donating to my ko-fi page. With your support, I can continue to create amazing art that captures the beauty of the natural world, with a touch of anime magic. Thank you for your support!

FAQ:

- Best Resolutions without HiResFix Requirement

I personally believe my model works best at either 512x832 and 576x768

- Recommended Samplers

DPM++ SDE Karras at 28 steps or DM++ 2M at 50 steps

- Recommended Steps

For SDE: 28 steps with 8 highres

For 2M: 50 steps with 15 highres

- Recommended CFG

Anything between 7 and 10.5

- Restore Faces Possible (Yes/No)

If it works for you sure, I personally haven't used it.

- Recommended VAEs

kl-f8-anime2 if you want vibrant colors, alternatively any VAE is good.

- Recommended HiResFixes Samplers

4x-UltraSharp is the best IMO

- Bonus

Look up Detection Detailer for better faces! My favourite~

Hentai Core_v1

Trained on 42 sex position with normally at least 30 images each one. I will update tomorrow for better hands consistency!!

Jesus this took me a while but is presentable now, i will continue training this fot improvement in the next days. I also have a dataset of 22 artists ready but the results are meh, if i get it to work i will update this model or make other.

If you like you can buy me a coffee https://www.buymeacoffee.com/Norian01

Good for train and use of loras i suppose, i just used one of Aqua and worked well with a weight of 0.4

I used a fox girl in the sample images to show the flexibility of the model, so you see it isnt overcooked

I will upload a Lora extracted version of this model, check that in my profile or in this page. That Lora only works with anime models, even RevAnimated works weird https://civitai.com/models/54901?modelVersionId=59294

Specifications of training:

1280 images

10 repeats

8 epoch

2 batch size

base model: AnyHentay_v2.0 https://civitai.com/models/5706/anyhentai

merged low ratio with grapefruit because lack of good style https://civitai.com/models/24383/grapefruit-hentai-model

List of concepts with activation words with possible situations

!! left are just names, right are activation words + possible situations like pov, from side, standing, etc !!

USE CLIP SKIP 2!!!

• aftersex: aftersex, ass, cum in pussy, cum in face, fellatio, missionary, buttjob

• breast grab guided: guided breast grab, pov, guiding hand, breast grab, holding another's wrist

• breast grab pov: breast grab, pov

• own breast grab: grabbing own breast

• cheek bulge: cheek bulge:1.4, fellatio

• cowgirl position, cowgirl position, pov

• reverse cowgirl position: reverse cowgirl position, pov

• squatting cowgirl position: squatting cowgirl position, pov

• doggystyle: doggystile, pov, from side, standing

• doggystyle+fellatio: doggystyle, fellatio

• fellatio: fellatio, pov, from side

• fingering: fingering, vaginal, anal

• masturbation: masturbation

• imminent penetration: imminent penetration, missionary, cowgirl position, squatting cowgirl position

• Jack'O Challenge: Jack'O Challenge, sex from behind, deep penetration

• licking penis: licking penis, tongue, open mouth

• lying, leg up: lying, sex, leg lift, lef up, on side

• mating press: mating press

• missionary: missionary, pov, legs up

• paizuri: paizuri, penis

• piledrive: piledrive position

• prone bone: prone bone, from side

• suspended congress: suspended congress

• reverse suspended congress: reverse suspended congress

• standing_leg_up: standing, sex, split legs, leg lift

• triple fellatio: triple fellatio, 3girls

• x-ray: x-ray cervix, cross-section, internal cumshot, cum

• x-ray fellatio: x-ray mouth, fellatio, cross-section, from side, cum

• Animation sheet (work in progress) will give specification in next days

  You probably will need (animation sheet) in negative prompt

中规中矩的基础模组，搭配我的LORA用起来也是中规中矩。

如果不知道用什么模组，就用这个吧。

在拟真的基础上，对于2.5D的理解也很好。

The basic module is quite satisfactory, and it is also quite satisfactory when used with my LORA.

If you don't know what module to use, try this.

On the basis of realistic, the generation of 2.5D is also fine.

• Vae 840000

• Magazine Mix here: https://civitai.com/models/53862/meganimevae

• Model1.5 realistic

• Civitai Shortcut

Beautiful!

• https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111

This is a mix of one of my previous mix and AOM3.

Works well with characters and backgrounds!

Share your effects and enjoy!

Tips:

I personally use these settings
using EasyNegative embedding is recommended

This model was put together to make anime like stuff
Clip = 1

Try it out here! https://mobians.ai/

Join the discord for updates, share generated-images, just want to chat or if you want to contribute to helping perfect the model! https://discord.gg/KfAuP72uCd

USE:

The model uses Danbooru tags similar to anime models. I would not recommend just generating a character with no follow-up prompts as the results won't look very good. A good negative prompt also helps greatly.

Limitations:

Some characters generate less consistently. This problem will hopefully be resolved in the full release of V3 as I gather more training data.

Versions: LORA SD1.5 | LORA SD 2.1-768 | LYCORIS SD2.1-768

Beta1 sub models, models can be conditioned to provide additional flavours.

• Zenkai Ultra Hermit released 30 April 2023
  variation weights using my oldest successful recipe

• Zenkai Ultra Shift TBA
  variation using a new recipe

• Zenkai Ultra Instinct TBA
  variation using newest recipe

JDZ Beta1 - 22 April 2023
Beta1 is released for testing and evaluation. While convergence was reached, there are 230 styles/concepts contained within. This was trained on my original datasets used to create many CKPT / LORA already released on Civit. The Training Data contains combined instructions on making the most from this new versatile model format.

UPDATE - added new "combined-instructions.zip" now includes Instructions.txt and wildcards.zip. Wildcards are now included for use with Dynamic prompts.

JDZ-CaptionList.txt - ~5000 Example prompts used in captioning the 230 sets
JDZ-StyleTriggers.txt - ~230 "Trigger words" used for each style

JDZenkai-GenPrompts2200.txt - ~4200 "GPT2 Generated Prompts" trained on the Caption List

(Above pictured) Wildcard files now included

• Future releases will be trained to Epoch 40 (~200,000 steps)

• The Beta1 model is currently trained to Epoch 19 (~97000 steps.

• There are many ways to use this model as it contained all my CKPT styles in one model.
  

1. You do not need to use Triggers and can prompt normally.

2. You can prompt using words listed in DJZ-Styletriggers.txt to anchor to a specific style, or use multiple triggers with attention weighting to control the output.

3. You can use one or more prompts as shown in the DJZ-CaptionList.txt.

4. You can use the DJZ-CaptionList.txt and/or JDZenkai-GenPrompts2200.txt as a wildcard with Dynamic Prompts, simply place it into the /wildcards/ folder and rename it to your choosing.
   
