L.O.F.I: Limitless Originality Free from Interference

• 🧐 No special face alignment

• 🚀 Improve line details

• 🚀 Improve prompt understanding

[update:230414] LOFI V2.1

• 🧙more realistic details from more iterative training (200epochs)

• 🩹a little bit facial effect rollback (to LOFI V2pre)

• 🖌️add inpainting model (base on LOFI V2.1)

Note: because I used some of the optimization of torch2.1, it is possible that you could not generate exactly same image with same parameters (but quality was good)

[update:230325] LOFI V2

Finally, v2 is released

Based on the LOFI-v1 model, finetuned 80,000 steps / 300 epochs

• 📷more camera concept

• 🎨exact palette

[update:230224] LOFI V2pre

version 2 pre-release. This is a pre-release version, not just the merge model, it also includes some model parts that I have fine-tuned through training, including the previously released TAF and some other train models that improve image quality

But it may cause the model to focus on “portrait of characters”, so this is only a pre-release version and needs to be updated. LOFI model aims to be general and high-quality.

Prompt suggestions

• Since the text-encoder used is enough trained version, do not use a very high attention to control weight, which will cause some wrong drawing, it is recommended that all attention weights be no higher than 1.2

• If there is no special composition requirement, there is no need for a lot of negative prompts, such as "missing hands", which may affect human body drawing, and DeepNegativeis enough for negative prompts

• It is strongly recommended to use hires.fix to generate, Recommended parameters:

  • (final output 512*768)

    Steps: 20, Sampler: Euler a, CFG scale: 7, Size: 256x384, Denoising strength: 0.75, Hires upscale: 2, Hires steps: 40, Hires upscaler: Latent (bicubic antialiased)

Most of the sample images are generated with hires.fix

Note: that if you use hires.fix, you may not be able to reproduce the image with the same set of parameters in webui, because hires.fix introduces double randomness

busty girl with glasses and pink lingerie

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

You just need to write to write "BL2 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.

IronCatFateToon

Making models is a thing I do for fun. So if you like what I do? You can help me out by just using the model and leaving a review!

IronCatFateToon's goal is to be able to create nice looking anime style images with vibrant colours, more simplistic anime (cartoony) style look and with influence of Fate/Stay Night: Unlimited Blade Works styling.

In order to achieve what's just been promised I used MeinaMixas the base checkpoint, I then merged two of my style LoRA's into MeinaMix; Anime Vibrancy & Simplify Style (Merge ratio of 0.4`) & Fate/stay night: UBW Style (Merge ratio of 0.6`) . And finally I merged the result of that with the checkpoint Mistoon_Anime. Where I did a weighted merge, 70% was used from the results of me merging the LoRA's into MeinaMix. And 30% was used for Mistoon_Anime. (Note:I didn't use the exact same Anime vibrancy & simplifier Style LoRA when merging, I retrained on the same dataset from that LoRA for a more nuanced look that complimented MeinaMix better i.m.o.)

Some things you might recognize from Fate/stay night: UBW in this checkpoint, is in the way how eyes, scenery, city, cityscape, sky, grass, trees, certain 2.5D angles et cetera are generated. The anime/cartoony look is a combination from my Anime vibrancy & simplifier Style LoRA and Mistoon_Anime. While MeinaMix as a base gives in my own opinion a really sollid base for good looking art.

I haven't used any embeddings in the example images so you should get a good idea, in what the checkpoint can create on its own. I generated the images using some wildcard libraries, so the prompts might sometimes seem random.

Advice:

• I would recommended a clip skip of 2

• This checkpoint requires a VAE, I recommend kl-f8-anime2, I've also used it for the examples.

• I personally like lollypopas an upscaler a lot, it gives anime images a nice soft upscaled touch.

• I am personally a fan of using DPM++ SDE Karras with 15 steps and when Hires fixing, I like to use 10 for Hires steps and a denoise of about 0.35 or 0.4. Example images had 512x768 as base resolution.

This model wouldn't have been possible without these awesome creators:

• Mean for MeinaMix

• Inzania for Mistoon_Anime

This is a merge I made to try to get a good range of ability and overall quality with anime/hentai models.

