More Art of OZ

[Stungiskhan]

The little story behind this picture is: two young women were arrested as drug mules and put in a women's prison. There they discovered their love for each other. Unfortunately, lesbian interactions are strictly forbidden in this prison and the punishment is 50 strokes over certain body parts. For each.

That’s pretty rough even for my standards, but for sure serves the purpose well.

I prefer caning from behind, where tip of the cane can deal with the tender parts without risking to brake bones, and a whip from positions like this.

Rubber hose or piece of extension cord works very well in this type of position.

 

[tdmaster]

Thanks a lot for your really kind and useful answers. About the questioned female really I meant this one:https://www.cruxforums.com/xf/attachments/00099-1591284783_oz_ai-png.1357012/

 

[tdmaster]

I don't know exactly which lady you mean. Do you mean Ariana? See attachment.
Either way, I don't get much work done with 3D at the moment, firstly I'm still experimenting with AI, and secondly I don't have the time and peace for other things. But someday... who knows... when the thrill comes....?



I work with DAZ Studio for the 3D work, with Photoshop and Topaz Denoise for post-processing.

The AI images I make with Stable Diffusion 1.6 Automatic 1111.
Here you can find a good tutorial von @fallenmystic for the first steps in the Cruxforum: Link.

Thanks a lot for your really kind and useful answers. About the questioned female really I meant this one:https://www.cruxforums.com/xf/attachments/00099-1591284783_oz_ai-png.1357012/

 

[EADGHE]

https://www.cruxforums.com/xf/attachments/00065-3934019654_oz_ai-png.1357011/

She is my favorite one...very impressive!

 

[EADGHE]

https://www.cruxforums.com/xf/attachments/00031-3144514851_oz_ai-png.1357008/

Would be nice to see this woman while tortured or/and raped...

 

[Zardoz]

https://www.cruxforums.com/xf/attachments/00031-3144514851_oz_ai-png.1357008/
Would be nice to see this woman while tortured or/and raped...

In fact, both are AI characters. It's quite difficult to get the same faces in multiple scenes (there are methods for this, but I haven't dealt with them yet). Besides, it's not yet possible to construct the kind of concrete scenes you envision with AI. AI always has a mind of its own.
I would have to recreate the figures in 3D and then use DAZ Studio to create the scenes, but that's something else entirely.

 

[EADGHE]

In fact, both are AI characters. It's quite difficult to get the same faces in multiple scenes (there are methods for this, but I haven't dealt with them yet). Besides, it's not yet possible to construct the kind of concrete scenes you envision with AI. AI always has a mind of its own.
I would have to recreate the figures in 3D and then use DAZ Studio to create the scenes, but that's something else entirely.

Sounds very elaborate and difficult. Thanks for the answer.

 

[Stungiskhan]

In fact, both are AI characters. It's quite difficult to get the same faces in multiple scenes (there are methods for this, but I haven't dealt with them yet). Besides, it's not yet possible to construct the kind of concrete scenes you envision with AI. AI always has a mind of its own.
I would have to recreate the figures in 3D and then use DAZ Studio to create the scenes, but that's something else entirely.

What I’d really like to get is the AI, which could alter the shape of 3d meshes, create skin textures, expressions, poses and background scenes.

I have made some figures from the scratch in blender, making a mesh I like and then the skin. It’s hard work, but kind of fun alternative for looking at TV. To see the shape of the body form and finally see what she become.

What is the most difficult (impossible) for me is to get proper expressions. I can alter the shape of the face by rigging or making shape keys, but even if she’s beautiful in her base form, expressions become unnatural and distorted. That’s propably because there are so many things which are related and must be right.

AI could learn to control expressions, and that would be the most helpfull. Other big thing would be to create all the small muscular contours and skin details on the body. Similar what some ready made rigged characters already have but better.

Some people are known to be able to draw very realistic expressions but I have many times wondered if they really know what is happening in the face or do they just have very good photographic memory and ability to alter what they have seen.

 

[Zardoz]

Some people are known to be able to draw very realistic expressions but I have many times wondered if they really know what is happening in the face or do they just have very good photographic memory and ability to alter what they have seen.

I used to draw a lot when there were no computers, and I did it very systematically for a while, trying to understand the relationships between muscle play, posture and facial expressions. My skills weren't enough to apply that in the end, but I know that good artists are also concerned with anatomy. Leonardo da Vinci dissected corpses to understand these relationships.