   I trained a GPT2 Generator which created many variations on the Example Prompts, an example of these are the JDZenkai-GenPrompts2200.txt

This is an ongoing project so expect a LORA version next. It will take some time to create the LORA version due to a difference in the captioning. This is trained using a Parallel Dreambooth Method which prevents concept/style bleeding, but you can still mix by interpolating trained terms/triggers in the prompt.


TLDR there is a missing comma, can you find it?

License

Because of the base models this model inherits some terrible licenses 😔

• Dreamlike Diffusion 1.0 License

  • https://huggingface.co/dreamlike-art/dreamlike-diffusion-1.0

• Seek Art License

  • https://huggingface.co/coreco/seek.art_MEGA/blob/main/LICENSE.txt

PROMPTING

The model is easy to prompt for, words like blurry background, depth of field are good for example:

Positive Prompt:

• 1girl, asian, blurry background, depth of field

Negative Prompt:

• worst quality, lowres, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, grainy, signature, cut off,

• I usually remove blurry or blurred from the negative prompt

Words like film, and photo are good as well:

Positive Prompt:

• 1girl, film, photo, depth of field, gradient hair,

Negative Prompt:

• EasyNegative,, worst quality, lowres, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, grainy, signature, cut off,

• I also removed blurry or blurred from the negative prompt for this image

Because of the realism of this model I used it to replace the person in this video with emma watson using TikTokDiffusion you can see a side by side comparison

• 

• 

  • What Is TikTok Diffusion

• 

well, checkpoint trained in references from Defoko, i need say something more?

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,

A sister project of WyvernMix, geared more towards NSFW erotic portraits.

Works well with 5-7 CFG, prefers higher native res (768x960)

Version comparison (first row: V1, second: V2, third: V2.1):

Doesn't do too well with more hardcore stuff, but maybe it just needs better prompting, let me know if you have good results.

Also let me know in the comments what improvements I could make in future versions.

It is another previous model what I did last month.

Those models are very similar.

• You can also use controlnet,

• It's good to do the poses.

I can't use controlnet if it can fail.

• how difficult the crooked hands....! There's no other way.

It's a bit complicated. That's how the models are, the ads are more difficult to find, the ads are different and they keep changing. And well, I look for a realistic photograph, they are like photos. there is no other way of real. but the shapes of the models are beautiful: anime, realistic, cartoon, drawings, magazines, robot, warrior, hero, characters, luxury houses, fantasies, whatever they can do, LORA, textual...

And why don't you show me your work?

I hope you like them!

• if you can, or save your hard disks or disks or pen drive... with safetensors, because they are large MB, or if some can delete the files, your documents

Hires. fix

resize: from 440x480 to 901x983

scaler: R-ESRGAN 4x+

hire steps: 44

noise suppression force: 0.7

Exclusive by: 2.05

Resize width to: 0

Steps: 44, Sampler: DDIM, CFG Scale: 7, Size: 440x480, Denoise Force: 0.7, Upscale Hire: 2.05, Step Hire: 44, Upscale Hire: R-ESRGAN 4x+

Thank you.

I trained this model with dreambooth using 500 images of living rooms and interior design, this is a test model, and it will be refined in the further versions.

As usual you guys a model with good hands and details 95% of the time ... im really picky like that!
this one has OVER 50 MERGES !! please consider supporting me on my kofi link https://ko-fi.com/horrid111

unlike other models of this style i made sure my model requires very little prompt engineering to achieve maximum results! so have fun using it please share your results!!! inpainting model included!

05-01 : v03

In version v2.3, the anime tendency was slightly visible, so I tried to make it more realistic again.

04-24 : v2.3

This version has changed mixing ratios compared to the previous v2.2 and includes some experimental attempts. As always, extensive testing has not been conducted.

I uploaded the pruned version separately, but it is currently downloading only the full version maybe there is a problem with Civitai. I will solve the problem and upload the pruned version within the 25th

Upload pruned model finished

04-10 : v2.2

I attempted to improve the skin tones and lighting and shadows, which were previously awkward. I have noticed a difference in the image, but I have not had the opportunity to thoroughly test the changes due to my limited skills.

VAE : vae-ft-mse-840000-ema-pruned

Steps: 50, Sampler: DPM++ SDE Karras, CFG scale: 20, Size: 512x768, Model hash: a8907e32c0, Denoising strength: 0.35, Clip skip: 2, Hires upscale: 2, Hires steps: 20, Hires upscaler: R-ESRGAN 4x+, Dynamic thresholding enabled: True, Mimic scale: 10, Threshold percentile: 99.95, Mimic mode: Cosine Up, Mimic scale minimum: 0, CFG mode: Half Cosine Down, CFG scale minimum: 15, Eta: 0.2, DDetailer prompt: "", DDetailer neg prompt: "", DDetailer model a: "bbox\mmdet_anime-face_yolov3.pth [51e1af4a]", DDetailer conf a: 30, DDetailer dilation a: 4, DDetailer offset x a: 0, DDetailer offset y a: 0, DDetailer preprocess b: False, DDetailer bitwise: None, DDetailer model b: "None", DDetailer conf b: 30, DDetailer dilation b: 4, DDetailer offset x b: 0, DDetailer offset y b: 0, DDetailer mask blur: 4, DDetailer denoising: 0.35, DDetailer inpaint full: True, DDetailer inpaint padding: 32, DDetailer cfg: 11

04-01 : v2.0

As always, I didn't conduct many tests. However, I did notice a slight improvement in prompt recognition.

I'm not entirely certain if it's an actual upgrade from the v1 version or not.

Steps: 40

Sampler: DPM++ SDE Karras

CFG scale: 20

HIRES. FIX

R-ESRGAN 4X+

STEP : 20

DENOISING : 0.35

UPSCALE : 2

Dynamic Thresholding 

MIMIC CFG SCALE : 10

PERCENTILE : 99.95 OR 99.9

MIMIC SCHEDULER : COSINE UP

MINIMUM VALUE OF MIMIC : 0

CFG SCHEDULER : HALF COSINE UP

MINIMUM VALUE OF CFG : 0

Ddetailer

bbox\mmdet_anime-face_yolov3.pth

Denoising strength(Inpaint) : 0.35

Mimic CFG Scale : 15 (custom ddetailer setting)

----------------------------------------------------------------------------------

03-17 : v1.0

I have updated the model with the extension "safetensors" upon the request of someone else, while keeping it the same as the existing model.

----------------------------------------------------------------------------------

This is a realistic model that I personally merged.
There is limited information due to a small number of tests conducted.