The pictures use detection detailer (anime face) and most were upscaled using ultimate sd upscale with 4x-UltraSharp.

I used these models in the merge:
Abyssorangemix3AOM3_aom3a1b
AitiaInstantcafeNsfw_aitiaInstantcafeE
AlabasterBlend_v10
Counterfeit-V2.5
Degenerate_hassakuV1
FurryVixens_v20
GalenaBlend_v12

This is a pretty large merge based mostly on 526mix with various model merge alphas and then a final merge with a 1.5 trained model at 50 50. I make no assertions about what or what is not in this model. It was just a fun project and seems to do well at mostly anything as long as you prompt for it correctly.

Optimized the previous V1 version, it is a bit too watercolor, a bit girly, I don't like it.

This version is closer to the riso I want, but could be better.

The first thing to say is that this version is very compatible. You can draw a lot of content you want to try, including people, animals, and objects. Of course, I don’t recommend you to use it to draw dinosaurs.

If you have friends who have successfully drawn dinosaurs, please share them in the comment area.

In terms of tags, the same as the VI version, I suggest "Slight retro color, flat, grainy, riso print style, illustration, cute, white space, low saturation," and the rest are free to play.

at last,

To become a golden alchemist is my generation.

I wish all fellow Taoists good luck.

chilloutmix + lora

EnvyMix is an anime and semi-realistic model that is a cosine merge of several merges of high quality models. It's reasonably good at hands, it does nice backgrounds, and it can handle LoRAs without looking overcooked a lot better than most other models.

The model was trained based on John Blacksad.

The bottom model is tmnd, which replaces the painting style of the original model, and then retains the color and detail processing. This is the first version. There may be flaws. Welcome to give me feedback. I strongly recommend that you open high-resolution repairs and output pictures. High-resolution pictures Turn on high-resolution and turn on high-resolution repair when outputting the image. The sampling method is recommended DDIM, and the number of sampling steps is recommended to be more than 50 and less than 50 steps. This sampling method will lose details, and the redrawing range is 0.2-0.3. The zoom-in scheme: R- ESRGAN 4x+ Anime6B, redrawing and sampling: 20 steps, the follow-up image generation is enlarged with the plug-in SD upscale, the equipment is good, and the equipment is generally not recommended.

The bottom model is tmnd, which replaces the painting style of the original model, and then retains the color and detail processing. This is the first version. There may be flaws. Welcome to give me feedback. I strongly recommend that you open high-resolution repairs and output pictures. High-resolution pictures Turn on high-resolution and turn on high-resolution repair when outputting the image. The sampling method is recommended DDIM, and the number of sampling steps is recommended to be more than 50 and less than 50 steps. This sampling method will lose details, and the redrawing range is 0.2-0.3. The zoom-in scheme: R- ESRGAN 4x+ Anime6B, redrawing and sampling: 20 steps, the follow-up image generation is enlarged with the plug-in SD upscale, the equipment is good, and the equipment is generally not recommended.

Welcome to the exciting world of LewdPunk SKY Edition - an innovative stable diffusion model that brings you the best of nude, skimpy clothing styles and futuristic sceneries in realistic 3D or photography style with different lighting modes. Immerse yourself in an exquisite world of beautiful, captivating images and explore the endless possibilities of creativity with our special blend of models. Take a look at our sample images and see for yourself the amazing results we can offer. Try Lewdpunk SKY Edition today and unlock the potential of your imagination!

Gaping/Large Insertion model

This is not my model, it was taken from hereand made according to the instructions, only instead of the usual Anything was taken V4.

gaping/large insertion/other model trained on danbooru, gelbooru, sankaku: all of gaping anus,gaping pussy,large insertion, huge dildo, fisting, prolapse, all the way through + some others. based on NAI nsfw.

prefer gelbooru (NAI) tags as the main ones were mapped when training on sankaku data, namely use gaping pussy instead of gaping vagina.
add letterbox in the negative prompt, all letterbox'd images were tagged in training so this helps a bit.
danbooru tags are a subset of gelboorus, for example danbooru only has gaping, where as gelbooru has gaping, gaping pussy and gaping anus

AnythingGape
Diff merge of Gape60 and AnythingV3. Much easier to use than Gape60 and more aesthetic! This is all I use these days.
Construction is: AnythingGape = Anything + (Gape - NAI)

It's a model mostly trained on free range non artist sourced women. Some men. I tried like contrast here with metallic metally things looking metally and dark thing things looking darkish. Mostly it's on women tho cause I like making pictures of women personally so there you go.

