• Hi Guest!

    We are extremely excited to announce the release of our first Beta1.1 and the first release of our Public AddonKit!
    To participate in the Beta, a subscription to the Entertainer or Creator Tier is required. For access to the Public AddonKit you must be a Creator tier member. Once subscribed, download instructions can be found here.

    Click here for information and guides regarding the VaM2 beta. Join our Discord server for more announcements and community discussion about VaM2.
  • Hi Guest!

    VaM2 Resource Categories have now been added to the Hub! For information on posting VaM2 resources and details about VaM2 related changes to our Community Forums, please see our official announcement here.
Create sexy art using VaM + Stable Diffusion

Guides Create sexy art using VaM + Stable Diffusion

JaxZoa

Invaluable member
Featured Contributor
Joined
May 6, 2021
Messages
296
Reactions
4,201
JaxZoa submitted a new resource:

Create hentai with VaM + Stable Diffusion - Workflow i use to create hentai artwork

View attachment 181802 View attachment 181803
This is a small guide on how i create hentai artwork using VaM and Stable diffusion, with little to no drawing skills (Thx again @Barcoder the the infos about stable diffusion!)

First, install Stable Diffusion, lots of guides out there, i recommend using a one-click installer if you don't know what you are doing ( you need one with a UI, and a way to change the model used!)

- Use...

Read more about this resource...
 
Hello !
Thanks for the tutorial, but my native language is not English, I do not even understand how to install ...
I found this: https://github.com/CompVis/stable-diffusion but at no time is there an exe in it...
how does it install? do i have to install python? :unsure:

Hey, i take it from your profile picture that you speak french ? '^^ If you want direct help, DM me on Discord!
 
Hey, I installed the StableDiffusion using https://github.com/cmdr2/stable-diffusion-ui which worked perfectly. I follow your guide and use the custom models, but I can't get the same results as you do. Could you please give me some literatal prompts you used to generate the pictures you used?
 
Hey, I installed the StableDiffusion using https://github.com/cmdr2/stable-diffusion-ui which worked perfectly. I follow your guide and use the custom models, but I can't get the same results as you do. Could you please give me some literatal prompts you used to generate the pictures you used?

If you look at prompt crafting guides, forums etc, you'll see that it's really deep, especially with good models.
Prompt crafting is the actual "skill" and personal touch here :) i can give you ideas, but i'm hesitant in sharing my full prompts!

The way i see it is the following: The words you use in the prompts are image tags to narrow the inspiration of the AI, for exemple, writing "A girl sitting on a sofa" will work, but "1Girl, Sitting, Sofa" is better! Don't write it like you would for a human, see it more as a google search.

And there is a sweet spot for prompts. If you are not specific enough, the generations will be mostly crap, and same if you specify too much! My prompts are usually between 20 and 50 words.

Generate ton of images! I generated around 200 images per artwork (just for the base character) then maybe 200 more for the details, background etc. (background usually needs way less generations)

Oh and one last thing, as i said in the guide i modified each one of these artworks in photoshop afterward! The original generation is almost never good enough!
 
I get it and thanks for your elaborate answer, but are there any art styles you mention? Or drawn “by artist X”? (I mean, I get that you don’t want to share kinks, I am more interested in the technical stuff of how to get the western hentai/anime style).

200 images is a lot, good to know that you used that to choose from.

This is an awesome tool by the way. Thanks for pointingme to it. Had no idea that they actually shared the neural model to run it locally.

Do you know of any NSFW / porn like pretrained neural models that you can use like the Any… v3?
 
Last edited:
I get it and thanks for your elaborate answer, but are there any art styles you mention? Or drawn “by artist X”?

200 images is a lot, good to know that you used that to choose from.

This is an awesome tool by the way. Thanks for pointingme to it. Had no idea that they actually shared the neural model to run it locally.

Do you know of any NSFW / porn like pretrained neural models that you can use like the Any… v3?