This is an example of a prompt used

ultra realistic details, sharp focus, detailed skin, (21 years old pretty korean girl:1.2), (kpop idol:1.2), (korean beauty:0.8), (korean mixed),

(large breasts),

Negative prompt: Drawings, abstract art, cartoons, surrealist painting, conceptual drawing, graphics, (low resolution:1.4), (blurry:1.3), (strabismus:1.1),

(worst quality:1.3), (low quality:1.3), (thick thighs:1.2), collage, (makeup:0.8), nsfw, bad proportions, earings, floral print, loli, big eyes, (watermark:1.2), letter, (abs:1.2),

Steps: 40, Sampler: DPM++ SDE Karras, CFG scale: 7.5, Seed: 3203463706, Size: 768x512, Model hash: 1d8d6e1204, Denoising strength: 0.35, Clip skip: 2, Eta: 0.2

HIRES. FIX

R-ESRGAN 4X+

STEP : 20

DENOISING : 0.35

UPSCALE : 2

Dynamic Thresholding 

CFG : 20

MIMIC CFG SCALE : 5

PERCENTILE : 99.95 OR 99.9

MIMIC SCHEDULER : COSINE UP

MINIMUM VALUE OF MIMIC : 0

CFG SCHEDULER : HALF COSINE DOWN

MINIMUM VALUE OF CFG : 10

Long time no see. :>

VAE : ft-mse-840000

Clip skip : 2

Steps: 30

Sampling method : DPM++ SDE Karras

Hires steps : 14

Hires.fix upscaler : R-ESRGAN 4x+Anime6B

CFG Scale : 6~10

别问，问就是对对对

Renders fantasy art in a style similar to Larry Elmore / Dragonlance, both groups and portraits

这个模型在人像和风景场景中表现良好，具有广泛的效果，用户可以自行探索。

我上传的大部分图片都经过了TEXT2IMG-Hires Fix--TO---IMG2IMG的处理过程。对于IMG2IMG步骤，我启用了Tiled Diffusion进行图像放大。如果您想要增加图像的细节，还可以启用face editor。但是，有时候这两个插件同时工作可能会导致图像出现问题，因此用户可能需要相应地调整数值。

如果你觉得这个模型不错，请将你的图片分享出来

This model has been tested extensively and performs well in both portrait and landscape scenes. Additionally, there are many more effects that users can explore on their own.

Most of the images I uploaded went through the process of TEXT2IMG-Hires Fix--TO---IMG2IMG. For the IMG2IMG step, I enabled Tiled Diffusion for image enlargement. If you want to add more details to the image, you can also enable the face editor. However, sometimes there may be issues with the image when both plugins are working together, so users may need to adjust the values accordingly.

If you think this model is good, please share your images.

推荐参数

Recommended Parameters:

Sampler: DPM++ 2M Karras alt Karras or DPM++ SDE Karras

Steps: 20~40

Hires upscaler:4x-UltraSharp,RealisticRescaler 4x,R-ESRGAN 4x+ Anime6B

Hires upscale: 2

Hires steps: 15

Denoising strength: 0.2~0.5

CFG scale: 6-8

clip skip 2

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 model (v1: chilled regeneric v2, v3: Muse), and then transformed to toon-like one.

If you use my model "CityEdgeMix", you may notice that same prompt/seed as CityEdgeMix also produces similar images as CityEdge_ToonMix. CityEdge_ToonMix is based on CityEdgeMix, which is photorealistic merge model, and 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（v1: chilled-regereric v2, v3:Muse）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。

This is a merge I refined in the last few week using some models I found in this wonderful community - you can now choose between normal and bakedVAE version.

✦ Comics Vision - Quick Start ✦

1_____ I highly suggest to use these embeddings:

• easynegative https://civitai.com/models/7808/easynegative

• bad-hands-5 https://huggingface.co/yesyeahvh/bad-hands-5/tree/main

2_____Size: 640 x 768
(all uploaded pictures are upscales of these original dimensions).

3_____Prompt: Start from the prompt of the pictures I've uploaded and then customize the areas focused on Character, Setting, Emotions, Colors and Aesthetic.

• for example:
  (inked original-art:1.2), (lineart:1.33),1woman,space-viking society, blonde braid strong manly woman barbarian,wrinkles,(depth),(movement:1.4),cinematic-composition,snob smirk,(hyperdefined),intricate-setting,undercut,yellow white orange,winter futuristic-city-streets,(thin-lines),viking-sci-fi aesthetic,neon-lighting,stars,giant-robots-machines,(fine-details:1.3),RAW-format,emotional,(muted-colors)

PS: if you find weird that the model is called "Toontanium" in the picture's data it's just because that was its previous name, but it's the same version you have here, don't worry!

4_____Upscale: to get crisp results like the one shown in the pictures use HiResFix and Real-ESRGAN-Anime. I suggest to use this step only on the preferred pictures, since it's quite heavy on the performance side.

I hope you like my merge and I can't wait to see what you all are going to create with it!

For future update, about the AI Editor I'm building with an italian team and my models you can check:

https://www.instagram.com/aledelpho/
____________________________________

The models used to create this merge are:

• Icomix - https://civitai.com/models/16164/icomix

• SleeplessMix - https://civitai.com/models/37097/sleeplessmix

• Mistoon_Sapphire - https://civitai.com/models/32022/mistoonsapphire

• Mistoon_Emerald - https://civitai.com/models/28322/mistoonemerald

• + Some of my own stuff

Without these models, this one would not have been possible.
If you like Comics Vision, remember to check these other models and consider to support their creators!

v2这个版本改善了远景构图以及远景的梦幻感，里面加入了我经过调试的新的lora-cinematic的参数，以及一些其他的辅助lora，而近景变化不大（手好像没那么稳定了= =）

This version of v2 improves the composition of the vista and the dreaminess of the vista, adding my new debugged Lora-Cinematic parameters, as well as some other auxiliary LoRa, while the close-up scene doesn't change much (the hand seems to be less stable = =)

v3改善了脸部和手部，以及一些肢体结构，还改善了画面精细度。泛用性提高了不少，现在能兼容更多lora了，污染也减少了。

The V3 improves the face and hands, as well as some limb structure, and also improves the fineness of the picture. Versatility has improved a lot, more LoRa is now compatible, and pollution is reduced.

04-30

Compared to the v1.0 model, the 2.0 model has been expressed in a more anime-like style. This model is derived from the henmix_real_v3 model, which is expected to be uploaded soon. As a creator, I don't have much to say since I'm not familiar with using 2.5D.

-------------------------------------------------------------------------------------

This is a 2.5D model that I personally merged.
There is also limited information due to a small number of tests conducted.

used vae-ft-mse-840000-ema-pruned.ckpt

This is an example of a prompt used

1 korean cute girl, (21 years old cute Korean girl:1.2), (kpop idol:1.2), (korean beauty:0.8), (korean mixed), (large breasts),

Negative prompt: (low resolution:1.4), (blurry:1.3), (worst quality:1.3), (low quality:1.3), (thick thighs:1.2), collage, nsfw, bad proportions, loli, (forehead:1.5)

Steps: 40, Sampler: DPM++ SDE Karras, CFG scale: 20, Denoising strength: 0.35, Clip skip: 2, Eta: 0.2

HIRES. FIX

R-ESRGAN 4X+

STEP : 20

DENOISING : 0.35

UPSCALE : 2

Dynamic Thresholding 

CFG : 20

MIMIC CFG SCALE : 5

PERCENTILE : 99.95 OR 99.9

MIMIC SCHEDULER : COSINE UP

MINIMUM VALUE OF MIMIC : 0

CFG SCHEDULER : HALF COSINE DOWN

MINIMUM VALUE OF CFG : 10

More anime-like style, better anatomy on clip skip 1.