All test images are made with no embeddings. But of course those help so use the ones you like as many are out there.

This is a SD 2.1 768 model. I use it in the automatic 1111 web gui. You should not need a yaml with the updated version but should you then rename a yaml to the checkpoint name.

SD 2.1 stuff people have problems running and some vodoo can be needed. I run my batch file like this:

set COMMANDLINE_ARGS= --no-half --xformers --no-half-vae --disable-nan-check

That is the "bat" file and the commandline I use.

To use type a string of words in the box and see what happens.

Also does some NSFW. type nsfw in the little box with the words.

Merge credit dont credit whatever

SR-Omega is a merge of a few handpicked checkpoints/mixes which aims to help generate semi-realistic images.

This was supposed to be an update to SR-Alpha, but the style of this merge's outputs was way too different to be classified as such, as, in my opinion, the images generated with this merge have drastically more realistic proportions compared to the older model.

I highly recommend inpainting for best results and to unlock the full potential of the merge. Separate inpainting model not required. Face restore (GFPGAN) has been used to just bring out a bit of realism to the outputs. I have not tested using Codeformer as it doesn't seem to work on my current set-up (3050 - 4GB VRAM). This does not have a significant impact when inpainting is integral to the workflow, but it does allow for better outputs from raw txt2img, though your mileage may vary.

Recommended Settings:

• Sampling Method: DPM++ 2M Karras (txt2img), or Euler A (inpainting)

• Sampling Steps: 30

• CFG = 10

• Denoising Strength = 0.3-0.5 (img2img/inpainting)

*I only use denoising strength values greater than 0.4 if I want to drastically change the appearance of a particular feature/region of the image. I mainly use this if I want to fix hands that seem to be too deformed to be saved with lower denoising strength values.

Suggested Embeddings:

• https://civitai.com/models/17083/bad-picture-negative-embedding-for-chilloutmix

• https://civitai.com/models/11772/verybadimagenegative

• https://cdn.discordapp.com/attachments/1032948846197747731/1069660323709190195/bad-hands-5.pt

*Orangemix VAE baked in

Credits to the respective original creators of the resources used.

original description

Anime版本 不做对比了 就是更平面和动漫风格 看看例图就好了

-2.0的一些版本提升在于 强化了人物皮肤衣服的透亮感，强化了暗部的细节，并稍微加强了一些手部关节 质感提升更多

1.5的一些对比 质感光影提升 色彩对比很棒

请注意 我的负面词汇并不是单纯的词汇 而是一个个embeddings 你们可以在这个网站下到这些词汇并在负面提示词中用上 可以帮助你们达到我想要的效果

Anime version does not compare, it is more flat and anime style, just look at the example picture

Some contrasting textures of 1.5 enhance the color contrast and are great

Please note that my negative words are not pure words but embeddings one by one. You can download these words from this website and use them in negative prompt words to help you achieve the effect I want

This mix is more focused on (un)realistic creatures/monsters, merging concepts, but seems to be a quite versatile/creative base model. Can be merged with more specific models to steer its direction.

Check out WyvernMix for realistic portraits, or SuccubusMix for more nsfw portraits.

You can see news about Model and my other stuff on my Twitter: https://twitter.com/Amoralchiklofi

UPDATED 04/13/23:
v3: marge of revAnimated, cetus, lainmixv22, darksushimix

UPDATED:
Lainmixv: better as v1, same size, little changes, and a new version of cetusMix

Lainmixv22: smaller than v1 and xv according to the version of cetusMix, also I not using revAnimated_v11 again, and change it to viewer-mix_v1.7

And for both, I added my custom orange mix with Grapefruit41 and oldfish_v1.1

v1 and xv still need a vae from pastelmix, v22 baked with vae

Recommend:

• Clip skip: 2

• Sampler: DPM++2M Karras

• Steps: 20+

• CFG scale: 4-8

• Highres.fix: SwinIR_4x Hires / Latent

• steps: 10+

• Denoising strength: 0.4+

• Upscale by: 1.5+

Nicknamed: C3

Credit to justmaier

This model is a mix of these:

In no particular order:
https://civitai.com/models/7279/colorful

https://civitai.com/models/9114/consistent-factor

https://civitai.com/models/5062/clarity

Clarity:V2

ConsistentFactor:V4.0

Colorful:V2.6

Some of the images might use embeddings or Loras.

