Check the version description below (bottom right) for more info and add a ❤️ to receive future updates.
Do you like what I do? Consider supporting me on Patreon 🅿️ to get exclusive tips and tutorials, or feel free to buy me a coffee ☕
Available on the following websites with GPU acceleration:
Live demo available on HuggingFace (CPU is slow but free).
New Negative Embedding for this: Bad Dream.
NOTES
Version 6 adds more lora support and more style in general. It should also be better at generating directly at 1024 height (but be careful with it).
Version 5 is the best at photorealism and has noise offset.
Version 4 is much better with anime (can do them with no LoRA) and booru tags. IT might be harder to control if you're used to caption style, so you might still want to use version 3.31.
V4 is also better with eyes at lower resolutions. Overall is like a "fix" of V3 and shouldn't be too much different.
Results of version 3.32 "clip fix" will vary from the examples (produced on 3.31, which I personally prefer).
I get no money from any generative service, but you can buy me a coffee.
You should use 3.32 for mixing, so the clip error doesn't spread.
Inpainting models are only for inpaint and outpaint, not txt2img or mixing.
After a lot of tests I'm finally releasing my mix. This started as a model to make good portraits that do not look like cg or photos with heavy filters, but more like actual paintings. The result is a model capable of doing portraits like I wanted, but also great backgrounds and anime-style characters. Below you can find some suggestions, including LoRA networks to make anime style images.
I hope you'll enjoy it as much as I do.
Official HF repository: https://huggingface.co/Lykon/DreamShaper
Suggested settings:
- I had CLIP skip 2 on some pics (all of them for version 4)
- I had ENSD: 31337 for basically all of them
- All of them had highres.fix or img2img at higher resolution.
- I don't use restore faces, as it washes out the painting effect
- Version 4 requires no LoRA for anime style. For version 3 I suggest to use one of these LoRA networks at 0.35 weight:
-- https://civitai.com/models/4219 (the girls with glasses or if it says wanostyle
)
-- https://huggingface.co/closertodeath/dpepmkmp/blob/main/last.pt (if it says mksk style
)
-- https://civitai.com/models/4982/anime-screencap-style-lora (not used for any example but works great)
NOTE: if you find that the prompts below look "familiar" it's because I've taken them from other reviews and models here, to basically compare my model to other examples. Credits to the original authors. Thanks for the benchmark.
model of shivji in style of anime.
bnha
ske
Nothing to See Here.
Move Along.
This model is versatile, applicable from 2D to 3D models. However, being universal also implies that it's not specifically optimized for any particular type of model. Please keep this in mind.
The model uses 'photorealistic' and 'anime key visual' as keywords in the prompt. By weighting 'low quality' as a negative prompt, you can adjust the details of 'photorealistic' and the style of 'anime key visual'. Example images according to these prompts are provided below.
Feel free to experiment with a variety of other options beyond 'photorealistic' and 'anime key visual'. Remember that exploring various other options might necessitate adjustments to corresponding elements such as the sampling method and upscaler.
Here I share one of my favorite self-merged model. It's merged from a complex triple-merging method. Mainly focusing on Sci-Fantasy boys (but still works with girls and general arts) with the best intricate every single detail.
You may get DDetailer extension to get the best result
Recommended Prompt
Absurdres, Intricate Details, Masterpiece, Best Quality, High Resolution, 8k
Recommended Settings
Sampling method: DPM++ 2M Karras, DPM++ SDE Karras
Clip skip: 2~6
Hires.fix upscaler: 4x-UltraSharp / Latent
CFG Scale: 7~9
VAE : vae-ft-mse-840000-ema-pruned
Merge Material
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
https://www.patreon.com/GorillaMonsoon
anemone is one of my first trained model, at first I wasn't satisfied with the results I got, then some time ago I watched some old anime and movies (especially Kill Bill movie LOL), and then an idea occurred to me to retrain, by adding some images in the existing dataset.
Dataset has been preprocessed using Aspect Ratio Bucketing Tool so that it can be converted to latents and trained at non-square resolutions.
