Huggingface: https://huggingface.co/sheldonxxxx/MechaDream
Support me on Ko-fi to help me buy more GPU time: https://ko-fi.com/techadventurex
This is a all in one checkpoint (base model + lora + vae) for convenient use.
Please refer to https://civitai.com/models/47025 for the Lora release
Sampler: UniPC (Personal preference)
CFG Scale: 15
Negative Embeddings: EasyNegative, bad-hands-5
Lora Weight: 0.8
Hires.fix with scale >= 2 is highly recommend
Step 1: Batch generate images
Step 2: Select the best image and inpaint any flaws
Step 3: Upscale using MultiDiffusion
🡳 Some comparison images between original image and after the above workflow 🡳
MechaDream v1 will be the base model for generic mecha image generation
New Lora models train on different anime Blu-ray will add flavors into my MechaDream
List of WIP anime:
Gurren Lagann (Dataset is ready, will start training when I'm less busy)
Code Geass
Macross Frontier
Evangelion
Genesis of Aquarion (Maybe)
Valvrave
Eureka Seven (Maybe)
Gundam
Armored Core
Making models can be expensive. Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕ Ko-fi
I warmly welcome you to share your creations made using this model in the discussion section. If you encounter any issues, please feel free to add a comment😊, and both the community and I will do our best to help and provide solutions❤.
Resolutions: 512x768, 512x1024 for Portrait!
Negative (worst quality:1.5), (low quality:1.5), (normal quality:1.5), 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, painting by bad-artist
more easy, more simple
1 hour to train
5 mins to gen
please use 840000 vae
my test model
Fran
I made a reasonably good merge model, so I'm going to publish it.
I need your help 🥰🥰🥰 please support me! 😘
https://ko-fi.com/uncle_emilia
very pretty and beautiful female
beautiful background
NSFW
Sampler : DPM++SDE Karras
Hires. fix ON
Upscaler : SwinIR 4x
Upscale by : 1.5 ~ 2
Denoising : 0.3 ~ 0.5
VAE : vae-ft-mse-840000-ema-pruned
1girl,
(finely detailed beautiful face),(8k eyes),(8k pupils),
(photorealistic:1.4), masterpiece, ultra quality, insanely detailed, (ultra high res), absurdres,(Unspeakable-Horrors-Composition-4v), (EasyNegative:0.8), (bad_prompt_version2:0.8), bad-hands-5, (low quality, worst quality, normal quality:1.7), sketch, illustration, (monotone:1.2), (monochrome:1.2), (grayscale:1.4), watermark
First of all, thanks for all the downloads and likes.
The goal of this v2 update was:
Improving the age issue
Improving eye clarity
Improving backgrounds
Not breaking too much
If you are switching from V1 to V2 and want to try your old images with the new checkpoint:
Please try out using: Extra -> Variation seed: -1 -> Variation strength: 0,1–0,2
Under normal circumstances, a variation of your old image should be possible to recreate.
Feedback is always appreciated. New Discord thingy here.. still don't really know how it really works, but if your interested or need help:
Further infos down below!
Within the last few weeks, I really enjoyed discovering stable diffusion, and because of this community, I really got great results out of it.
But as someone who likes to tinker around and personalize things, I wanted to try creating my own mixes and loras.
And because this great community has given me so much, I wanted to give something back.
Depending on the overall feedback and interest, I will update and upload further mixes and loras in the future.
Keep in mind, this mix is made with NSFW related content in mind!
Prompt: best quality, masterpiece
Negative: (worst quality: 1.4), (low quality: 1.4),
* Adjust the strength and prompts to your liking, but keep in mind that the negative prompts are very important!
** Use (nsfw:value) positive/negative prompt as needed.
*** Consider adding age tags like: mature female, teen etc.
Most samplers work pretty well, so try out what works for you and pick your favorite.
I'm personally a huge fan of DDIM because it's very fast and I can test prompts and seeds very quickly and make adjustments without much delay.
My "personal" basic settings:
Sampler: DDIM
Steps: 30
Clipskip: 2
CFG Scale: 10
VAE: Included in Checkpoint
HiRes fix: R-ESRGAN 4x+ Anime6B
HiRes Steps: 25
HiRes Denoise: 25
Upscale by 1x for prompt or seed testing and higher for final results.
Sometimes they can help your results a lot, but they can also screw you over. So try adding or removing them from your prompts for different results.
Most of them should work out pretty well, but I would suggest to start with low values like 0.1/0.2 start increasing until you get the result you want.