If you want more creative results, go for Clip Skip 1. On the other hand, Clip Skip 2 is your best bet for consistent outcomes. As of version 8.0, Clip skip 3 is just as (if not more) reliable as Clip Skip 2.

UseThisVAE if you have a desaturated image.

Info:

You can use this model for everything, and I mean EVERYTHING. Sell it, send the file, use it in generation services, and do what you want with it. We are an open-source community, and I want to embrace that fully!

I forgot to keep track of the models used in this process, so I don't know which ones were merged. However, starting with version 2, I'll make sure to keep a record and update it here.

Concept transplanted model

Concept 'vivid color, linestyle' from

TMND-MIX by gxf159357233

to

BlueMix by Happy 930126496

Sample images are generated using the vae attached with this model. kl-f8-anime2 is a bit too strong for my taste.

original description

Golden mirage,海市蜃楼第一弹,又称黄金答辩💩幻象,炼金术的尽头是金色海市蜃楼

vae:stabilityai/sd-vae-ft-mse-original · Hugging Face

mirage系列：

Golden mirage, the first bullet of the mirage, also known as the golden answer 💩 illusion, the end of alchemy is a golden mirage

vae:stabilityai/sd-vae-ft-mse-original Hugging Face

Lunar Diffusion 1.26

This is the first public release of mooncryptowow's Lunar Diffusion model. This model is a high quality blend based off 1.5 and was lovingly crafted to create the most detailed images possible while retaining a large degree of creativity in the renders.

For best results render your images at sizes higher than 512x512. All preview images were created at 576x704 x 2x Swin2sr hires fix with extremely simple prompts on a locked seed (no cherry picking).

Model contains a baked "vae-ft-mse-840000-ema-recommended"

Check me out on Instagram at: https://www.instagram.com/mooncryptowow/
Facebook at: https://www.facebook.com/profile.phpid=100087374733528

I also have a Patreon at https://www.patreon.com/mooncryptowow where I share all of my prompts and settings, uncensored art, custom models and files, as well as general knowledge and tutorials.

This model falls under the RunwayML Stable Diffusion 1.5 license which you can review here: https://huggingface.co/runwayml/stable-diffusion-v1-5

The AI virtual model trained based on the prototype of Dilmurat, all the pictures appearing in the account are generated by AI software Stable Diffusion, and the characters in the pictures do not exist in reality. Any theft or reproduction of the pictures from this account for fraud or other commercial purposes is the responsibility of the infringer.

以 Dilmurat 为原型基础训练的 Ai虚拟模型，账号中所出现的所有图片均由ai软件Stable Diffusion生成，图片中人物非真实存在。任何盗用或搬运本账号图片用于诈骗或其它商业用途的，责任由盗用者本人承担。

Anime style focused blend. Generates images somewhere between 2d and 2.5d, but it is capable of both with the right prompts.

This blend has been made by merging a few different anime and 2/2.5d models to various degrees.

Its strength lies in fantasy but it is also capable of more realistic or futuristic settings.

Highly recommended to use the following negative textual inversion:

• badhandv4

• bad-artist

• bad-artist-anime

• bad_prompt_version2

VAE has been baked in.

The examples were made by a combination of txt2img and img2img, inpainting and upscaling techniques.

Model tends to be slightly perverted when it comes to female characters. It is highly capable of nsfw. Keep nsfw tags in the negative prompt to prevent.

Thx to Alex Hogrefe who is a wonderful artist in architecture visualization, and I trained this model based on his works to help us all can get the beauty from him!