Negative Prompt: rMadaNegative4 (SD 1.5) | Stable Diffusion TextualInversion | Civitai

VAE: stabilityai/sd-vae-ft-mse-original at main (huggingface.co)

Optional: rMadArt3_NoiseOffset (SD 1.5 Lora) | Stable Diffusion LORA | Civitai

Please select the version you like, not the latest.

• W means wide, aims to be a multifunctional model which also has stability and prompt-controllability. (sfw+nsfw)

• Prompt can be long or short, individual words or long sentences, the painting style will be affected.

• Commercial use? Ask me for the mixed model list and go ask their original author. (Just a suggestion)

• The example is generated with wd-2 vae and "No xformer","use cpu all" (or webui-colab default)

• CLIP has incorrect positions (not a big problem) (later on the fixed ones)

Please share your amazing works here if you would like to :)

If you would like to see the recipe or more detail, feel free to ask.

badtimeforeveryone

since I use these amazing to merge (which aren't belong to me), I won't use these models to make money (of course you can if you got the permissions from their original author)

if you really like my works and want to give me some kind of encouragement, consider click on the link to find out what it really is, I will be grateful.

• this model is made to make furrys, either male or female, and with sufficient specification, you can change the proportions of certain aspects of the body. Is mostly adapted to make a single person, with two or more it can become a bit messy. (Both SFW and NSFW)

• As the name indicates this is a combination of other furry models, some are yiffyMix, BB95, and FluffyRock, among others, I highly recommend you try those, especially BB95 if you want to simulate realistic fur and FluffyRock if you want to imitate artist styles

• I also recommend anyone interested in furry content to join the Furry Diffusion discord to see more creations, prompts and content in general related to AI image generation.

2023.4.30 version 2.0 updated.

Reduced model size. Made it cleaner and brighter than V1.0.

ckpt version here.

2023.4.27 Change model to safetensor.

Click here if you want ckpt version.

This model is created by merging many LoRa and other models to produce pictures of cute furry girls. I hope you like it. (*^▽^*)

融合了一些模型和Lora得到的模型，可以生产可爱福瑞图。希望各位用得开心。(*^▽^*)

Recommend settings:

512×768, DPM++ 2M Karras, Steps: 20-30, and add “furry, animal ears, tail, " to prompt.

Used models:

NineKeyMIX5

Mistoon_Anime

MIX-HIRES
Used Loras:

Squeezer - Experimental

Rikose Style

Nukunuku Nigirimeshi
Furry girl

ToonYou - Beta 1 is up!

Silly, stylish, and.. kind of cute? 😅

A bit of detail with a cartoony feel,

and it keeps getting better!

⬇Please read the information below 4/30

////////////////////////////////////////////////////////////////////////////////////

Recommended Settings - VAE is included starting with Alpha2

• Clip skip : 2

• Hires. fix : R-ESRGAN 4x+ Anime6B / Upscale by 1.5+ / Hires steps 14 / Denoising strength: 0.4

• Adetailer - face_yolo8n.pt

• Sampler : DPM++ 2M Karras / DPM++ SDE Karras / UniPC

• CFG : 8 Steps : 25

• Prompts

  • (best quality, masterpiece)

  • Neg : (worst quality, low quality)

////////////////////////////////////////////////////////////////////////////////////

Do you like my work? check out my profile and see what else!

And A cup of coffee would be nice! 😉

////////////////////////////////////////////////////////////////////////////////////

For anything other than general personal use, please be sure to contact me

You are solely responsible for any legal liability resulting from unethical use of this model

BeenYou - B17 is up!🥳

For Small breasts & Skinny lovers

Back to the feel of the early versions

Of course, the problem came back too!🥲

⬇Please read the information below 04/30

////////////////////////////////////////////////////////////////////////////////////

Known issues - B17

• Ignore hair color, mixed colors

• Pupil expression issues (can be minimized with the settings below)

• Some instability caused by Niji learning

////////////////////////////////////////////////////////////////////////////////////

Recommended Settings - VAE is already included

• Clip skip : 2

• Hires. fix : R-ESRGAN 4x+ Anime6B / Upscale by 1.5+ / Hires steps 14 / Denoising strength: 0.4

• Adetailer / Just select face_yolo8n.pt

• Sampler : DPM++ 2M Karras / DPM++ SDE Karras / UniPC

• CFG : 8 Steps : 25

• Prompts

  • (best quality, masterpiece), looking at viewer,

  • Neg : (worst quality, low quality)

////////////////////////////////////////////////////////////////////////////////////

Do you like my work? check out my profile and see what else!

And A cup of coffee would be nice! 😉

////////////////////////////////////////////////////////////////////////////////////

For anything other than general personal use, please be sure to contact me

You are solely responsible for any legal liability resulting from unethical use of this model

Playing around with models and I wanted one that could take the minimum of prompts and produce acceptable results. The prompt: "award winning photo of a beautiful woman trending on flickr, 50mm f1.2" I had a few different versions and I settled on this one. It's based on all SD 1.5 models and has a little of the each of the following plus some personal photography work baked in: URPM, PornVision, PurePornPlusMerge, Clarity2.0, ProtoGenX3.4 and Photosomnia... so it's a merge of a many merges. The model has an affinity for brunettes and certain faces. It's obviously capable of NSFW content but not unsolicited from my experience so far. Since I have trained into this model some personal photographic work it's not intended for commercial use.

A test project for a new anime model, made at the same time as CAN

Hello, hello. I have returned with a holy grail for y'all.. folks. With a model that surprised me when I had first merged it. It has a Abyssorangemix2_hard feel to it, but at the same time the saturated anime esk feeling goodness. So its detailed, highly anime like feeling model.

I was going for this but I didn't expect it on my first go around when I was making it. Heres the "specs" for y'all people wanting to test some stuff out a bit.

Bloodymarrymix 30%

Anything V4.5 30%

MeteoriteDiffusion 10%

And lastly AOM3 30%

For the last merge, name the model created from the first thing anything you want then make the slider for merging 0.3. This is for A1111 users. Anyone else using Gradio or shit like that Im leaving you in the dark on this one.

Anyways, here is the recommended settings for da model.

Steps (As usual) 20 for test, 34 for illustration. 50 and above.

CFG Scale: 8-12.

Sampler: (THIS IS GONNA BE A SHOCKER) DPM ++ SDE Karras

Upscaler: Latent Nearest Exact Your pick.