Unfortunately, I did not keep track of every resource author I used when I first started mixing and testing... Therefore, credit and huge thanks to all Civitai creators!
If you like my mix it would mean a lot to me if you leave a like, support me with a ko-fi or share your results / prompts with this great community!
More easy, More simple
Good day, everyone!
Thank you for your interest in my model, this model is based on my previous model "Pika's Animated Mix"
But what's new?
I trained this model on a style which is very close to the style of Niji V5, it has a strong cinematic effect and requires very grammatic input prompt to get quality images!
Also the model is very flexible in style, try it yourself!
The model comes without VAE!
Here are my recommendations:
Sampler: Euler a, DPM2 a Karras, UniPC
Steps: 28-50+
CFG: 4.5-8
VAE: Any Anime VAE you like.
Clip Skip: 1,2
Negative prompt: (easynegative:1.1), (verybadimagenegative_v1.3:1), (low quality:1.2), (worst quality:1.2)
Enjoy using it and be sure to share your results!
P.S: I hope your results are better than mine! :D
Looking forward to your feedback <3
more easy, more simple
Vox-mix
Vox-mix is a chekpoint optimized for creating images influenced by Victorian and Art Nouveau styles.
Model Description
The model was created by merging Well-known models. (Waifu Diffusion, Novel AI, Anything 3.0, etc)
A cup of coffee : https://ko-fi.com/voxmix
The fleeting radiance, like a momentary brilliance in the fleetingness of life, passes by in a flash, yet leaves an indelible impression deep in the soul
Chinese:
一个泛用的，比较清爽的二次元模型（自认为）
在融合过程中，我追求细节，结构与稳定
不过，请注意，FR在融合过程中，从未、以后也不会使用到2.5D和真人模型，
这是一个纯粹的二次元模型
我推荐使用最新版本的FRMiX，因为在迭代过程中，我尽力让新版本表现得比先前更好
不过，如果你喜欢先前某一版本的画风，也可以下载对应版本
在设置上，我推荐高分辨率，竖图（如600*960），高清修复（算法我常用4x-UltraSharp和R-ESRGAN 4x+ Anime6B，重绘幅度0.3-0.4）
采样器我推荐 DPM++ SDE Karras，步数30-60（我常用60，细节更到位）
提示词相关性5-7
vae我使用的是clearvae，不过我并没有将它整合入模型，你可以选用你喜欢的vae
如果你需要fp16版本，请告知我（否则我仅会在最终版发布时同时发布一个最终版的fp16版本）
希望大家能够反馈自己生成的图片，这可以帮助我改进FR，谢谢大家！
我的bilibili（欢迎私信）：
https://space.bilibili.com/351739534
English:
My English level is very poor, so I had GPT translate this for me:
A general, relatively refreshing 2D model (at least I think so).
During the fusion process, I pursue details, structure, and stability.
However, please note that FR has never and will never use 2.5D or human models in the fusion process.
This is a purely 2D model.
I recommend using the latest version of FRMiX, because I have tried to make the new version perform better than the previous one during the iteration process.
However, if you like the style of a previous version, you can download the corresponding version.
For settings, I recommend high resolution, portrait orientation (such as 600*960), and high-definition repair (I often use the algorithms 4x-UltraSharp and R-ESRGAN 4x+ Anime6B, with a Denoising of 0.3-0.4).
I recommend using the DPM++ SDE Karras sampler, with 30-60 steps (I often use 60 for more accurate details).
The CFG Scale should be 5-7.
I use clearvae for VAE, but I haven't integrated it into the model. You can use your favorite VAE.
If you need an fp16 version, please let me know (otherwise, I will only release an fp16 version with the final version).
I hope everyone can provide feedback on the images they generate, as this can help me improve FR. Thank you!
FRMix 1.0 version(Attached with seven rendering images)：
GracefulMix is a model for making mature and elegant asian women.
Ideal for portrait-oriented image generation. NSFW is not known.
You can use this model as a material to merge your own models.
2023.4.21
我将胶片风Lora小比例融合进了Fantasy模型中，优化了此模型下皮肤的质感。但请注意，要获得完美的胶片质感，仍需要搭配胶片风Lora使用，只是使用比例可以降低0.1~0.15左右。
Fantasy 2.0模型更偏幻想风格一些，Reality 2.0模型更偏写实风格一些。两个模型质量都不错，大家可以根据需求下载使用。
2023.4.20
更新了reality2.0版本，这个版本相比之前两个版本，较大幅度提升了皮肤和样貌的真实感。建议在生成人像时，搭配胶片感Lora（1girlmix）使用，真实感更佳。
同时最重要的，新模型搭配兔狲Lora使用，生成兔狲逼真图像的成功率大幅提升！（没错，我搞这个模型，其实主要是为了训练猫猫）
chilloutmix模型与本模型的reality版本、fantasy版本的对比：
Comparison between the chilloutmix model and the reality and fantasy versions of this model:
我比较推荐用reality版本的模型，在色调、颜值和真实性上比较平衡。建议将clip skip设为1，这样生成的图会更美观。
I recommend using the reality version of the model, which is more balanced in tone, appearance and authenticity.