Enjoy and this is Alex's page: Visualizing Architecture | by Alex Hogrefe

This is the model where i focus on the faces and body shapes as well as the skin tone according to the asian aesthetic like me. As an easy to use model with simple prompts, but with great emphasis on negative prompts, so it's best to copy my negative prompts to get the same results as in the sample image.

please support me by becoming a patron:

patreon.com/duchaitenreal

illustration, 3d, 2d, painting, cartoons, sketch, (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, vaginas in breasts, ((monochrome)), ((grayscale)), collapsed eyeshadow, multiple eyeblows, (cropped), oversaturated, extra limb, missing limbs, deformed hands, long neck, long body, imperfect, (bad hands), signature, watermark, username, artist name, conjoined fingers, deformed fingers, ugly eyes, imperfect eyes, skewed eyes, unnatural face, unnatural body, error, bad image, bad photo

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

This CKPT mainly emphasizes a bright and refreshing painting style. Compared with oil painting and impasto models, the area of ​​the color block representing the shadow is relatively controlled, and the picture is neat and clean, which is very suitable for the display of fashionable clothing in the example picture.

Notice:

特别感谢zap无私分享的的LoRA模型 fashion girl使我得到了这样的画面效果，由于这个CKPT已经融合了fashion girl，Please do not add it again during use, otherwise it will cause overfitting

This CKPT mainly emphasizes a bright and refreshing painting style. Compared to oil painting and thick paint models, it controls the area of the color blocks that represent shadows, making the picture clean and neat. It is very suitable for showcasing fashion clothing in example images.

NOTE:

Special thanks to zap for sharing the LoRA model "fashion girl" selflessly, which has enabled me to achieve this kind of visual effect. Since this CKPT has already incorporated the LoRA, please do not add it again during use, otherwise it may cause overfitting.

This model can generate images at 1600x704 with no highres upscaling required.

Use with Automatic1111 or InvokeAi on RunDiffusion where you can generate your favorite citizens models.

This merge was created to achieve clear and unrivaled image quality, color and sharpness; no more blurry or foggy filter over your image.

Join Civitai-Crystal beta on discord!

Join RunDiffusion on discord!

EXTRA:

Most of the images are rendered at 768x1216 with NO HighRes Optimization and NO Upscaling of any sort. The faces were not adjusted in any way. These images are raw results. As you can see, the model handles the prompt very well even at large sizes.

VERSION 1b:

In overly colorful pictures some details get lost in the intensity. Toned down the color intensity a little for a clearer and crisper image. Also, there are now extra details in your renders when compared to the previous version of Crystal-Clear.

Try your LORAs through this model and post your results.

Feel free to compare with other models.

Don't forget to post your results!

Model merged with intent to balance it's behavior between neutral and fairly stable, yet creative and slightly different from the "normal" sd/nai results.

As for prompt style, Nai's "masterpiece, best quality" looks like a good start.

Introduction

In the world of anime, style and aesthetics play a huge role in creating memorable and captivating characters. From bold and vibrant colors to intricate and detailed designs, there's no limit to the creativity that can be expressed through anime. In this model, we embrace the anime style and bring it to life with our own unique spin. Each character is carefully crafted to embody the essence of the anime universe, with exaggerated features and dynamic poses that capture the imagination. So if you're looking for a model that truly captures the spirit of anime, look no further than our collection.

How to install

1. Download and install Stable Diffusion, I personally suggest the executable version by AUTOMATIC1111.

2. Download Falkons model in .safetensor format and insert it into the "ckpt" folder.

3. Start Stable Diffusion.

4. Select the "Falkons.safetensor" model from the "Stable Diffusion checkpoint" dropdown menu.

How to use and examples

Of course you can try it with the prompt, sampling methods, steps, CFG, etc. you desire, I use to generate images with:

• Sampling methods: Euler a, DPM++ SDE Karras or DDIM

  • Euler a for a more toon effect

  • DPM++ SDE Karras for an higher realistic effect

  • DDIM for faster generation but higher risk of artifacts and "strange things"

• Sampling steps: I usually set a minimum of 20 steps, more steps often help to generate a more detailed image, but it's not always true, and typically there is less difference as the number of steps increases.