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.
v1.0-fp16: added fp16 (no EMA) model
LoRA/LoCon/LyCORIS user who are looking for a compatible checkpoint
Creative prompters
People who needs Manhwa model
Almost compatible with all sampling methods available, but Euler a and DPM++ SDE Karras are my recommendation
Use at least 20 sampling steps, more than 40 is not giving any improvement
CFG: 4.5-7
Less complicated, short negative prompts are better
Preffered VAE
If using Hires:
Hires step: 10-20 steps
R-ESRGAN 4x+ Anime6b, if you're prefer Latent upscaler, use Latent (nearest-exact)
After download this model, try to share with us about your findings.
Please give star reviews with your generated picture if you like it.
I'm also accept commisioning for LoRA/Checkpoint
Kanade
Negative prompt: (worst quality, low quality:1.4), (bad_prompt_version2:0.8), EasyNegative, badhandv4, text, name, letters, watermark
Steps: 20,
Sampler: DPM++ 2S a Karras,
CFG scale: 7,
Denoising strength: 0.5,
Hires upscale: 2,
Hires steps: 20,
Hires upscaler: R-ESRGAN 4x+ Anime6B
Clip skip: 2,
VAE: kl-f8-anime2.vae.pt
man
Would appreciate images, feedback, requests, reviews, and of course comments in the comments. Please check out my other models if you like this one.
this is a Blend of EveryDayArt [EDA] (a checkpoint I am still working on), CounterfeitV3, MenaiHentaiV3 and Paizuri FocusV1 (can't remember the exact percentages - but if I find them I will put them) [so it's half original and half merge, not that that matters, just thought it was cool]
this is an illustration/anime model, can do nsfw and sfw, and gives good paizuri images if that's your thing.
for the paizuri stuff visit my model and/or my LoRas - Paizuri Focus - Paizuri-FocusV1 | Stable Diffusion Checkpoint | Civitai
for everything else, well if your downloading this you know the drill, get creating.
found bugs: pov likes to result in paizuri unless fought or your more specific (Ie: put pov in both negative and positive and/or specify more of the scene like doggystyle, from_behind, pov, etc), and even then I found that genning in batches usually only the first or first two images would be paizuri, but could have just been lucky
note - some of the images were used with orangemix Vae, so that's why some are less saturated, extra note - think this is wrong, was going to use those images but then didn't, also vergil model is burnt, not a good showcase of the quality. extra extra not - yes first note was right, some did use orangemix
authors note - I am terrible with names, would like any that you guys come up with, preferable blood or red related
A bit of detail with a cartoony feel,
and it keeps getting better! so close to a full release!
////////////////////////////////////////////////////////////////////////////////////
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_yolov8n
Sampler : DPM++ 2M Karras / DPM++ SDE Karras / UniPC
CFG : 8 Steps : 25
Prompts
(best quality, masterpiece)
Neg : (worst quality, low quality)
////////////////////////////////////////////////////////////////////////////////////
Why is my image different from yours?
Use prompts that match the danbooru tag
Tags like Realistic, HDR, etc. have the side effect of forcing the model's characteristics to change
Use Neg only when necessary
Refrain from excessive negative embedding
Use Adetailer for eyes
Use similar settings
////////////////////////////////////////////////////////////////////////////////////
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
idol
マスターチーフ
Merge various checkpoints
2.5D anime style, very good at generating rich details in human portraits, also quite realistic.
It can be used to generate realistic photos together with chilloutmix (details will be richer).
2.5D动漫风格,在生成人像的丰富细节上非常好,也比较接近照片写实。
可以用来配合chilloutmix来生成写实照片(细节会更丰富)。
This model is not only capable of generating NSFW portraits, it is good to generate any content through prompt.I've been using this model for a month and it feels good so I'm sharing it now.
Thank you for using, This is my first model. Welcome to provide evaluations and useful criticisms or feedback.
该模型不仅可以生成人像,通过提示生成任何内容都不错。
感谢使用,欢迎反馈问题和建议。
This is a model created with the desire to produce an image that is close to reality but still carries a hint of fantasy. At the same time, it is intended to have the ability to blend well with many different LoRa and texture elements, as much as possible. Therefore, I would greatly appreciate your feedback in order to improve this model further in the future. Thank you very much.