本模型由多个checkpoint模型合并而来，具体有哪些我也记不太清了。以下是MoonMix与chilloutMix模型在生成人像时的一些对比。总的来说，MoonMix相比chilloutMix，真实感和锐度有所下降，但美感与丰富度有所提升。
This model is the result of merging several checkpoint models. I can't remember exactly which models there are. Here are some comparisons between MoonMix and chilloutMix models in generating portraits. In general, MoonMix is less realistic and sharper than chilloutMix, but the overall beauty and richness is improved.
演示样例中所使用Lora模型“1girlmix”可以在此处获取：https://civitai.com/models/33208/lora
The Lora model "1girlmix" used in the demo example is available here: https://civitai.com/models/33208/lora
Hello friends,
after midjourney release new niji style expressive, i really fall in love with this style and it make really really good 3D model. So I had make a new model to try mimic as much as i can to the expressive style of niji
This model was trained with about a few thousand images I made with niji
It can generation mostly in the style of niji expressive.
you can use it with pretty much all Lora
For negative prompt:
Deep Negative
EasyNegative
are the best embeddings i suggest everyone should use it with this model
Pls give me some feed back to fine tune this model if you can
3D model
This model is a merge of Lodestones "Fluffyrock" model (fluffyrock-576-704-832-960-1088-lion-low-lr-e17) with a 3.6% addition of a merge of Yiffye18 and AnyEasterSnack V1. It is geared towards broadening the styles and Results possible in the very good Fluffyrock model without losing too much of said quality.
I am not Lodestone and would advice all to check his original work at https://huggingface.co/lodestones or on the Furry Diffusion Discord https://discord.com/invite/JKS7UttReS
As its mostly their model, with only a minor tweak, i would give all credit to him, with derpyness being caused by my modification.
配合米山舞LoraYoneyama Mai Style | Stable Diffusion LORA | Civitai有比较好的效果，是自己闲的没事练的模型，不喜勿喷。
建议搭配EasyNegative | Stable Diffusion TextualInversion | Civitai
欢迎大家使用
Cooperating with Lora Yoneyama Mai Style | Stable Diffusion LORA | Civitai has a better effect. It is a model for myself to practice when I have nothing to do. Don’t spray it if you don’t like it.
It is recommended to use EasyNegative | Stable Diffusion TextualInversion | Civitai
welcome to use
Babylon clothes are called in Arabic (الازياء البابلية) Dream Booth model trained with Stable Diffusion, we can now create artificial intelligence art generation images using trained images. In this model, we can create images of women wearing clothes called Babylonian fashion style emanating from the Babylonian civilization in the country of Iraq, it is popular clothing for Babylonian women in ancient times as famous images, or anything you can think of Test the concept via A1111 Colab fast-Colab-A1111
Works well with simple and complex inputs! Use (nsfw) in negative to be on the safe side!
Work in progress , a morerefined version may appear soon... I hate the be the one but I am making very little doing this, but want to be doing way more, please help me rent GPU time...
https://www.patreon.com/nucleardiffusion
or please consider an easy one off donation here: https://ko-fi.com/nucleardiffusion
Notice: This free work is intended for art and illustration purposes of emergent technology and not intended nor' trained for making illicit images, as the author I've taken care to not use any private data in the training process or designed it for such purposes. By downloading this model you are agreeing with the aforementioned intent and that you will not use my work to create / share images of a potentially immoral and / or illegal nature.
This model gives you the ability to create whatever you want.
Attention, the model requires VAE from Stability AI: vae-ft-ema-560000
or you can use VAE from Stability AI: vae-ft-mse-840000
NSFW Art Designers
Character designers
Professional prompters
Art designers
I hope you enjoy the results of my efforts.
Thank you very much XpucT. My first knowledge came from this man.
This model is based on his Deliberate V2 model.
I don't know why, but this model becomes clearer or simpler depending on the prompt.