Hope y'all have fun with this model. Its very good and it can create high quality images very easily, a lot like BMM and MeteoriteDiffusion.

This is my first and likely only attempt at merging models. I have no idea what I'm doing and just ran through some tutorials and hit buttons. This model is a merge of 3 different 'cartoon' style models to give a blend of more 'western' stylizing while gaining a lot of responsiveness to prompts. Seriously, no need for a prompting guide, it responds to all styles, including anime tags and 'natural language' sentence fragments. This model works well with most LoRAs and negative embeddings. It struggles with hands a bit still and the occasion extra limbs, but otherwise outputs are solid.

All images were generated using only prompting, via an advanced wildcard extension called Umi AI. One of the unpublished source models used for this merge was created by a member of their team. Check out their published model Macross V2.

There is no VAE baked in. I recommend the standard Stable Diffusion VAEs, Clear VAE, or my own Anime VAE.

I stand on the shoulders of giants. Credit to CivitAI for providing this space for us, and to the community that seeks ever greater levels of perfection.

Like all checkpoints since the originals released by Stable Diffusion, this model is responsive to age related keywords. It is also capable of producing NSFW content. What you do with this model is your choice. I suggest not making questionable images by using negative prompting as needed. I am marking this model as intended for mature audiences because of this.

一个与之前maturemalemix模型的融合，加入一些美式卡通风格的成分，拥有粗犷的线条和卡通风格，发型和衣服可能不太稳定，浅玩一下。

A fusion with the previous maturemalemix model, adding some American cartoon style components, with bold lines and cartoon style, hair and clothes may be less stable.

https://discord.gg/MDpfe3JBQX

https://ko-fi.com/bgmodel

Please read the description for the models i put out. There is some good information that may increase the quality of your output when using my models.

Use clip skip 2 btw

A test project for a new anime model, made at the same time as CAB

Combined Many models with orange3 at 40% and dozens of other models at 60%.

If you like my model you can support me at

https://www.patreon.com/JumyJam

https://jumyjam.fanbox.cc/

I do R18 illustrations and doujinshi with AI.

This is a blend of models that I've been working on for a while, with the primary goal of creating realistic character portraits in a variety of genres, but it's a well-rounded model that's been good at everything I've thrown at it. It's probably not very good at anime, but if that's what you're after there are plenty of other models to choose from.

This model inherits the CreativeML Open RAIL-M and Dreamlike License. By downloading, you agree to both.

No need to use a VAE. The model has vae-ft-mse-840000-ema-pruned.ckpt baked in. I also strongly recommend using the following negative embeddings: neg-sketch-2 by JPPhoto and neg-anime by dwringer.

The example images were done with InvokeAI. If you want to reproduce them with a1111, you'll need to convert the syntax.

This model can be used to make a lot of fun images (see examples)

CLIP skip 1
VAE - vae-ft-mse-8400000
https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.ckpt

-Update 4/29/2023

betterMix3.0_Anime and betterMix3.0(with_cetus)_Anime is added.

3.0 is a general improvement on 2.0, focusing on better eyes, anatomy (and a bit better hands) while maintaining most of the features of 2.0 like wonderful color and light and shadow.

3.0 (with cetus) mixed cetusMix inside. Compared to 3.0, it sacrifices some color and light effect in exchange for better anatomy.

All the example images provided here are using hiten1 and EasyNegative these 2 embeddings.

The link to hiten1 embedding:

https://civitai.com/models/15430/hiten-style

The link to EasyNegative embedding:

https://civitai.com/models/7808/easynegative

-

This model is made by merging PastelMix, Anything3.0 and OrangeMix together.

I mainly use this model to cooperate with the "hiten1" embedding (textual inversion) I made, the link to it is above.

Suggested basic parameters:

• Sampler: Euler a & DPM++ 2M Karras

• Sampling steps: 20

• CFG scale: 7

Recommend to use this model with highres fix on.

Suggested highres fix parameters:

• Upscaler: Latent

• highres steps: 20

• Denoising: 0.6

• Upscale from 640 360 to 1280 720

I've also uploaded this model to hugging face, link is below

https://huggingface.co/SakanakoChan/BetterMix2.0_Anime

Ascendance Of A Bookworm

Authors: Juusoz, Tylenol, Goldkoron and Expl0dingCat

Feel free to join our Discord community for updates on current models or models on other shows.

Prompt Format

The prompts we trained on are available in the corresponding T{version} prompt list.txt file.

T3.2 Prompt list.txt

Format: Parent tag, description prompt 1, description prompt 2, etc;

The parent tag serves as the primary label for the overall theme of the image and acts as the main subject. The description prompts are a comprehensive list of attributes, objects, actions, and other elements that describe the subject in detail. Each subject is separated by a semicolon (;) while the individual attributes, objects, actions, and elements are separated by a comma (,).

Example:

Positive prompts: MFB Style; 1boy; 1girl; Lutz; Myne, poor, yellow eyes, blue hair, hairstick; (masterpiece); detailed; reflective; depth of field; 8k; photo;

Negative prompts: (bad hands, error), bad anatomy, low quality, worst quality, watermark, signature, poorly drawn face, noise, grain, nude, naked, upskirt, blind, AOB Style, solo

Just because we haven’t trained on something doesn’t mean the base AI model doesn’t already know what it is, so go crazy.

Style tags:

You can specify what style the image should generate in using the following tags.

MyneFactoryBase style, for more artistic pictures MFB Style

Ascendance of a Bookworm style, for more anime accurate pictures AOB Style

Recommended Settings

• Sampling Steps 40

• Sampler DPM++ SDE Karras

• Image Size At least 768x768

• CFG 7

• Clip skip 1 (settings, Clip skip)

Socials

Website | Discord | Patreon | CivitAI

AnyLoRA

Add a ❤️ to receive future updates.
Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕

Remember to use the pruned version when training (less vram and no baked vae).

I made this model to ensure my future LoRA training is compatible with newer models, plus to get a model with a style neutral enough to get accurate styles with any style LoRA. Training on this model is much more effective conpared to NAI, so at the end you might want to adjust the weight or offset (I suspect that's because NAI is now much diluted in newer models). I usually find good results at 0.65 weigth that I later offset to 1.

This is good for inference (again, especially with styles) even if I made it mainly for training. It ended up being super good for generating pics and it's now my go-to anime model. It also eats very little vram.

The first version I'm uploading is a fp16-pruned with no baked vae, which is less than 2 GB, meaning you can get up to 6 epochs in the same batch on a colab.

Just make sure you use CLIP skip 2 and booru style tags when training.

Remember to use a good vae when generating, or images wil look desaturated. I suggest WD Vae or FT MSE. Or you can use the baked vae version.

The model is assembled from a huge number of different art in the style of Western comics.

I didn't use anime-style art at all to avoid the Asian cartoon style of drawing.

The default model, unless otherwise specified, is prone to big tits, powerful buttocks and underwear.