Use LoRA by additional Network for more eff efficiency
masterpiece, best quality, best illustration, HDR, beautiful details, intricate details, detailed scenery, dramatic light, best shadow, 1girl, 18 year old, medium breast, wearing ultra detailed dress, detailed skin texture, tyndall effect
(worst quality, low quality, normal quality:1.4), lowres, wrong anatomy, bad anatomy, deformed, (extra arms), (missing arms:1.3), extra hands, extra fingers, (missing fingers), (red eyes, text, watermark, signature, username, artist name), error, blurry, jpeg artifacts, cropped, disfigured, ugly
Step 30 | Hires Step 10 | Denoising strength 0.35 | Scale 8 | Sampling method Eluer A | Upscaler 4x_UltraSharp | CLIP skip 2
trigger word : urnm
please, no commercial use
urnm
If you want to support my efforts donations are gladly accepted for time put in. https://ko-fi.com/asmrgaming
Mega Model 1 with an additional 80 merges or so on top of the 500 of v1. Basically a big mix of civitai models, good bad and ugly all averaged out in an attempt to get a very complex neural model going. this is a 16 fp file and it is still 5.8gb or so. the full version at 32fp is 12.8gb. So for this one i will upload the condensed version to guage community interest
Trans community charity support in australia, a great group championing for trans rights : Transcend Australia | Support for Trans, Gender Diverse and Non-Binary Children and their Families
Merged my favorite models with a few tweaks and here it is!
Vae: kl-f8-anime2
Sampling method: DPM++2M Karras
Sampling steps: 28
Upscaler: R-ESRGAN 4x+ Anime6B
Hires steps: 13
Denoising strength: 0.4-0.6
CFG Scale: 8-10
my ko-fi(if u wanna buy me a coffee :)): https://ko-fi.com/yotta53891
Thank you!
im still learning how to make and merge models, please feel free to give me any tips!
RuviStyle is a stylized anime art style. 🎨
Great at doing full body shots with scenery. (Hires Fix is needed)
Great at creating landscapes.
Great at making upper body portraits.
✅ Recommended Parameters ✅
Sampler: Euler a: 20-30 steps.
CFG Scale: 5-7
Hires.fix: Latent, 20 steps, 0.37 to 0.45 denoising OR Latent(Nearest to Exact), 20 steps, 0.5 denoising
Clip Skip: 2
Negatives: (low quality, worst quality:1.4), (bad anatomy), (inaccurate limb:1.2), bad composition, inaccurate eyes, extra digit, fewer digits, (extra arms:1.2), (EasyNegative)
*Model was brought back, there was an issue that lead to the disappearance of the model
All of my stuff will be free to use! Support me on Kofi~ ☕
Hide
With your support, I can continue to develop.
You can also participate in the testing of the model under test.
皆さんの応援があれば、私は開発を続けることができます。
また、テスト中のモデルのテストに参加することもできます。
[Other series include]
BreakDomain (illustration output models )
https://civitai.com/models/50520/breakdomain
BreakDomainAnime (illustration output models that are more animated)
https://civitai.com/models/72675/breakdroidanime
BreakRealize LoRA (LoRA that adjusts to a realistic representation)
https://civitai.com/models/61810/breakrealize-lora
BD_Marumi LoRA (LoRA to create an animated expression)
https://civitai.com/models/70877/bdmarumi-lora
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。
《AI-游戏图标研究所》"AI-Game Icon Research Institute"
This is a model of a large test game ICON, which belongs to the beta version. Welcome to join the QQ group: 489141941 to discuss and give comments, which is convenient for model developmentThis version is mass-produced by the Icon Academy model, and everyone is welcome to use and comment for better optimization in the future.
After the release of version 3.1, test it in the QQ group
1. On the text generation map, it is easier to have the desired effect
2. The image quality of the version is better, 3.1 period effect ratio 2.0 quality improvement
3.Because the model is a personal suggestion and free of charge, there is a large amount of demand for the picture.
4. Hoping landlords to enter the group, help the game map research institute, provide picture resources.
5.You can ask in the QQ group for specific writing methods for prompting words.
game icon
描述编辑中……
trigger word : kimstyle
Kim Hong Do Style
kimstyle
inspiration from https://civitai.com/models/25494/brabeautiful-realistic-asians-v5
Thanks for Bankai dataset from:
https://huggingface.co/datasets/BanKaiPls/AsianDataset1/tree/main
例图效果推荐latent放大。例图数据我先发再补,中间断了三次上传。例图数据又被吞了。。。
DPM++ SDE Karras, 50 to 60 steps, clip skip 1 is recommended.