MegaDistortedSerenity_V2... Why V2? Because I've iterated too many times to keep track of and my naming scheme is a mess. This is a continuation of model merging based on way too many models, but I like it. Have fun, merge it to your own, whatever, it's just another toy for ya'll to play with.
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:
SweetBoys 2D is a model that can make slender and cute boy characters.
Fall into the very fantastic and delicate world of beautiful boys!
This model tends to ignore Prompt in Clip skip 1 and draws an abstract illustrations. Clip skip 2 or higher is recommended.
***
If you support me, I will make many AI male models and try to improve them more beautifully.
If you would like to see more and better versions of Bras please consider supporting (tryna recoup gpu cost lol)
https://ko-fi.com/bankaiplease
Trained on countless pictures of Beautiful Asian women. (some nsfw images included)
This is the result of around 3 Months of training and failing and training and merging. After a lot of failures and mild successes I have finally come to a point where it is good enough to release to the public.
Trained mostly on BLIP and WD, so suggest to use similar style prompts.
This model is hosted online and can be run on the following website.
https://randomseed.co/model/18
Example:
Pretty asian in a black swimsuit leaning against a wall
Korean with long black hair wearing a red shirt and a red polka dot blouse with long black hair
a woman with long hair and a black top is posing for a picture with her hand on her chin, 1girl, solo, long_hair, looking_at_viewer, brown_hair, simple_background, black_hair, jewelry, earrings, necklace, lips, black_shirt, ring, realistic
Negative Prompts:
use negative TI embeddings as well as (lowres,worst quality) etc
Feel Free to use this for merges or mixes, just credit.
Feel Free to add me on twitter if you have questions, feedback, requests for improvements or want to ask for commisions or whatever:
A mixture of multiple loras and checkpoint.
Some of them are trained by myself. And I haven't uploaded them yet.
Hi guys!
This is my first SD 1.5 Mix. I trying to give a little bit of comic, anime and semirrealism style.
This is the models contained in this mix:
I recomend this VAE for your experiments.
Greetings! :8)
insertNameHere
Prompt : (best quality), (high_quality), (intricate_details), (hyper detailed:1.3), (ultra-detailed), (ultra highres:1.3), (illustration), (distinct_image), (delicate illustration:1.3)
negative : (worst quality:1.4, low quality:1.4, normal quality, low resolution, lowres:1.4), EasyNegative
EasyNegative | Stable Diffusion TextualInversion | Civitai
Sampling : DPM++ 2M Karras or DPM++ SDE Karras
Sampling steps : 20 ~ 30
Clip skip : 2
Hires. fix : R-ESRGAN 4x+Anime6B or 4x ultrasharp
Upscale By : 2
Hires step : 12 ~ 15
CFG Scale : 7~8
use Xformer
use dddetailer
GitHub - Bing-su/dddetailer: Detection Detailer hijack edition
I trained this model mainly for assisting 3d style graphic effect design, to help designers to realize and express their ideas more easily, so that they can put more energy into some real design things (LOL).
Compared with txt2img, this model works better on img2img.
Here is some example of img2img work：
Use the word "Blender，Lumio，Arnold" to enhance the rendering texture of images
Negative prompt I like to use："drawing,painteing，Mosaic, low quality，Waste quality，photo"
Here is an example of a business project that I used this model to assist with:
3d render ofstyle of graphicreator
V3.5
Made this at like 1am so I forgot what I did.
Notes:
USE THE PROMPT STARTER PLEASE (your gens will look better)
there is a mediafire link to all the TI's in the prompt starter. All credit goes to those who made these TI's.
you need these TI's to get a better image
CFG Scale | 8
Euler A | 44 steps
DPM ++ SDE Karras | 28-44 steps
unipc | Whatever works (I use 132 because of extensions, most people get by with 12 somehow. I can't tho so experiment lol)
If you use a lora and it looks funny. don't use that lora or lower the strength of it
POSITIVE PROMPT STARTER
masterpiece, best quality, perfect, highres, absurdres, hyper detailed, unity 8k wallpaper, perfect lighting, perfect shading, perfect shadows, (solo), 1girl, (colorful), perfect details, volumetric lighting, detailed face, perfect composition, HD, (large eyes, cute face)NEGATIVE PROMPT
(painting by bad-artist-anime:0.9), (painting by bad-artist:0.9), easynegative, pointy ears, NG_DeepNegative_V1_75T, bad_quality, bad-hands-5, bad-image-v2-27000, bad_prompt_version2, cropped, bad mouth, 3 hands, (blurry:1.4), out of frame, verybadimagenegative_v1.1-4800, dick, penis, badv3, neg_grapefruit, tan lines, fat, obese, boring_e621, small eyes, poor detail, (animal ears, hat)Mediafire link to the TI's since it took me like 20 mins to find em online and I know y'all don't wanna do that lol.