As in any Western comics, this model has high contrast, bright outline, glare. The main drawback is that the model likes to sculpt logos in the style of comic book covers. Approximately every fourth drawing.

Модель собрана из огромного количества разных артов в стиле западных комиксов.

Я вообще не использовал арты в стиле аниме, чтобы избежать азиатского мультяшного стиля рисования.

Модель по умолчанию, если не указано иное, склонна к большим сиськам, мощным ягодицам и нижнему белью.

Как в любых западных комиксах у этой модели высокая контрастность, яркий абрис, блики. Главный недостаток - это то, что модель любит лепить логотипы в стиле комиксных обложек. Примерно каждый четвертый рисунок.

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<<

Amixx

This is a mix ; Amixx.

It does clothing well, which is obviously what I was looking for when I merged it, and is quite versatile in my opinion.

Let me know what you think of it!

Enjoy my mix, Amixx!

Vae Pruned 840000 ema

Realistic

Photo

MIX https://civitai.com/models/49592/fantasy-mix-v15

https://civitai.com/models/53862/meganimevae

My personal-use blend, which I decided to release. Tuned for cute characters in bold styles - very good at gothic lolita fashion, princesses, and other similarly ornate clothing. Performs better on detailed, interesting backgrounds than other models of a similar style.

Recommended VAE is blessed2 - which is a direct, superior replacement for kl-f8-anime2.

If you omit the "high detail masterpiece" etc. stuff it becomes more generic.

Pruning this model only knocks off 1GB, so I don't recommend it.

---------------

我个人使用的混合，我决定发布。为大胆风格的可爱人物而调整--非常擅长哥特式洛丽塔时装、公主和其他类似的华丽服装。在详细、有趣的背景上比其他类似风格的模型表现得更好。

推荐的VAE是blessed2--它是kl-f8-anime2的直接、优越的替代品。

如果你省略了 "high detail masterpiece "等东西，它就会变得更加普通和平淡。

不建议对这个模型进行修剪，因为它只减少了1GB。

---------------

私の個人的な使用のためのミックスは、私が投稿することにしました。 ゴスロリファッションやプリンセス、その他似たような装飾のコスチュームがとても似合う、大胆なスタイルで素敵な人物になるように調整しました。 他の似たようなスタイルのモデルよりも、詳細で興味深い背景でより良いパフォーマンスを発揮します。

推奨されるVAEはblessed2 - kl-f8-anime2の直接的で優れた代替品である。

「high detail masterpiece」などを省くと、より一般的で当たり障りのないものになります。

このモデルをプルーンしても1GBしか減らないので、お勧めしません。

Anime Pastel Dream

CHECK BOTH VERSIONS BELOW

Add a ❤️ to receive future updates.
Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕

Available on Sinkin.ai with GPU acceleration.
MY MODELS WILL ALWAYS BE FREE.

This is my attempt at making a good anime pastel model. I love the end result and I hope you enjoy it. This should also work great with most anime models.

I made 2 versions:

• Hard has a stronger pastel style and gets a lot more creative.

• Soft has a subtler pastel style and is much easier to control.

Consider supporting model creators of the models you like!

Suggested settings:

• CLIP skip 2

• I had ENSD: 31337 but it's optional

• NO restore faces

• Highres fix with either a general upscaler and low denoise or Latent with high denoise (see examples)

• Be sure to use Auto as vae for baked vae versions and a good vae for the no vae ones. Avoid anythingv3 vae as it makes everything grey.

• Steps and upscale denoise depend on your samplers and upscaler. See the examples.

just another mix with fluffyrock in it so expect all the same tokenization as fluffyrock.

formuler:

fluffyrock-576-704-832-960-1088-lion-low-lr-e18-offset-noise-e3 at .75

furry vixens at .25

Here

https://civitai.com/models/52441/megaanimerv5

https://civitai.com/models/52297/rmega-anime-v7

https://civitai.com/models/51797/meganime-v5

Vae and Model V1.5

Realistic

Anime

"Everywhere Mix CKPT"

If you can change with safetensors or Lora,

CHECKPOINT MERGE

"Aira Lumina" is a huge additive model merge, as can be seen by its ~4gb fp16 size. It carefully combines well over 200 models, including both realistic and anime checkpoints.

This has been a labor of love, often doing and undoing merges based on the results. Probably the most impressive thing is the integrity of the results -- you have less control over whether you get anime/photorealistic photos, but it passes alexds9's overfit/corruption test with flying colors.

Because the merge is mostly additive, it performs well at both clipskip 1 and 2, and flipping back and forth can yield interesting results.

Enjoy!

I don't speak English so I'm translating at DeepL. Let me know if the English is weird.

What would happen if I mixed my favourite model, Nostalgia, with a realistic model?
The result was this model, which I think it quite interesting.

A:nostalgiaMix_nostalgia.safetensors
B:braBeautifulRealistic_brav3.safetensors
I mixed the 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0

私の大好きなNostalgia、これにリアル系モデルを混ぜたらどうなるんや・・・？
というわけで結果的に面白いモデルができた。

A:nostalgiaMix_nostalgia.safetensors
B:braBeautifulRealistic_brav3.safetensors
を、0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0で混ぜたよ

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

This new version has been trained with more than 270 images. (Including the nsfw of course) If someone wants to teach me how to do a LoRa I'll be happy to learn Now it seems that this new model understands the red drawing on the shirt better.

that's a Checkpoint from kasane teto! i searched and didn't found any of kasane, so i made my own! well, this maybe can be bad, i used only 20 photos as reference--

The models are the same. You probably don't want the no VAE version.

This model was originally a dreambooth training. It has since been through more merges with more models than I care to remember or count and has been fine tuned with images from various artists meeting the style. I have marked this as "Trained" because, despite it containing merges, it did in fact have a good amount of training too.

It's made to be incredibly cute and adorable while being capable of taking on basically any Lora with some form of efficiency. I would recommend using lower weights for your Lora with this model than you would any other model. This model also does very good scenery!

Please do not use this model for NSFW purposes. I highly recommend placing sfw in your positive prompts and nsfw in your negative prompts. I'm sure you can understand why this is from the previews.

Previews are generated using automatic1111's Stable Diffusion web UI. No Lora's, Hypernetwork's or anything else was used in the preview generations aside from negative embeddings. Previews were generated without using Hires. fix at 576x1024 for tall images and 2048x1152 for wide images. It is unlikely you can perfectly replicate my previews as I am using xFormers + ToMeSD.