Suggested negative prompt: negative_hand-neg, bad structures, bad clothes, watermark, bad ratio, bad toe, bad finger, bad foot, clone, big breasts, giant, long neck, extra finger, mutated finger, duplicate, bad hands, wrong limb, extra limb, low res, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, penis, extra hand, bad toes, bad fingers, ugly, (incorrect hands), (incorrect fingers)
Valkyrie_Mix is meant to be a beautiful and colorful/bright anime and evnironmental mix. This is meant for SFW but NSFW is possible!
Lets just say I needed a bit more awesomeness in my life, well here we are, something to satisfy my need for greatness, any maybe yours too!
-
This is a custom creation! (with the help of others!)
-
Secret VAE!?! : O
-
It is recomended to use (512x768), (Euler A), and a (CFG Scale of 7)
(768x768) images are still good and possible :
Example Generation
Positive prompt: (extremely_detailed), The Final Stand
Negative prompt: (worst quality, low quality:1.4), easynegative, bad anatomy, bad hands, cropped, missing fingers, missing toes, deformed, disfigured, distorted face, mutation, ugly, poorly drawn hands, out of focus, monochrome, symbol, text, logo, door frame, window frame, mirror frame, elf
Steps: 30, Sampler: Euler a, CFG scale: 7, Seed: -1, Size: 768x768,
-
Thank you the creators and mods of Yodayo, AI generation website, as without them i would not be here today making good images and checkpoint merges.
-
Special thanks to you, for downloading! : D
Seraph_Mix is meant to be a beautiful and colorful/bright hentai mix.
Something with the Civitai website broke with the original model, sorry to those who have it, but hey, now you have exclusive models! This is a Re-Upload, and has been fixed/improved.
-
-
It is recomended to use (512x768), (Euler A), and a (CFG Scale of 7)
(768x768) images are still good and possible :
Example Generation
adding (extremely realistic) or (extremely detailed) can change the style!
Positive prompt: (masterpiece, detailed), detailed anime visuals Advanced stunning epic magical picture full of details, majestic scene, beyond beautiful ethereal majestic illumination, simultaneous contrast, stunning artwork, beautiful atmosphere, layered topography, (perfect face), (2 wings, angelic, purple seraphim, nude), (nice hands, perfect hands, good hands), <lora:GoodHands-beta2:1>, purple lotus flowers
Negative prompt: (worst quality, low quality:1.4), easynegative, bad anatomy, bad hands, cropped, missing fingers, missing toes, deformed, disfigured, distorted face, mutation, ugly, bad hands, out of focus, monochrome, symbol, text, logo, mirror frame
Steps: 30, Sampler: Euler a, CFG scale: 7, Seed: 3771725356, Size: 768x768,
Use the prompt above, its pretty cool!
-
LORA's from "ddp12" are pretty cool.
I also recommend using this LORA : "BetterHands" - By Envy
-
Thank you to the creators of both Yesmix, and AOM3, without them i would not have been able to create such a beautiful checkpoint merge.
Thank you the creators and mods of Yodayo, AI generation website, as without them i would not be here today making good images and checkpoint merges.
-
Special thanks to you, for downloading! : D
To begin with, English is not my mother tongue, so I apologize in advance if there is any mistake.
I think this checkpoint really likes doing portraits.
The images I posted have been inpainted at least one time. (0.15 strengh).
I think 512x512 and 512x768 are better, but please, experiment.
Please, don't use restore face.
Since I did't do a lot of testing, I'm looking forward to see what you guys will create.
Please, show me what my checkpoint is capable of !
Hi, this is a checkpoint merge of several models. Notably it's an older verion, but still great.
List:
-Chilloutmix
-Lofi
-Realistic v13
-SeekArtMega
-Grapefruit
-HM? hentai mode? not sure???
It's a 70/30 mix so 70% realism and 30% anime.
InToThe2KGothClub
A stylized version of my realism mix. This mix is more anime-style.
More sample pictures and noise-offset variant model coming soon!
After tons of experimenting, I have updated the already-fantastic SB1.6.3 to meet the ever-increasing quality standards of models. I would love for folks to test this model vs. the famous A-Zovya Ultra model. The biggest difference between this model and that one is that this model is more dreambooth-compatible than Zovya's, a fantastic model otherwise marred by inflexibility.