Tried every NSFW Anime style models that i could find. The only ones that are close-ish to Anything v3 are the last version of HentaiDiffusion, and the last version of WaifuDiffusion. And they are not as good, at least to generate female characters (almost impossible to generate naked male/futa characters in Anything v3, so for this maybe HentaiDIffusion is better ?).

The only thing i think is better than Anything v3 is the Unstable diffusion discord bot, but generating NSFW art in a public channel feels kinda weird to me (but their model is pretty impressive).
 
I get it and thanks for your elaborate answer, but are there any art styles you mention? Or drawn “by artist X”? (I mean, I get that you don’t want to share kinks, I am more interested in the technical stuff of how to get the western hentai/anime style).

200 images is a lot, good to know that you used that to choose from.

This is an awesome tool by the way. Thanks for pointingme to it. Had no idea that they actually shared the neural model to run it locally.

Do you know of any NSFW / porn like pretrained neural models that you can use like the Any… v3?

For the style i usually use a brand reference, but i don't refer to it directly.You need to use the word that could narrow the style the most. For exemple, if you want the style of the Black Lilith studio, Typing "Style of black lilith" is a bad idea, as black lilith is a reference to mainly unrelated stuff (succubus stuff, and everything linked to the color black). But if you use Taimanin, it will mostly (only ?) be linked to their characters! This way you can capture their style with better precision. Using the authors real name is the same problem. For exemple if it's a name given to another well known person, or to a fictional character.

Another way to capture a style in your prompt is describing it's main features, for exemple if you like BunnyWalker hentais, they have a way to draw skin that is pretty unique. Desribing that is a better way than typing Bunnywalker (you'll have bunny ears everywhere, and bunnies on leashs XD)

After that i use a description of the style, like "HD", or "Line art" (if you want bigger outlines), or "Cel Shading" if you want cartoon like shadows etc.
 
Last edited:
For the style i usually use a brand reference, but i don't refer to it directly.You need to use the word that could narrow the style the most. For exemple, if you want the style of the Black Lilith studio, Typing "Style of black lilith" is a bad idea, as black lilith is a reference to mainly unrelated stuff (succubus stuff, and everything linked to the color black). But if you use Taimanin, it will mostly (only ?) be linked to their characters! This way you can capture their style with better precision. Using the authors real name is the same problem. For exemple if it's a name given to another well known person, or to a fictional character.

Another way to capture a style in your prompt is describing it's main features, for exemple if you like BunnyWalker hentais, they have a way to draw skin that is pretty unique. Desribing that is a better way than typing Bunnywalker (you'll have bunny ears everywhere, and bunnies on leashs XD)

After that i use a description of the style, like "HD", or "Line art" (if you want bigger outlines), or "Cel Shading" if you want cartoon like shadows etc.
This is one of my best so far, no editing.

This was my prompt:

"1 girl, blonde hair, big breasts, western realistic art style, full length, skimpy blue bikini, thick drawing lines, black background, standing, black high heels, smiling, big eyes, slender, photo realistic, pony tail, large head, concept art."
1_girl__blonde_hair__big_breasts__western_realistic_art_style__full_length__skimpy_blue_bikin...jpeg
 
This is one of my best so far, no editing.

This was my prompt:

"1 girl, blonde hair, big breasts, western realistic art style, full length, skimpy blue bikini, thick drawing lines, black background, standing, black high heels, smiling, big eyes, slender, photo realistic, pony tail, large head, concept art."

Nice!

Can you try this one "1Girl, standing, blond ponytail, big breasts, fullbody, skimpy blue bikini, high heels, Comics style, hand drawn, black background, Cel shading, Thick line"

The 1Girl, 1Men style prompt is a special tag, you have to write it like that, without space iirc.

Not tested, just how i'd write it (maybe it's crap sorry xD)
 
For the style i usually use a brand reference, but i don't refer to it directly.You need to use the word that could narrow the style the most. For exemple, if you want the style of the Black Lilith studio, Typing "Style of black lilith" is a bad idea, as black lilith is a reference to mainly unrelated stuff (succubus stuff, and everything linked to the color black). But if you use Taimanin, it will mostly (only ?) be linked to their characters! This way you can capture their style with better precision. Using the authors real name is the same problem. For exemple if it's a name given to another well known person, or to a fictional character.