• Restore faces, Tiling and Hires. fix turned off

• Dimensions: I usually pick 640x960, but I have noticed good results with other resolutions such as 512x512, 512x704, 512x960, 624x1176, 720x1024, 1024x1024, while I have experienced poor results with ultrawide resolutions

• Batch size: my GPU can handle a Batch size of 5, which means 5 images generated per run, it really depends on your GPU's VRAM, so find the highest number you can handle

Here are two examples:

You can start with very easy prompt, such as:

Prompt: a girl,
Negative prompt: (worst quality, low quality:1.4), watermark,
Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 2118282899, Size: 640x960, Model hash: 460b71c1c2, Model: Falkons

Or you can try something more detailed (for example if you want to generate the same character in different poses or in different outfits):

Prompt: masterpiece, best quality, best resolution, girl, nude, perfect anatomy, tight, fantasy worlds background, fantasy, determined, power, flames, intense,night sky, outdoors, moon, stars, clouds, wind, trees, night, sorceress, witch, profession witch, witch uniform, with heels, summoning, glowing runes, neck ribbon, hair ornament, heart jewelry crystal, lace feather white frills, gloves, thighhighs, skirt, thighboots, neckline, extrovert personality, favourite color blue, blue eyes, blue hair, short hair, large breasts, small waist, shiny skin, visible nipples, showing boobs, crotch tattoo, tattoo on underbelly, pussy juice, looking at viewer, magic circle, standing in the center of the circle,

Negative prompt: worst quality, low quality, EasyNegative, comic, sketch, not masterpiece, not best quality, not best resolution, poorly drawn, bad anatomy, wrong anatomy, sketches, mutated, mutilated, monochrome, grayscale, greyscale, logo, watermark, duplicate, artifact, noise, film grain, blur, blurry, deformed, distorted, disfigured, fused fingers, extra fingers, twisted, malformed hands, ((((mutated hands and fingers)))), deep depth of field, simple background, abstract background,

Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 648403476, Size: 640x960, Model hash: 3624a75e34, Model: Falkons

Use this model

You can use this template for commercial purposes, I would just prefer not to see it being sold directly and to be provided with the URL of this page. Thank you!

There is a problem with the file just now, it has been uploaded again

The content is no different from version 1.0

GOOGLE：

There is a problem with the file just now, it has been uploaded again

The content is no different from version 1.0

Fusion model:

Anything V5

fallengel

basil_mix

animePastelDream_hard

museV1_v1

Del Toro has long been an inspiration to me since I first saw his 2006 masterpiece "Pans Labyrinth". A fantastic fusion of horror and fantasy, fully displaying his creative genius and unrivalled to this day. As a tribute to his work, I decided to take his freakiest creation, "The Pale Man" and try to capture its entirety in dreambooth. Built on 55 images taken from a lifesize model, a lifesize bust and some denoised screenshots. Trained at 768 resolution for 5k steps with 2k class images.

I don't expect a lot of interest in this sort of freakish model but, I will be happy to see it reach even just a few like minded fans of fantasy macabre and thought it was worth sharing here.

I've spent countless hours optimising my workflow for this type of full figure checkpoint in dreambooth and I have quite a few models I will now start releasing on a weekly basis.

A model that I trained over ~500 portrait photographs merged with my favorite models for versatility; Hassanblend, Aeros, RealisticVision, Deliberate, sxd, and f222.

Happy generating!

huggingface repo here

If you like the model, consider supporting on Patreon / Coffee / Paypal

Japanese is in the second half

Chrysanthemum Mix

This is a 2.2D style model with a slightly more anime-like feel, based on Alstroemeria Mix.

By adding anime, illustration to Positive and realistic to Negative, you can achieve a more illustration-like expression.

Models used for merging

The following models were used for merging.

I express my truly gratitude to all the authors of these models and their ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

Additional information

The high-poly LoRA used for generating sample images can be downloaded from the link below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the actual values of the LoRA Block Weight used for generation are as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

• NP: 1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0

Please also check out sister models:

• Alstroemeria Mix

• Bougainvillea Mix

Chrysanthemum Mix

Alstroemeria Mixをベースに少しアニメ調に寄せた 2.2D スタイルのモデルです。

Positive に anime, illustration, Negative に realistic を入れることで、よりイラスト風の表現ができます。

model used for merging

I used the model below for the merge.