 

[Stungiskhan]

I used to draw a lot when there were no computers, and I did it very systematically for a while, trying to understand the relationships between muscle play, posture and facial expressions. My skills weren't enough to apply that in the end, but I know that good artists are also concerned with anatomy. Leonardo da Vinci dissected corpses to understand these relationships.

I know. Artists are different breed. The problem for me, I'm not an artist, just a man who can try to visualize some fantasies and is quite good with technical stuff and have an eye for a beauty. With my free-hand I can't draw more than a stitch-man.

 

[tdmaster]

I used to draw a lot when there were no computers, and I did it very systematically for a while, trying to understand the relationships between muscle play, posture and facial expressions. My skills weren't enough to apply that in the end, but I know that good artists are also concerned with anatomy. Leonardo da Vinci dissected corpses to understand these relationships.

It is not clear how the muscles of the human head are divided between face, head and neck but it seems now quite well established that those of the face are 36.
Considering that each of them can be completely tense or completely relaxed but then also have dozens of intermediate tensions... I would say that the possible combinations that produce a facial expression are billions and billions.

 

[Loxuru]

View attachment 1357140 I like the last one … it’s the atmospherics/lighting that grabs me. What uncertainty and foreboding she must be feeling.

Having passed the test or facing an uncertain ordeal?

 

[tdmaster]

Slowly - very slowly - I am improving my AI skills. Sometimes I think I've found the method to make a photo out of my drawings with a few adjustments, and then everything is different again in the next picture.... Meanwhile, it also confirms to me that not every picture is good for this. Therefore I have to disappoint @wulf´s hope that I could convert all my pictures..... I am not a bottle genie after all.
In the next post (tomorrow or so) I'll show a few variations that illustrate how different the creation of a usable result sometimes is.

I am trying to learn and I'm

I have somewhat improved my technique of transferring original images to AI images - even though I have only partially succeeded in doing this systematically and in a controlled manner so far. For some images it works better, for others not so well, though I haven't quite figured out yet what makes the difference. At the same time, transferring drawings and sketches works better than 3D images. At least according to my current level of skill. Either way, it's usually quite tedious, but the results are pretty good, I think.
Here you can see a few transfers of my own drawings into AI-generated images. More to come.

View attachment 1332889View attachment 1332890View attachment 1332891

View attachment 1332892View attachment 1332893View attachment 1332894View attachment 1332895

View attachment 1332896View attachment 1332897View attachment 1332898

HI. I'm practicing AI and trying to follow your works and teachings.
For example, I managed to get some results from the vintage photo of the French women in Paris at the end of WW2.
I'm testing with the free Stable Horde and I used img2img, put the original b&w photo as a reference then the prompt you posted.
The prompt certainly helped me a lot to understand...
But now I'm struggling with trying to turn a drawing into a colorful image like you did.
I guess once again you use the img2img function with the drawing as a reference.
But to stay on a really concrete example, could you, for example, tell me the prompts and any other parameters to go from design 013b_OZ to version 013b_OZ_and_AI and then 013b_OZ_and_AI_2...
and/or 100_OZ to 100OZandAI_01 then 100OZandAI_02?
It would be a gigantic help for me.
Am I asking too much?? Thanks anyway in advance.

 

[fallenmystic]

I am trying to learn and I'm

HI. I'm practicing AI and trying to follow your works and teachings.
For example, I managed to get some results from the vintage photo of the French women in Paris at the end of WW2.
I'm testing with the free Stable Horde and I used img2img, put the original b&w photo as a reference then the prompt you posted.
The prompt certainly helped me a lot to understand...
But now I'm struggling with trying to turn a drawing into a colorful image like you did.
I guess once again you use the img2img function with the drawing as a reference.
But to stay on a really concrete example, could you, for example, tell me the prompts and any other parameters to go from design 013b_OZ to version 013b_OZ_and_AI and then 013b_OZ_and_AI_2...
and/or 100_OZ to 100OZandAI_01 then 100OZandAI_02?
It would be a gigantic help for me.
Am I asking too much?? Thanks anyway in advance.

ControlNet (e.g. lineart, depth) is essential in such a workflow. It can be achieved without it by repeating img2img with a very low denoising strength also, but it would be a tedious proces.