The model does work with NSFW prompts, since it is mixed with the old SenBlend, which was in part mixed with the UberPorn model. Of course, there is a tradeoff between accuracy and quality that must be made for a more generalized model like this.
The SenBlend 1.8 checkpoint is a blend of the following models:
Edge of Realism: 24.2%
Cyber: 24.2%
SenBlend 1.6.3: 18.2%
Analog Madness: 10.37%
Consistent Factor: 10.37%
Zovya Pruned: 5.94%
Beautiful Asians: 3.78%
TheAlly: 2.97%
COMING SOON: SenBlend 1.8 Dark Version, a version of this model finetuned with LoRA's such as LowRA for those who prefer darker pictures in general instead of the more exposed ones you typically get.
Happy diffusing!
mixed a few furry models with anime and hentai models to get the results I wanted
anthro
body fur
This model is not mine, is a merge between Chillout Mix fp16 and Realistic Vision v2, so....enjoy it.
Hugging Face: jr-ai-dreamer/Chillout_Vision_v1 · Hugging Face
This model belongs to the SD2.1 series. It's an enhanced model that underwent additional training based on the hakoD model.
A distinct feature of this model is its propensity to generate images of boys.
Please include "sepia" as a negative prompt.
In the hakoMay series, there are ABCD series of girl models, but currently only D is published on civiitai.
The use of Negative Embedding from "Mayng" is recommended.
hakoMayA/B/C and Mayng Link
https://huggingface.co/852wa/hakoMay/tree/main
*Example images' prompts in InvokeAI format, not A1111 format. If you don't use InvokeAI, don't copy/paste my prompts without first adjusting the syntax.
Animated version is less heavily anime and thus more general-purpose. Anime version is more focused towards anime, and has some extra specialization in chaos and scenery, and much less tendency for uncanny valley.
New - Anime-V2 (still SD 1.5 because I can't run 2.1 768x models lol): New version with a bit of a style change. Reduction in the giant/titan issue that appeared in some prompts, and the model is less overbaked and cleaner at default settings with a tradeoff of being a little more needy for good negative prompting. Has a slight leaning towards generating women, but still won't fight you if you want dudes because that's annoying as hell and I wouldn't allow that. Also, anime is more entrenched and doesn't need to be explicitly prompted anymore.
USAGE TIPS, BEST PRACTICES, AND SETTINGS (Anime-V2 version):
The recommended negative embeddings are negative_hand, and then either a choice of bad-artist (the base version), <neg-anime>, or <neg-sketch-2>. negative_hand
doesn't interfere with the model and has an okay success rate. bad-artist has less effect on model imagination than <neg-anime>
or <neg-sketch-2>
, and provides brighter and more colorful results, though may not increase detail as much. <neg-anime>
or <neg-sketch-2>
are the opposite. More impact on imagination, yet more detailed and less change in composition. <neg-anime>
seems to lean towards darker images in Anime-V2. Luna's KHFB and AuroraNegative are also safe to use, though I don't use them personally.
Other negative embeddings (e.g. bad-artist-anime, EasyNegative, bad_prompt_version2) are generally not recommended, as they impact art style and model expressiveness and have little, if any, actual benefit.
When not using any negative embeddings, negative_hand
aside, using (low quality, worst quality, lowres)
with a 1.3 or 1.4 weight does the job for most prompts if you get any subpar definition or color.
Does not respond very well to LoRAs trained mainly 2.5d, 3d, or photorealistic-style images, such as dogu_cat's it's Skyler White Yo guy LoRA.
Upscaling via img2img / High-Res Optimization in InvokeAI is ideal, especially at a strength like 0.47-0.55.
Example images were generated without CLIP skip. CLIP skip 2 has interesting effect, though does reduce stylization.
This model uses the Waifu Diffusion 1.4 VAE with the pixel layer modified for -6% contrast. img2img at low strength may reduce quality unless you swap the VAE.
USAGE TIPS, BEST PRACTICES, AND SETTINGS (Animated version):
negative_hand
and bad-artist
are the only negative embeddings I can comfortably recommend using for this model, and not even that is a requirement. <neg-anime>
or <neg-sketch-2>
are also great, but tend towards a semi-realistic style in this model, so be wary of the uncanny valley if you use them.