Thanks to the authors of each model and all its ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

others

The high-poly LoRA used to generate the sample images can be downloaded below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the real value of the LoRA Block Weight used for generation is as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

• NP: 1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0

Thank you very much for your sister model.

• Alstroemeria Mix

• Bougainvillea Mix

Japanese is in the second half

Bougainvillea Mix

This is a model with enhanced contrast, based on Alstroemeria Mix.

It is suitable for night scenes and strong, cinematic lighting with heavy shadows.

Models used for merging

The following models were used for merging.

I express my truly gratitude to all the authors of these models and their ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

• Lyriel

Additional information

The high-poly LoRA used for generating sample images can be downloaded from the link below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the actual values of the LoRA Block Weight used for generation are as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

Please also check out sister models:

• Alstroemeria Mix

• Chrysanthemum Mix

Bougainvillea Mix

Alstroemeria Mix をベースにコントラストを強化したモデルです。

Suitable for night scenes and cinematic lighting with strong shadows.

model used for merging

I used the model below for the merge.

Thanks to the authors of each model and all its ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

• Lyriel

others

The high-poly LoRA used to generate the sample images can be downloaded below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the real value of the LoRA Block Weight used for generation is as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

Thank you very much for your sister model.

• Alstroemeria Mix

• Chrysanthemum Mix

Japanese is in the second half

Alstroemeria Mix

This is a 2.5D style model based on my beloved model doll774 and merged with others.

Models used for merging

The following models were used for merging.

I express my truly gratitude to all the authors of these models and their ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

Additional information

The high-poly LoRA used for generating sample images can be downloaded from the link below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the actual values of the LoRA Block Weight used for generation are as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

Please also check out sister models:

• Bougainvillea Mix

• Chrysanthemum Mix

Alstroemeria Mix

最愛モデル doll774 をベースにマージした 2.5D スタイルのモデルです。

model used for merging

I used the model below for the merge.

Thanks to the authors of each model and all its ancestors.

• doll774

• Camellia25D

• Fantastic mix

• dalcefo painting V3

others

The high-poly LoRA used to generate the sample images can be downloaded below.

https://civitai.com/models/8730/hipoly-3d-model-lora

Also, the real value of the LoRA Block Weight used for generation is as follows.

• NCNF: 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,0,0

Thank you very much for your sister model.

• Bougainvillea Mix

• Chrysanthemum Mix

StarryLie For Large Image Size

Fine tuning training based on RAT5 model

[2023/4/6]

• StarryLie model update

  为获得最佳效果，请确保您的最终分辨率> 1568x1568。这对于生成好图是必要的，并且可以使用Hires.fix。

  For best results, make sure your final resolution is greater than 1536x1024. This is necessary to generate good graphics and can be used with Hires.fix.

• It is recommended to use the VAE uploaded here

  VAE is necessary to download, I believe 99% of people do not have this VAE, so it is recommended to download it

  VAE is necessary to download, but most people should not have it, so it is recommended to download it

• The effect of this model will be poor if you run a smaller picture, so it is not recommended to use a too small picture and a magnification of less than 2 times.

  This model runs poorly on smaller images,Therefore, it is not recommended to make images too small and magnification less than 2x.

Special thanks:

• GitHub - bmaltais/kohya_ss：训练UI

• Automatic1111: SD-WebUI

• Danbooru: marker, help standard training set

• ClearVAE：Channal版本替换的VAE

• Elemental Codex Authors Group: Here it is that got me thinking

Have fun everyone <3

• Feel free to merge this model elsewhere, but if you share the merged model, don't forget to tag it

  We allow you to merge with another model, but if you share that merge model, don't forget to add me to the credits.

A model that generates B-end style elements, which needs to be used in conjunction with the corresponding version of DDicon lora.

The V1 version has more details, and the v2 version is more concise, please choose according to your needs. Used in conjunction with DDICON_lora.

A model for generating B2B style elements is required to be used in conjunction with the corresponding version of DDicon lora.

The V1 version has more details, while the V2 version is more concise. Please choose the version that fits your needs.Used in conjunction with DDICON_lora.

https://civitai.com/models/38558/ddiconlora