Prompting is less important in such a case but adding a few words in the negative prompt to prevent the result from looking like the b&w source image can help.

By the way, if the difference between the source image and the desired output is drastic, I would recommend using txt2img instead of img2img.

 

[tdmaster]

ControlNet (e.g. lineart, depth) is essential in such a workflow. It can be achieved without it by repeating img2img with a very low denoising strength also, but it would be a tedious proces.

Prompting is less important in such a case but adding a few words in the negative prompt to prevent the result from looking like the b&w source image can help.

By the way, if the difference between the source image and the desired output is drastic, I would recommend using txt2img instead of img2img.

... consider that I am a total beginner. I do not know what you meand with "ControlNet(e.g. leneart, depth) . Can you specify more?I'm trying to find some results with those paramenters on Stable Horde:

 

[fallenmystic]

... consider that I am a total beginner. I do not know what you meand with "ControlNet(e.g. leneart, depth) . Can you specify more?I'm trying to find some results with those paramenters on Stable Horde: View attachment 1362891

From the screenshot, it looks like whatever frontend you're using now provides only the bare minimum features of Stable Diffusion, which is insufficient for the task you want to perform.

I haven't used Kobold Horde much but the last time checked, there were multiple frontend options including a modified version of A11. Try searching up what other frontend options we have now and choose one that supports ControlNet at least.

 

[tdmaster]

From the screenshot, it looks like whatever frontend you're using now provides only the bare minimum features of Stable Diffusion, which is insufficient for the task you want to perform.

I haven't used Kobold Horde much but the last time checked, there were multiple frontend options including a modified version of A11. Try searching up what other frontend options we have now and choose one that supports ControlNet at least.

I will. Thanks a lot

 

[frenzyde]

perfect !

 

[tdmaster]

Hello everyone.
What is going on here? Is this thread closed?
I'm going to try and reactivate it a bit.
As you know, I am an absolute disciple of Oz. I started approaching DAZ3D even though I'm still 1000 years behind Zardoz. For the same reason I started to approach AI and I made a new friend who is fallenmystic.
As I promised him, I have been experimenting with different Stable Horde options for a few days now. For the time being, I am limiting myself to using an online interface because I want to better understand what AI is before I buy some new technology as a more modern Mac.
However, I believe that this one is essentially ZARDOZ's "thread" and therefore for AI I believe that in the next few days I will start a discussion on fallenmystic's "thread" to share my experiences but also to get opinions and advice from those who may be more expert than me.
At the moment, however, I think I should quickly return to 3D, because although it is still extremely long and tiring for the beginner that I am, I believe it is much easier to control. I can control the environment independently, every single character in the scene, one by one, with its own character, its own gender, its own physical forms, its own facial expressions.
On the other hand, in hundreds and hundreds of tests with the AI, although the results are often simply amazing, I have not yet managed to understand how to manage and build a scene with my control of the individual characters and the background.
Okay... as mentioned, the debate will be reopened in the fallenmystic thread in the next few days.
For now, I remain in favour of 3D and Zardoz is the best of the masters.
For this reason I hope that the desire to work in 3D will return to him and that he will want to publish again in this area which is specific to him.
For example, I hope that the ordeal of the publicly tortured MILF from the 2jealousy series (page 29) is not over, but just the beginning... or at least it would be nice to see a few more pictures of the same episode that have already been published, but from different angles or points of view.
Good evening everyone.
PS. Below are some of my AI experiments

 

[Zardoz]

Hello everyone.
What is going on here? Is this thread closed?
....
PS. Below are some of my AI experiments

Hi, my thread is not closed, but I don't feel like posting anything new here at the moment. Maybe I'll post some old stuff soon, but I don't know yet.

As you may have noticed in another thread, I'm currently in an in-between state between rendering scenes using AI (which I'm not really getting anywhere with yet) and working with good old 3D rendering, which I'm not really comfortable with anymore. Additionaly, I'm going through a very un-erotic phase right now and I'm prioritizing my time differently.

I think the results of your work with AI are great. You manage very well to implement your fantasies - which I know . I wish you much success in the further work on it.
At the same time, I'm glad if you open your own thread and do not post your further work in mine, there one can distinguish the further development of our different approaches also better. I assume anyway that you will soon get further with it than I can at the moment .