AI AI Art Creation Discussion

GuyFreely

Active Member
May 2, 2018
729
2,549
I couldn't seem to find a thread for discussing AI art generation, so I figured I'd start one. Tricks, tips, advice, questions, that sort of thing. I only just recently got into it properly and by that I mean local generation on my own PC. So I am just a newbie myself. I suspect we have some real experts on this site and it would be useful to collaborate. I'll throw a few things out there to start things off. [I've come from the future to report that most of what's below is just rambling, sorry.]

AI Models: Do people tend to find a model they like and stick with it or do you dabble around?

I'm currently in the dabbling phase. I find it interesting to try the same prompts in different models and see how the results vary. I probably have too many redundant models at the moment, but I often find myself switching and swapping to see if something else will achieve what I want. One thing that I've found very useful if I'm having trouble generating with a realistic model is to use one of the hentai focus models with heavy booru tagging then working back from there and converting it in a real life focused model.

LoRAs: This is an aspect I wasn't really aware of only having used some free generators before. For the uninitiated, you can think of them like add-ons to the model that focus on one aspect to reinforce more specifically that concept that the model might only have a vague idea about or no idea about. Common uses are: specific people (celeb fakes) or characters, outfits, sex positions, art styles, etc. Going through LoRAs is a real rabbit hole, again I've probably ended up with too many as I just like to try them out. I find them to be a double edged sword in a sense. Sure, they can inject a concept into the model, but they almost always fundamentally change other aspects. They can change faces, nipples, body types.

Inpainting: Hoo boy, does AI hate hands. Too be fair, I've heard a lot of real artists struggle with hands, so I guess it was to be expected. They are complicated things. The savior here is inpainting. This is when you have the generator redo only a small portion of the image. Very good for fixing faces and hands, which often need fixing. The funny thing is that sometimes I'll use the image that actually has the best hands, because I can inpaint the other aspects more easily.

Conversions: My very basic understanding of how the AI works is that it generates a random noise map and as it starts getting even the hint of a picture, it starts turning nothing into something. It's kind of like looking at clouds and imagining they are shapes of things you recognize. But you don't have to start from nothing. You can give the AI a bit of a nudge in the right direction to inform color, composition, etc. The most specific form of this being image-to-image conversion. If you give it an image that is already of the thing you want to see, it has a much easier time of making it. (This is essentially how inpainting works too.) You can limit the amount of creative freedom the model can take so that it only makes a few tweaks overall.

I find this most useful in trying to make very specific scenes in a realistic model by first making it in a hentai model. The anime/hentai sims strongly leverage booru tagging, which is perfect for machine learning. It's true that some of the real models use similar tagging, but I've just found the anime ones get there a lot quicker. There are models that handle both and most models do have some mix, but you can start running into problems of mixed styles. Your real-life people look too fake or your fake people look too real. Then you are left having to negatively prompt against the aspects you don't want to include and it can be a real chore.

Prompts: Man I wish I knew what the hell things were called. I've found that if you want to make good prompts you should brush up on your terminology. We've all been exposed to a great number of styles and techniques, but most of us have no fucking clue how to put that exact thing into words. If you're lucky, you can find some cheat sheets that give you a whole list of terms and examples of what they look like. The other side of this is the AI not knowing what the fuck you mean sometimes or not having the right context. I find this most frustrating with distance to subject. You have to remember that the model is trained on images in it's default size (512x512 or 1024x1024). If 80% of it's training data had a woman at about three to five feet away, trying to have an image of a woman at thirty feet away is going to be an uphill battle. There are some tricks, but I've very often found that directions around distance and angle are only loosely represented most of the time.

Anyway, feel free to share you own experiences, ask questions, or leave some advice.
 
  • Like
Reactions: fbass and VanMortis

GuyFreely

Active Member
May 2, 2018
729
2,549
What models are people using currently, possibly with some examples?

As I mentioned, I dabble in a lot, but one I've used quite a bit is cyberrealistic-Pony.




rogue2.png paintfit2.png
 

Mipecio

Well-Known Member
Oct 3, 2021
1,556
10,597
Models I'm currently using:
ZavyChromaXL:
Jib Mix Realistic XL:
Pony ForReal:

I have also played with Flux for some time now.

Some of the loras I use:
Detail Tweaker XL:
DetailedEyes_XL:
Midjourney mimic:
Kodak Film Grain:
Chiaroscuro (Contrasted) Lighting Style XL:

for Flux and SDXL:
Image Upgrader / Detail Maximizer / Contrast Fix:

and for Pony:
Pony Detail Tweaker:
Incase Style [PonyXL]:
Styles for Pony Diffusion V6 XL:
All Disney Princess:
 
Last edited:

Sharinel

Active Member
Dec 23, 2018
613
2,583
As I predominantly use Pony models this is a godsend for me.