Also available at 🤗: https://huggingface.co/Miyuutsu/kittenchow

这个融合模型可以生成多种多样风格的插画，整体风格偏向动漫写实渲染，对于面部，会根据你的词条改变风格（我个人比较喜欢可爱的风格，所以往里面融了一些大家喜闻乐见的脸模），这个模型对各种词条有还算不错的兼容性，如果你有什么好玩的想法，可以拿它试一试，希望你能享受这个模型带来的乐趣~

我常用的配置是:Steps: 24, Sampler: DPM++ 2M Karras, CFG scale: 7, Size: 512x768, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, Hires upscale: 1.5, Hires steps: 10, Hires upscaler: R-ESRGAN 4x+ Anime6B

以下内容为机器翻译/The following is machine translated:

This fusion model can generate a variety of styles of illustration, the overall style is inclined to anime realistic rendering, for the face, will change the style according to your words (I personally prefer a cute style, so into the fusion of some of the popular face models), this model has a pretty good compatibility with a variety of words, if you have any fun ideas, you can take it to try, I hope you can enjoy the fun brought by this model ~

My usual configuration:Steps: 24, Sampler: DPM++ 2M Karras, CFG scale: 7, Size: 512x768, Denoising strength: 0.3, Clip skip: 2, ENSD: 31337, Hires upscale: 1.5, Hires steps: 10, Hires upscaler: R-ESRGAN 4x+ Anime6B

This is a merge of PsyGon AniMerge v2 at 50% mixed with FluffyRock and FurryStaber at 25% each.

Her name is RUIS.

==ReadMe===========================================

The fun AI life we all enjoy

let's have fun

==================================================

Prompt

• (write down your imagination)

Negative

• (EasyNegative, badhandv4)

Default Setting.

• USE VAE "clearvae_main.safetensors"

• https://civitai.com/models/22354 with download

• Steps: 30,

• Sampler: DPM++ SDE Karras,

• CFG scale: 7,

• Denoising strength: 0.4,

• Clip skip: 2,

• Enable Dynamic thresholding enabled: True, Mimic scale: 7

• Script -> Detection Detailed -> bbox/mmdet ***

Step-by-step method of outputting as shown in the figure

First Step

• Hires. fix -> R-ESRGAN 4x+ Anime6B -> Hire Steps 10 -> Denoising strength: 0.4,

Second Step

• send to inpaint ->

• Paint both eyes of the character black with Mask Blur

• Enable Dynamic thresholding enabled: True, Mimic scale: 7

• Script -> SD upscale -> upscaler: 4x-UltraSharp,

To recap,

the first upscale is R-ESRGAN 4x+ Anime6B, the second upscale is 4x-UltraSharp after painting both eyes in inpaint. Upscale a total of 2 times to improve the quality.

A sample of commonly drawn images

Eclipse

Legal & Risk

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

• v1.0-Asami: First release, made from self-trained model, merged with various other models

• v2.0-Byakuren: updated from v1.0, added more hairstyles, more poses, more vehicles, etc, trying to fix hands by using hires trained image.

Who find this model perfect

• LoRA/LoCon/LyCORIS user who are looking for a compatible checkpoint

• User who have limited resources, because this checkpoint can generate good image start from 15 sampling steps, no need to wasting time for sampling steps

• Creative prompters

Usage

• Almost compatible with all sampling methods available

• Use 15-40 sampling steps

• CFG: 4.5-7

• Less complicated, short negative prompts are better

If using Hires:

• Hires step: 20 steps or same as sampling steps

• R-ESRGAN 4x+ Anime6b, if you're prefer Latent upscaler, use Latent (nearest-exact)

• Denoising: 0.3-0.7

VAE: https://huggingface.co/stabilityai/sd-vae-ft-mse-original

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

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

p.s. It's not necessary but If you like my creations 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 etc.

I don't know why, but this model becomes clearer or simpler depending on the prompt.

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

V1.1より以下に変更

XXMix_9

https://civitai.com/models/47274/xxmix9

本モデルも使用したモデルの利用条件に従う形になりますが、以下に関しては厳に使用を禁止いたします。

・暴力的な表現

・児童ポルノ

・未成年者の性的な表現

本モデルは『CreativeML Open RAIL-M』の範囲でラインセンスされます。 本モデルを使用した上での問題に関しては、当方は一切責任を持ちません。ご了承の上ご使用ください。 また、マージモデルのライセンス変更に伴い、公開を停止することがあります。

This model is also subject to the terms and conditions of use of the model used, but the use of the following is strictly prohibited.

・Violent expressions

・Child pornography

・Sexual expression of minors

This model is licensed within the scope of "CreativeML Open RAIL-M". We are not responsible for any problems caused by the use of this model. Please use this model with your understanding. We reserve the right to suspend publication of the merged model due to changes in its license.

一个简单的融合模型，使用了WolfBoys_2D和NijiV5style为主加上一些其他融合模型，并融合了一些成熟男性的lora，推荐使用muscular和facial hair为提示词。

A simple merge model, using WolfBoys_2D and NijiV5style as the main plus some other merge models, and incorporating some mature male lora, with muscular and facial hair recommended as prompt words.

V3.0 Notes:

This really jumped up to the next level with a few extra tweaks to the LoRa styles plus Merging in Galena. The example images are all done with NO negative prompt and just a handful of words in the prompt - usually just a name and a quick description of what they are doing.

This also takes LoRa's and TIs a lot better than the original. For version 1 I needed to back down the weights quite considerably, for this version leaving them at the default 1.0 (or the values recommended by the creator) tended to work really well.

You can get some cool looks and variations when combining TI and LoRa triggers with the names of known celebrities, too.

This version tends to go a bit more Porny/NSFW than the original, but a lot of that is keyed into the source of the training. Celebrities who tend to not do the sexy stuff aren't as likely to default to sexy stuff - while a porn star or nude model are more likely to default to nudity and sex. Call your shots and it works out well either way, though.

Upscale is always a good thing (though in these, I just did latent) while Face Restoration is often better to leave off (but not always).

Body types tend to always be similar - but you can call them using various bodyweight/style prompts.

Original Version Notes (much of which is still true):

I wanted a model or workflow process that allowed me to quickly and easily churn out art for games and/or photo stories. I didn't want long complicated prompts to create a certain art style, but I also didn't want to be forced to use LoRa's and TI's for every character. Unfortunately, most of the models where I loved the art styles didn't do so well of loading characters (celebrities for mixing, fictional characters, etc.) and the ones that knew tons of famous characters and people tended not take to the art styles I was going for (at least not without complex prompts).

There are some great Anime style art prompts that did a nice job, but I didn't really want Anime faces (big eyes, heart shaped head with pointy chin, etc.). A wanted more of a western style, with a hint of caricature in the faces, and a consistent look from a basic <character> is wearing <outfit> at <location> type prompt.

Unfortunately, since I didn't really expect this to work, I didn't start taking notes on what all went into this until I was fairly well along. It started with a Mega Model (either SD's AIO or Clarity - I forget) and then I merged Rev Animated and Babes into that.

To my surprise, it knew LOTS of characters and celebrities (even some obscure ones that I was surprised about) and it spit out a pretty consistent style - even though it was still a bit more realistic than I wanted. It took LoRa's and style tokens well, though... so I moved to the next phase.