To liven up color and lighting (when not using a negative embedding), I suggest putting desaturated
and pixelated
in your negative prompt with 2 or 3 units of down-weighting (that means [[[desaturated]]], [[[pixelated]]]
in A1111 UI, or desaturated---, pixelated---
in InvokeAI).
To lean output more towards an anime style rather than semi-realistic, include anime
in your prompt. Put realistic
in your negative prompt to further lean towards 2D, or by the front of your positive prompt to lean more 3D.
Example images were generated in InvokeAI, so you'll have to use your UI's weighting syntax (which means the + and - in my prompts likely won't do anything for you unless you're using InvokeAI).
This model will be great for:
People who want an anime aesthetic that is different from other SD 1.5 models
People who are more distant, casual enjoyers of anime, who might find this model more welcoming than one that's heavily leaned to modern anime art only
People who like more classic anime styles, ala Voltron and Ghost in the Shell
People who like generating in 2D and semi-real styles and liked 526Mix, who will probably enjoy the better lighting and touch of wackiness in this version
The "Animated" model is a straightforward mix of 526Mix-V1.3.5, and Nerfgun3's Macaron Mix, the latter having had the noise offset added at 0.70 multiplier. This is done with a weighted sum at a 0.3 multiplier with Macaron Mix. I always like Nerfgun3's art and embeddings, so I felt I could trust that model to be fairly in line with my own creative desires and expectations.
As always, I suggest going to the source models for the full experience, and Nerfgun3's Macaron Mix and Newartmodel4 aren't exceptions here.
Example images were generated in Invoke AI with the model converted to Diffusers format, hires fix on (0.45 strength, works like img2img), and the sampler DDIM (unless listed otherwise). This means unless you use Invoke AI, you likely won't be able to recreate my images exactly. Just learn from the prompts and modify the weighting in prompts as needed for the UI you use (if you use the A1111 UI, any (plus sign)+ is equal to one set of parentheses).
2.5D++ is the final version of SAV_MIX, improved for anime with more 2.5D and more faces. You can always upgrade to version 1.1 if you run into issues with the request. Even more detail and less body exposure compared to 1.1 and 2.5+.
I collected a dozen LORA models, stylization for anime (I'll be honest with the hentai), one realistic model - this mix strongly influenced and at the end I added an emphasis on landscapes. Landscapes, even with minimal elaboration, turned out surprisingly well. And yes, don't be surprised, it will suddenly give you monsters at v1.1.
No modifications to kohya-LoRA-dreambooth model test.
Aesthetics-focused anime finetunes. Waifu Diffusion 1.5-based, inherits that license.
Outdated and experimental variants can be found here.
Images generated with the Replicant2 version may be overly bright; if you find this to be the case, consider using this extension. This is not a problem with more recent versions.
many many checkpoints merged , including 2 of my own custom dreambooth creations..
clip skip 2 works well , i also am using 4x_NMKD-PatchySharp_240K upscaler , ( see second set of pictures i just made
The alpha model has been replaced! Because someone complain that it makes weird body or weird hands. And it's not too special to naming 'beta' , so I replaced the alpha model. Please understand I announced in description.
It's a merged model that focuses on the person rather than the background
I tried to simplify the line as much as possible
I simplified it and maintained a clear outline
If you don't add 1girl, you can only see the background, so add 1girl plz
I'm using Counterfeit-V2.5.vae, but it's okay to use the VAE you used before)
Recommend vae : Anything-V3.0.vae
Negative prompt: (worst quality, low quality:1.4), (bad_prompt_version2:0.8), EasyNegative, badhandv4, text, name, letters, watermark
Steps: 20,
Sampler: DPM++ 2S a Karras,
CFG scale: 7,
Denoising strength: 0.5,
Hires upscale: 2,
Hires steps: 20,
Hires upscaler: R-ESRGAN 4x+ Anime6B
Clip skip: 2
mecha
UPDATED: v2.2
Official Doc v2.2:
Updated the: 22/5/2023
Google Docs // Model Information
Update: Gigantic changes
New Universal Prompt Style
90% less Negative Embeddings needed
No more chitty generation, generation are now by default good (Even without Positive Prompt)
Less complicated Extensions
Prompt for camera angle, is working
Better Preset
Bug correction
Free commercial use