https://www.mediafire.com/folder/x89gbgqx6w7tz/all-of-em-in-one-folder
V3.0
Baller update. Figured out how to mix some other things into my model as well as VERY FINELY tweak all the technical thingies :)
Notes/todo:
make it have a style
possibly improve performace
V3.5 I will provide all my TI stuff and some general prompt stuff
V2.1
Tweaked a couple things to get breast sizes under control
V2.0
Same as V1.0 just made the last part 50/50 to fix breasts
V1.0
Mixed AOM3a1 with Babes (0.7-0.30) - AB
mixed AB with baka-diffusion (0.5-0.5) - ABB
mixed baka-diffusion with kawaii2d (0.65-0.35) - BK
mixed kotosmix with BK (0.75-0.25) - KBK
mixed ABB with KBK (0.75-0.25) - ABBKBK
ABBKBK is now called silvermix because yes
LORA-V4 VERSION HERE
A custom build model using SD1.5 as a base. Trained on 30k images ove it's 4 versions, of horror and psychological terror, with a very graining and dark aesthetic. The images were hand sorted to make sure they fit the style and theme I was looking for, and after auto-tagging were then manually tagged with more clear identifiers and themes to help push generation even further.
If you are looking for crisp and clean images, go elsewhere. This is one grainy boi.
Dread V4 adds upon 2 and 3(not posted) with 20k images, custom sorted and tagged, before training. It adds a greater variety of horror based subject matter, and more keywords specifically trained into it's data.
Certain keywords work well invoking feelings and moods in pictures generated with this model. Here are some of the tags used specifically in training data tagging to keep some effects consistent.
Try using any the following in your prompts:
found footage, disposable camera, trailcam: For that grainy look with washed-out colours and faded edges.
liminal, abandoned, hallway, house, forest, swamp: Indoor and outdoor locations that are empty, earie, and even a bit run down.
body horror, visceral, circulatory system, viscera, organs, blood, injury: Various words indicating different types of mutations or styles.
dendrophobia, plants, mushroom, fungus, vines, roots: Vines overtaking objects, creatures made of roots, mushrooms growing out of the skin. Anything plant related.
diagram, illustration: Medical and botany illustration like style, redacted documents, sketches
tome, grimoire, arcane, sigil, magic symbol: Unholy and eerie works or writings to behold
experiment, medical, research, lab, malpractice: Creepy labs, glass tubes, shady science, deranged doctors offices- these tags help with those sorts of settings.
entity, creature, being, monster, animal: Tags used for various oddities and scary things that go bump in the night.
trypophobia, holes: Lots of holes, lots of growths, lots of repeating circular shapes across things.
torture, suffering, pain: The screams of the damned at your command
religious, demonic, ritual, cult: circles of people gathered by candle light, alters and holy (unholy?) figures abound
and many more...
Use DREAD in your prompt, all caps, to reinforce the horror aesthetic. Only the best images were tagged with this, and it can change your generations greatly. This is not a required tag, however! All images for training contained the word 'horror' in their tags.
Lora extraction in the works, will be listed separate from this model, unlike last time.
Dread
Merge of Rev Animated and Orange 50/50 mix with some additional training on video game characters. Will be updated often and pruned as much as possible (goal is 2GB).
WARNING: This model horny, fr fr.
Great for anime/games/nsfw and portraits.
Works great for 512x768 and obviously 512x512.
Start prompts with: (masterpiece:1.1, best quality)
※このモデルにはいくつかのバリエーションがあることを見逃さないでください
Note:Don't miss that there are several variations of this model!
,
Note: If you like it, please give it a STAR MARKS !
,
The caulkinumV2_AR series is a model designed to beautifully portray anime characters and light.
This checkpoint was spawned accidentally through multiple iterations of hierarchical merging.
Therefore, the detailed recipe is already lost.
,
However, there is no doubt that Silicon28 is at the root of it and that it has inherited the influence of ElysiumAnime_V2.
If someone likes it I will be very happy.
,
The caulkinumV2 series has several variations.
"AR" is the first model, which is a modern anime/game style.
"AR2" is the normal evolution of the first model, and it can create better illustrations.
"A4FS" has a dramatic style, which is more in line with recent RPG games.
"ARCT" is an anime-style painting that makes the characters look young.
"ARc" has the same anime style, but is a bit more difficult to handle and a bit outlandish.