I also find that when running Pony models, the more realistic one is the more 'locked down' (for want of a better term) it is in terms of poses and backgrounds. To get around this I have started generating with base Pony and then refining with the realistic Pony at 50% - see here

1724539839934.png

This gets me the more varied Pony poses/backgrounds, but also the Realistic characters
 

Kretua

Member
Oct 21, 2020
125
1,051
Hi.
I posted a portrait of a woman the other day and was told she was too young. As a test image, I re-set the prompt with a slight modification.
The only difference between the prompts of the two pictures: big breasts.
Interesting "thinking" by AI, though.
 
Last edited:

Mipecio

Well-Known Member
Oct 3, 2021
1,556
10,597
Hi.
I posted a portrait of a woman the other day and was told she was too young. As a test image, I re-set the prompt with a slight modification.
The only difference between the prompts of the two pictures: big breasts.
Interesting "thinking" by AI, though.
I usually put "child, kid, teen" in negative prompts and I would get similar results like you have on the right side pic.
 
Last edited:

Kretua

Member
Oct 21, 2020
125
1,051
Oh, yes. I always neglect the negative prompt.
Here, for example, there was nothing, although the solution is obvious.
What bothered me was the prompt that began: The picture is a portrait of an 18-year-old girl with bright pink hair in a messy bun.
And I thought the 18-year-old girl prompt was enough.
Thanks for the tip.


Hehehe, I wanted to generate an image with a negative prompt, but Forge simply won't let me use it with Flux. I don't know if this is a bug or not. There's been a huge uprade going on with Forge for a few weeks now.
 
Last edited:
  • Like
Reactions: Mipecio

Sharinel

Active Member
Dec 23, 2018
613
2,583
Oh, yes. I always neglect the negative prompt.
Here, for example, there was nothing, although the solution is obvious.
What bothered me was the prompt that began: The picture is a portrait of an 18-year-old girl with bright pink hair in a messy bun.
And I thought the 18-year-old girl prompt was enough.
Thanks for the tip.


Hehehe, I wanted to generate an image with a negative prompt, but Forge simply won't let me use it with Flux. I don't know if this is a bug or not. There's been a huge uprade going on with Forge for a few weeks now.
Hiya, no that is not a bug. Flux works on a cfg of 1 (and then its own cfg scale) and when you have a cfg of 1 it stops you from using the negative prompt field. You could set it to 1.5 and it would appear again, but from what I understand it would slow it down by a lot.
 
  • Like
Reactions: Kretua

Mipecio

Well-Known Member
Oct 3, 2021
1,556
10,597
Oh, yes. I always neglect the negative prompt.
Here, for example, there was nothing, although the solution is obvious.
What bothered me was the prompt that began: The picture is a portrait of an 18-year-old girl with bright pink hair in a messy bun.
And I thought the 18-year-old girl prompt was enough.
Thanks for the tip.


Hehehe, I wanted to generate an image with a negative prompt, but Forge simply won't let me use it with Flux. I don't know if this is a bug or not. There's been a huge uprade going on with Forge for a few weeks now.
Yeah, I noticed that too. It’s a shame that you can't use negative prompts, but as I mentioned before, I use prompts like "21yo" and so on.
 
  • Like
Reactions: Kretua

GuyFreely

Active Member
May 2, 2018
729
2,549
Weird, I don't think I was getting notified about the posts here. Yeah, I find pony models work well, but they do often have a sort of overly polished look. I'm torn between more realistic and more perfect when it comes to models. I do often find myself using one model for the layout and styling then using img2img in another model to balance some things out. Another thing I see with Pony is that it seems to be heavily influenced by anime when it comes to expressions. You get some real cartoony faces if you use 'surprised' for example.

Sharinel Do you have some examples from that setting you posted?

I'm really impressed with what Flux can do so far, but it is not on board with a lot of NSFW stuff yet. If we can get a cross-breed of Flux's prompt adherence with Pony's tag library (and decent body parts) that would be golden.
 

Sharinel

Active Member
Dec 23, 2018
613
2,583
Weird, I don't think I was getting notified about the posts here. Yeah, I find pony models work well, but they do often have a sort of overly polished look. I'm torn between more realistic and more perfect when it comes to models. I do often find myself using one model for the layout and styling then using img2img in another model to balance some things out. Another thing I see with Pony is that it seems to be heavily influenced by anime when it comes to expressions. You get some real cartoony faces if you use 'surprised' for example.