First I added several LoRa styles that I liked the looks of and tweaked the weights on each. They were styles that needed no triggers - so hopefully once I merged them in, then my default prompt would spit out that style every time. (Fingers crossed). I set the values so they showed up, but were low enough that the style didn't affect the clothes or other compositional elements much (if at all). The styles and the weights I ended up at are as follows: <LuisapNijijourneyLORAV2_v2:0.7>,<bootyFarmStyleLora_v10Dim128:0.4>, <lora:hungryClickerStyle_v20:0.6>

To my shock and amazement - it worked spectacularly. It wasn't quite the style I wanted, but adding simply: (bold lines:1.1), (high color), ((smooth skin)), (Masterpiece Digital Artwork:1.3), to the start of the prompt got me there. I can handle a 4 token seed prompt.

Some notes on the model:

• It "mostly" defaults to SFW out of the box. i.e. If you just type in a person's name or character description with no clothes - it'll tend to put them in clothes - or at least a swimsuit. That said... with proper prompting, does nudes and porny stuff just fine, too.

• I didn't care so much about making sure resulting characters look "exactly" like the person you're calling out so much as wanting to make sure that if I called out that name/person again - they'd be consistent in the way they look from image to image. So, in other words, it's not so much about "accuracy" as it is about "consistency" - and that seems to come through. (This works well for my needs to create stories and whatnot)

• Depending upon the distance from camera (i.e. Closeup head shots vs. full body shots) the weight of the "bold lines" in (many of) my example prompts can be tweaked. The further back, the lower that should go and the closer in, the higher it can go.

• This model (probably due to one of the LoRa styles I added) tends to want to add tattoos - even for those characters who don't have them. Adding that to the negative prompt can help (see below). Don't forget to remove those if you actually want the tattoos there.

• I've added some comments to several of the example images - especially the ones without generation data.

For the sample images I used the Anything VAE (though it works fine with both the default and Anime VAEs that I tried with it). No textual inversions or LoRa usage in them - it's all coming from the checkpoint. Base images at 384x512, Euler a at 20 steps, clip skip 2, face restoration, and scale fix (2x size using latent bicubic/antialiased). The first few examples are pure raw with just a character name and/or character name + general outfit description.

Those first few examples have no negative prompt, either (just to show how well it generally behaves without them). A good neg prompt helps with various things like extra fingers/bad hands, various defects and so on. The one I use most often is: (snagged from a sample image somewhere at some point)

canvas frame, 3d, ((disfigured)), ((bad art)), ((deformed)),((extra limbs)),((b&w)), blurry, (((duplicate))), ((morbid)), ((mutilated)), [out of frame], extra fingers, mutated hands, ((poorly drawn hands)), ((poorly drawn face)), (((mutation))), (((deformed))), ((ugly)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), cloned face, (((disfigured))), out of frame, ugly, extra limbs, (bad anatomy), gross proportions, (malformed limbs), ((missing arms)), ((missing legs)), (((extra arms))), (((extra legs))), mutated hands, (fused fingers), (too many fingers), (((long neck))), ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, mutation, mutated, extra limbs, extra legs, extra arms, disfigured, deformed, cross-eye, body out of frame, blurry, bad art, bad anatomy, Scribbles, Low quality, Low rated, Mediocre, 3D rendering, Screenshot, Software, UI, watermark, signature, distorted pupils, distorted eyes, (distorted face), Unnatural anatomy, strange anatomy, things on face, ((tattoo)), ((tattoos)),

Obviously, some of those may want to be removed if you're going for something that has been excluded.

The purpose of phobias is to elicit a sense of realism, which is achieved by collecting DNA samples from various iconic models. These models are particularly effective when creating close-up portraits.

Caution: Phobias contain NSFW contents. Use with care.

The global illumination of the subject matter is a prominent characteristic of Phobias, which is then complemented by the texture and diversity of subjects.

Variation:

Fig.1. As observed, 'Phobias_c' still has problems with the eyelash.

Fig.2. The comparison between Phobias_b and Phobias_Pro

Ancestors:

• chilloutmix_NiPrunedFp32Fix

• orangehardBasil05_v10

• sacrumProfanum_spv1

• level4_v50BakedVAEFp16

• cyberrealistic_v13

Additional Download:

Phobias_a
Phobias_b
Phobias_c

Optional: VAE

🤗 https://huggingface.co/Phobias/

Thanks for your ❤, feedback ,and image review. Sure! A cup of coffee is great.

Phobias care about your privacy via Binance.

American Commentator/News anchor Tucker Carlson - High quality Dreambooth training using JP's DB repo.

Put "Tuckercarlson Person" in your prompt. Use prompt weighting if necessary for example "Tuckercarlson person ((wearing a santa hat))" would put more weight on the santa hat part in the event the model is too strong.

Yes, it's a full checkpoint as tagged. If you like lora stuff and feel that's on par w/ DB for likenesses or have limited resources, this download probably isn't for you.

For best results use Hires Fix. For ultimate results use in tandem w/ 30-60% the steps modified by Controlnet w/ Canny, Openpose, or shuffle. Controlnet.

SAMPLES WERE GENERATED W/ CONTROLNET (if you don't understand what that means you will probably be disappointed).

I find it fun to put actual movie still frames in the CN input and use it for the first half or the generation. If you have a quick system try turning on Generate forever w/ 2 or 3 selected for the batch. While it's running you can change the prompts and CN input/settings and A1111 will update and continue churning out generations.

The beggining of a great project !

Providence - Prelude 

• A precise chosen noise offset setting

• A large variety of subjects, I took care to to keep the flavour at maximum from genesis seeds feel

• No baked VAE (works fine without any tbh)

Hyper Realism

For good results please use the forAnalogPortrait negative embedding.

Digital painting

For my digital painting finetune, please use “character portrait” along “digital painting” for enhance the quality

Analog

use the forAnalogPortrait negative embedding aswell.

Troubleshoot :

NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check.

A 2.1 model require either --xformers and / or --no-half start argument from the webui-user.bat

set COMMANDLINE_ARGS= --no-half --xformers

check also if your config file is named correctly alongside the model

Happy tuning !

romerorz.art - Twitter - DeviantArt - Gumroad

New, updated version of Tensorfunk.

Now there are 6 different major trigger words:

tensorfunk: Just generally everything this was trained on.

mj_v3: Midjourney ~v3 images (some v2 in there as well)

ddfusion: Disco Diffusion

jxstch: JAX guided Diffusion stitching

vqgan: VQGAN+CLiP

deeper_dream: deepdream images

Some other words that I used often in training that I think affect the model:

microcosms: I used this to describe the look that some older AI gets where each little nook and cranny is it's own way-too-detailed thing.
recursion: Similar to microcosms. Tried to gear it more towards "fractally" things.

Enjoy!