"NCA" is a little bit special and unique, but it was created to depict a slime girl.
"NLPS" is a very special model, and if you master it, you will experience a completely different world.
,
A sample comparison is provided below.
It does not always work, but I am generally satisfied with the results.
Note: Underscores at the beginning of checkpoint names are assigned by AUTOMATIC1111 for sorting purposes and have no particular meaning.
,
If the samples are too hard to see and judge, you can dl and try them out.
Then post and show us your great results!
Or you can use ctrl+wheel.
,
,
※ 気に入っていただけたならぜひ★をつけてください！
,
caulkinumV2_ARシリーズは、アニメのキャラクターと光の表現を美しく描くために考案したモデルです。
このチェックポイントは複数回の階層マージを繰り返して偶発的に産まれました。
そのため細かいレシピはもう失われています。
,
しかしSilicon28が根底にあること、ElysiumAnime_V2の影響を受け継いでいることは間違いありません。
誰かが気に入ってくれたなら私はとても幸せでしょう。
,
caulkinumV2シリーズはいくつかのバリエーションがあります。
「AR」が最初のモデルでモダンなアニメやゲームの画風です
「AR2」は正常進化版でより良いイラストを錬成できます
「A4FS」は劇画調で近年のRPGゲームらしさが表現できます
「ARCT」はアニメ風の塗りでキャラクターは幼くなるでしょう
「ARc」も同様にアニメの画風ですが、やや扱いづらく突飛なところがあります
「NCA」は少し特殊で独特の画風ですが、スライム娘を描くために作りました。
「NLPS」はかなり特殊なモデルで、使いこなせば今までとはまったく異なる世界を味わえるでしょう。
,
下記に比較サンプルを掲載しておきます。
必ずしもうまくいくとは限りませんが、私はおおむね満足しています。
注釈 ) チェックポイント名先頭のアンダースコアはAUTOMATIC1111で並べ替えのために付与したもので特に意味はありません
,
サンプルが見えづらく判断に困る場合、あなたはDLして試用することが可能です。
そして投稿してあなたの素晴らしい成果を見せてください！
もしくは ctrl+wheel を使う方法もあります。
,
※ You can download the original image at HuggingFace
(HuggingFaceで元画像をダウンロードできます)
- https://huggingface.co/Shiyaku/caulkinumCheckPoint/blob/main/README.md
,
,
,
,
,
Coffee will increase work efficiency and motivation.
コーヒーは作業効率を高めモチベーションをアップさせるでしょう。
,
Thank you for reading to the end.
I hope one of you will use it.
,
最後まで読んでくれてありがとうございます。
使用してくれるか方が一人でもいることを祈っています。
2.5D Anime model based on AOM3 and Sakimichan style, realistic version is based on ChilloutMix, good at drawing anime characters that have a puffy look it come out with mature body, good with lora, good with NSFW, vae is baked in.
Positive Prompts:
(masterpiece), (best quality), HDR, intricate detail
For girls:
1girl, solo focus, skindentation, perfect face, puffy face, beautiful face, big eyes, puffy eyes, perfect eyes, eyelashes, __skin_color__, __eye_color, __hair__
Or
1girl, solo focus, skindentation, perfect face, puffy face, beautiful face, big eyes, puffy eyes, perfect eyes, eyelashes, __popular_character__
Use body parts for full body shot example: chubby body, narrow waist, big pelvis, wide hips
For Negative Prompts:
EasyNegative, bad-hands-5, (worst quality, low quality:1.4), low contrast, monochrome, text, watermark, thunder, skin imperfection
Make sure you have downloaded the embeddings (EasyNegative, bad-hands-5).
Use nsfw, naked, nude, nipples in the negative if you want clothes on your girls.
Recommanded settings:
Steps: 25
Sampler: DPM++ 2M Karras or UniPC
CFG scale: 7
Size: 512x768 or 512x512
Denoising strength: 0.45
Clip skip: 2
Hires upscale: 1.5
Hires upscaler: 4x-UltraSharp (can be found on huggingface.co)
Future improvement will be made to the model.
Known issues:
Third-party hands: I recommend cropping the photo using the Resize Width in the upscaler to avoid third-party hands.
Faraway eyes are not good.
Change log:
v1: first version.
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
Checkpoint model only for inpainting- takes 20-50 runs to get good results and step by step inpainting gets the best results. Trained on 1800 image set. The initial image can be generated by any model (I used realistic vision 2.0 or URPM). Not tested with anime.
cum panties
A photorealistic merge that does very well at portraits and closeup shots.