Another way to capture a style in your prompt is describing it's main features, for exemple if you like BunnyWalker hentais, they have a way to draw skin that is pretty unique. Desribing that is a better way than typing Bunnywalker (you'll have bunny ears everywhere, and bunnies on leashs XD)

After that i use a description of the style, like "HD", or "Line art" (if you want bigger outlines), or "Cel Shading" if you want cartoon like shadows etc.
Good tips, thanks. The short prompt also works much better.
Nice!

Can you try this one "1Girl, standing, blond ponytail, big breasts, fullbody, skimpy blue bikini, high heels, Comics style, hand drawn, black background, Cel shading, Thick line"

The 1Girl, 1Men style prompt is a special tag, you have to write it like that, without space iirc.

Not tested, just how i'd write it (maybe it's crap sorry xD)
Thanks for sharing your insight. It gives me some directions of how to write a prompt. This particular prompt didn't do too well, this is the best of 10.

Another question: is there some way to make like a series of images. Or to put it differently: how do you ensure continuity? Let's say I would like to make a series of pictures of the same generated girl in different poses. Is there a trick or way to go about that?
1Girl__standing__blond_ponytail__big_breasts__fullbody__skimpy_blue_bikini__high_heels__Comic...jpeg
 
Good tips, thanks. The short prompt also works much better.

Thanks for sharing your insight. It gives me some directions of how to write a prompt. This particular prompt didn't do too well, this is the best of 10.

Another question: is there some way to make like a series of images. Or to put it differently: how do you ensure continuity? Let's say I would like to make a series of pictures of the same generated girl in different poses. Is there a trick or way to go about that?

Unfortunately not easily. You can with a lot of photoshop work and a lot of generations, by basically redrawing the important features of the character for each image.
 
If you go for sfw realism, i think the best is midjourney v4 :)
 
00041.png00090.png00091.png00099.png
The hassan blend model is cool, but the real magic is in merging and training your own models imo ! This is my HyperBabes_v1 Model, made from f222, hassan,anything v3 and a few others!
 
I’m following this thread with great interest. My GPU is currently cloning a voice, but after that I’ll dig into this a bit more. It’s amazing.
 
I downloaded the Hassanblend model but I have a hard time creating the stuff looking as good as some of these pictures in here. Are these pictures like the best of 200 tries or so?
 
I downloaded the Hassanblend model but I have a hard time creating the stuff looking as good as some of these pictures in here. Are these pictures like the best of 200 tries or so?

often you have an arm or a second body that appears, despite the negative prompts ... for my part I do not insist too much, 4 - 8 tries, if it goes in all directions, I change the prompt. In general I noticed that the more you are precise, the more it looks like nothing lol

I also recommend adding an artist name like:
1670509174675.png


My negative prompt :
ugly, deformed, deformity, disfigured, malformed, ugliness, blurry, disfigured, mutation, mutated, extra limb, bad anatomy, long body, cropped head, cropped face, anatomical nonsense, malformed hand, long neck, missing limb, floating limb, disconnected limb, lowres, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, dehydrated, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, bad anatomy, bad proportions, extra limbs, cloned face, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, loli, lolicon
 
I'll delay the v2 guide a little, still learning so much i'd have to update it every day! But now i see how bad the v1 guide is ahah x)
 
Does anyone have a guide on how to install and then activate the models? This is all new to me and I do have SD running and working, but not sure where to put the models and then how to get them running. Any help would be awesome.
 
Just put the models in the "models" directory. And then refresh your browser. You can then select them in the GUI of Stable Diffusion!
 
Just put the models in the "models" directory. And then refresh your browser. You can then select them in the GUI of Stable Diffusion!
Thank you. Is it just the one file or do you need another? I saw someone mention a VAE file.
 
Back
Top Bottom