Sharinel Do you have some examples from that setting you posted?

I'm really impressed with what Flux can do so far, but it is not on board with a lot of NSFW stuff yet. If we can get a cross-breed of Flux's prompt adherence with Pony's tag library (and decent body parts) that would be golden.
I've started doing inpainting with Flux now over the Pony models. Here's one from today



Have a look at my deviant gallery (shilling, moi?), all of the last week+ has been the same process with flux inpainting. I think they come out looking quite realistic, in fact deviantart suggests tags when you upload anything and these always have 'photorealistic' as a suggestion.

Mikayla Couch Surfing 07.jpg

One other tip for flux inpainting - mask the naughty bits from your original pony/sdxl image then use the 'inpaint not masked' option. As long as you use a low denoise it will often give you excellent flux-style NSFW images
 

GuyFreely

Active Member
May 2, 2018
729
2,549
Yeah, I really think Flux delivers on believable scenes (within reason). It would be nice if it wasn't a two step process.
 

Sharinel

Active Member
Dec 23, 2018
613
2,583
Yeah, I really think Flux delivers on believable scenes (within reason). It would be nice if it wasn't a two step process.
Need to wait on NSFW checkpoints coming out, which looks like a long way off with the hurdles of flux
 

angel__666

Newbie
Nov 24, 2023
54
464
Which AI do you recommend for a similar result to these cartoonized pic below?

I'm a pretty decent python developer, but I suck doing artwork. I have been working in this personal project of mine for a while. It have been satisfying so far, but my plot is ready and I don't wanna just adapt or change my plot just because I can't find similar real porn scene for my MC. I'm planning to learn these AI tools aiming to rework all artwork in my project, and have more flexibility in my game development. Thanks in advance
 
  • Like
Reactions: fbass

GuyFreely

Active Member
May 2, 2018
729
2,549
I would say your best bet is to find keywords for the style you want and use a model that is trained for drawn art instead of realism. Most are geared toward more of an anime look, but you can push it the right way with prompts. If you can find a LoRA that matches the style you want, that's your best bet. Below are some examples that are not quite the same, but in a similar direction made from a realistic original image and converted. A lot of making AI art is trial and error, seeing what works and what doesn't. Good luck!

Keywords used: oil painting, painterly, brush strokes, highly detailed.
1730523840355.png

Keywords:hand drawn, pencil art, full color, detailed.
1730524086552.png
 

Gundamu0079

Well-Known Member
Sep 19, 2018
1,702
2,438
Which AI do you recommend for a similar result to these cartoonized pic below?

I'm a pretty decent python developer, but I suck doing artwork. I have been working in this personal project of mine for a while. It have been satisfying so far, but my plot is ready and I don't wanna just adapt or change my plot just because I can't find similar real porn scene for my MC. I'm planning to learn these AI tools aiming to rework all artwork in my project, and have more flexibility in my game development. Thanks in advance
probably flux

but getting their superior pro/dev version is hard, i haven't seen it.

you cant try civitai and browse for models.
 

DW Games

New Member
Dec 13, 2024
4
0
Every time I dive into AI, I feel like I’m explaining a flip phone to my grandma—completely lost and ancient.
To be honest I just learned about civitai.com from this thread; before that, I was fumbling with tensor.art and similar platforms. So, thanks for starting this thread, and I’ll take the opportunity to ask—what would you recommend for 2D NSFW animation?
 

Sharinel

Active Member
Dec 23, 2018
613
2,583
Every time I dive into AI, I feel like I’m explaining a flip phone to my grandma—completely lost and ancient.
To be honest I just learned about civitai.com from this thread; before that, I was fumbling with tensor.art and similar platforms. So, thanks for starting this thread, and I’ll take the opportunity to ask—what would you recommend for 2D NSFW animation?
Either Pony or Illustrious. I think Illustrious is the better at NSFW concepts but Pony is the more mature (hah!) ecosystem. Have a look at civitai models for both.
Flux is the far far superior model but is absolutely shite for NSFW, which is why I have disregarded that one.
 
  • Like
Reactions: fbass

GlavGlav

New Member
May 20, 2021
11
5
Please help find some working free AI site where generate 18+ pics? Tried find by search there, but no luck.
Tested Tensor.Art is very difficult to write a request, it generates from the ceiling.
Civitai, too, but the mnyu has ended their points for 2 or 3 photos.