Works with both male and female characters but seems to be better at female characters.
Recommended settings:
Steps: 20-25
Sampler: Euler a or DPM++ 2M Karras
CFG scale: 4-7
Size: 512x768 or 512x512
Denoising strength: 0.45-0.5
Hires upscale: 2
Upscaler: 4x-UltraSharp
No VAE required.
Making models takes a lot of time and hard work. If you like this LoRA, please click the like button!
製作模型需要花費很多的時間和精力。如果您喜歡這個LoRA，請幫忙按一個“讚”！
Today, I'm presenting to you "Juggernaut," a side project that emerged during my work. In the past, I was always annoyed when I had to switch models to achieve different styles, so I had been working on another model (StaVis) for some time. From that, this side project emerged, which didn't fit my original vision but was still too good to be thrown away.
Juggernaut is an extremely versatile model that can serve several style directions (paintings, digital, realism, anime, NSFW) and delivers absolutely sharp quality. The sample images are the raw output from Stable Diffusion. The standard VAE was used, so you can experiment with your own VAEs.
It started a couple of weeks ago when i started training my own model (with round about 5k images (50 % Hand tagged/50 % AI Tagged). It was still not the model i searched for so I merged a couple of models into this (Dont ask me which models, i simple didnt write it down) . This Model is more "NSFW" as the Original Project (StaVis) and prob will be updated more into nsfw in the future. But in the End you guys decide ;)
The model works very well with weighted prompts but has some difficulties with shorter prompts.
I hope you have as much fun with "Juggernaut" as I do :) I would be happy to see some of your images in the gallery below.
Edit: Added 2GB Pruned Version
Version Logs:
niji3d V1: general 3D style, released.
niji3d V2: add circular plinth, released (New!).
This model is inspired by https://civitai.com/models/25995/blindbox, but not limited to blindbox style. We train with diverse images and support more general 3D styles.
The tagger word is "<niji-se-3d>". We train on https://civitai.com/models/7371/rev-animated. For convenience, we provide our prompts for easy beginnings and you can try with some simple prompts such as "<niji-se-3d>, 1girl". Both pickle (.ckpt) and safetensors formats are provided.
It would be very appreciated if you could share with your interesting results and star our work. For any problem, feel free to ask in this post, we will reply ASAP. If you like other styles, also let us know.
celsoricardo1
This model has no trigger words, but I highly suggest to use negative prompt, which let you refine your creations, I am going to leave an example.
Example:
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
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.
Download and install Stable Diffusion, I personally suggest the executable version by AUTOMATIC1111.
Download Falkons model in .safetensor format and insert it into the "ckpt" folder.
Start Stable Diffusion.
Select the "Falkons.safetensor" model from the "Stable Diffusion checkpoint" dropdown menu.
This model is compatible with LORAs, Textual Inversion embeddings and Stable Diffusion's extensions, just download your favourites and put them rispectively in stable-diffusion-webui/models/Lora, stable-diffusion-webui/embeddings and stable-diffusion-webui/extensions.
Each add-on has its own trigger words or tags to be added in prompt in order to be triggered. In Stable Diffusion Web UI, you can click on the "show/hide extra networks" button, right below the Generate button, here you can click on Refresh, then on Textual Inversion tab or Lora tab, finally you can click on your wanted add-on to automatically add the trigger words or tags.
I highly suggest EasyNegative and bad-hands-5, download EasyNegative.safetensor and bad-hands-5.pt from https://huggingface.co/datasets/gsdf/EasyNegative and https://huggingface.co/yesyeahvh/bad-hands-5/, copy them in stable-diffusion-webui/embeddings, and use the words EasyNegative, bad-hands-5 as negative prompt to trigger them.
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 and still good quality, I renew you the suggestion to use EasyNegative embedding (read the paragraph above)
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 suggest you to choose multiple of 64. I usually pick 640x960, but I have noticed good results with other resolutions such as 512x512, 512x704, 512x960, 624x1176, 720x1024, 960x1280, while I have experienced poor results with bigger resolutions you can still generate low-resolution images first, then use stable diffusion img2img to increase their resolution through upscaling techniques, e.g. you can generate a 640x960 picture and then upscale it to 1280x1920, increasing both quality and details; I will update this readme with more details later
Batch size and batch count: my GPU can handle a Batch size of 5, which means 5 images generated per run at once, it really depends on your GPU's VRAM, so find the highest number of batch size you can handle. With batch count, Stable Diffusion starts a new run automatically after the previous one, so you get the multiplication result of the two numbers (e.g. batch size: 5, and batch count: 10, means 50 output images)
Example 1
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
Example 2
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, 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, 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, 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
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!
hoi4
柔らかなアニメ塗りを目指したマージモデルです。
サンプルは、Euler a/DPM++ SDE Karras・CFG:8・Step:30・Latent (bicubic)・kl-f8-anime2で作成しました。
（この設定が推奨というわけではありません）
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.
QQ交流群： 704574483
模型仅作为科研兴趣交流使用，如有侵权请联系删除，谢谢！
获得原作者的同意 水一个模型 majic 我把胶片风女孩lora的画风融进来了 哈哈
For those of you who don't have a pc or a weak computer, you can consider using my model via sinkin and mage website using the link below:
Please support me by becoming a patron:
GODofSIMP is an anime model in the style of cutscenes from anime and hentai movies. It's neat, clean and simple, easily produces beautiful photos, and especially with very few errors, even if it's something as difficult as a hand. With a resolution of 704x1024 you can even see nails very clearly.
method: DPM++ 2M Karras
Steps: 30-60
Resolution: (768 or 704)x1024
CFG: 11
negative prompt:
(worst quality:1.5), (low quality:1.5), (normal quality:1.5), 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, painting by bad-artist
Merge of some of the top models
Lora, TI, VAE and other documents for indoor use
This data model has a corresponding LoRa model. If you wish to use it, please download it:
XSarchitectural-7Modern interior | Stable Diffusion LORA | Civitai
XSarchitectural-38InteriorForBedroom | Stable Diffusion LORA | Civitai
If you think it's good, give me a five-star review
If you have any questions, please contact me at xsarchitecturalstudio@gmail.com
Anime painting style model integrated according to my own needs.
A slightly more mature painting style than the general anime model(face/body).
Able to generate a full body image with a background.
Recommend: Clip skip 2 Sampler:DPM++2M Karras Steps:40
CFG scale:8 Vae:Just use one(Anything/Yozora/Pastle)
Highres.fix:SwinIR_4x Hires steps:20 Denoising strength:0.3-0.4(upper body) 0.5-0.6(full body)
(Samples are all t2i without ControlNet,can get better effect when using it.)
根据自己需要，采用分层融合Lora手法融合的二次元模型
画风比起主流的二次元模型稍微成熟（人体比例，中庭长，眼高，脸型）
不用控制网出全身人体也还比较稳，背景也能出（因为素材库关系，如果不描述有时候会出来一些类似方舟、舰娘系二游里那种莫名其妙的元素堆砌，尤其是科幻的东西，还在想如何解决）
之前一直在吃百家饭，第一次分享自融合模型，没什么经验还请大家指导。
【关于1.8版本】
【不一定比1.5版本更好，只能说是另一个选择，很可能是过渡版本】
如诸君所见，选择了一些之前1.5版本类似的prompt，对比下来最大的差别应该是色彩，会更加鲜亮一些；
因为用了face editor所以全身的时候脸会更好一些（用1.5搭配face editor也一样）
顺便说一下关于本系列模型自己摸索出来的一些用法：
主要人物是按正常少女体型出的，想要让角色体型更幼的tag：
short girl,short female,loli,
想要让角色体型更成熟的tag：
tall female,tall girl,mature female,
可以通过调权重来灵活调控
如果觉得皮肤不够白可以white skin,还不够可以再加上(pale skin:0.1)
This is another choice of Salome_Mix.
Brighter but not 100% overcome the 1.5version.
Can use [face editor] to make full body pics.
Can use [short girl,short female,loli,tall female,tall girl,mature female] to adjust the oldage of girls.
Can use [white skin,(pale skin:0.1)] to make a white skin.
[ENG version Below]
这个系列CKPT是由明快制作过程中得来的2D混合3D效果，具体为近景和人物、服装倾向于绘画风格，背景和其他元素倾向于照片风格，略带一些焦外的感觉。总体仍然比较适合各类服饰Lora的展现，但适合添加背景描述。
请注意：
一共有三个版本，分别将不同的流行模型作为主混对象，请依据你的喜好选择版本.
This series CKPT is a 2D mixed with 3D effect obtained during the production of "CrispMix". Specifically, the close-up shots of characters and clothing tend to have a more illustrative style, while the background and other elements tend to have a photographic style, with a slight bokeh effect. Overall, it is still suitable for showcasing various types of fashion Lora, but is more suitable for adding background descriptions.
Please note:
There are three versions in total, each using a different popular model as the main mixing object. Please choose the version according to your preference.
