AI [AI Art] - Show Us Your AI Skill *NO TEENS*

[Sharinel]

While that approach may have some effect, it's not necessarily as precise as it could be since there's been no positive reinforcement of the model or textual embedding vectors.

A nice trick I was told was to use a male celeb and mark it as female, seems to be quite consistent

[Chris Pratt | henry cavill] as a 25 year old sexy gorgeous woman, (((masterpiece))), full shot body photo, opened mouth, [[smirk]], eyebrow up, (((transparent outfit))), graceful, slim, athletic, large breasts, sexy, (nsfw), scenary, detailed and intricate, absurdres, detailed background, complex background, Intricate, High Detail, photorealistic, hyperdetailed



The only problem is sometimes you get Chris Pratt with boobs

 

[Jimwalrus]

A nice trick I was told was to use a male celeb and mark it as female, seems to be quite consistent

[Chris Pratt | henry cavill] as a 25 year old sexy gorgeous woman, (((masterpiece))), full shot body photo, opened mouth, [[smirk]], eyebrow up, (((transparent outfit))), graceful, slim, athletic, large breasts, sexy, (nsfw), scenary, detailed and intricate, absurdres, detailed background, complex background, Intricate, High Detail, photorealistic, hyperdetailed

View attachment 2630789 View attachment 2630790

The only problem is sometimes you get Chris Pratt with boobs

I have noticed that SD seems to be more consistent and accurate when generating male celebs than female.
Not sure why - perhaps society's push for the female ideal (and airbrushing of published photos) results in celebrity women looking more like one another than men?

 

[Splendiddxxx]

While that approach may have some effect, it's not necessarily as precise as it could be since there's been no positive reinforcement of the model or textual embedding vectors. A likely better approach would be to render out a set of images that consistently depict the characteristics you would like to encode for and then reinforce those characteristics through training. Koiboi posted a pretty good overview video on YouTube concerning this:

You must be registered to see the links

Edit: Attached a few more renders, fully aware that there are some technical flaws in some of them, still working on prompt engineering.

I would be glad if you say which checkpoints and which prompts you use at the renders you made. I want similar renders but i think im doing something wrong.

 

[PandaRepublic]

Sommer Time


​

 

[Sgravo05]

 

[Deleted member 395648]

 

[Sharinel]

This is how the Spice Girls should have been marketed

 

[Deleted member 395648]

testing on playing with existing renders from existing games
so much remastering potentials for so many games if given enough time

 

[Birulei]

Rias Gremory:

 

[Dagg0th]

 

[NachoCheese]

I would be glad if you say which checkpoints and which prompts you use at the renders you made. I want similar renders but i think im doing something wrong.

Look at my previous post in this thread, specifically +:, -:, and PS (postive, negative, and additional settings, such as checkpoint, etc). Also, assuming the images posted were direct from automatic1111 (as mine are), you can download them and load them into the PNG info tab to read out the parameter data and copy into the various modules.

And now, for something completely different: I have included some of my NSFW test results in this set of attachments, but nothing overly explicit. I'm still not comfortable with the inability to control for apparent age, probably the best approach will be to develop a custom checkpoint to fulfill the requirements of a given game/character set. Working on this in my very little off-time.

Also, consider that my prompts work for this checkpoint, and that you'll probably need to experiment and tune your prompts for each checkpoint you make use of.

 

[felldude]

Im still trying to learn Cuda and Cudnn, but for now Im just using the AI scaling with blender and GimP

A side thought you know that it took 5 years for chrome to get super sampling on browser videos despite ISR being open source.

Also you can still download the files used to make Disney's Moana...one model is 100GB and the whole island is over 10TB so good luck making use of it.

Below is an example of using AI plus 3D to create a new perceptive on a static photo(30-60 degrees with a 12k image)

 

[Microtom]

 

[felldude]

Also celebrity fakes

 

[Birulei]

Another brazilian which some brazilians will recognize(sorry for these flying stuffs in this photo)

 

[felldude]

Animation from a single image via AI Super Scaling (20k) and 3D tools. (Original was 4k but to big for server)

 

[bogumil]

Are ai generations of real models allowed in this gallery?
If not I delete it. Just inform me.

My take on Hannah Palmer. Since she is not eager to loose her clothes in real life I managed to play some time with Stable Diffussion ( online)

I have more of it. If u like it I upload some in the future

*Edit: Okey, nobody kicked me out . I am uloading the rest od Stable Diffusion Online experiment.
I ve just installed SD on my pc. Unfortuantely I have only 2060 gpu, so I cannot train my model. I tru to use colab. hope it works.

But seriously I hope somebody with Supreme GPU will make some nice models of her in future. Girl surely deserves some attention of AI creators.

 

[Deleted member 395648]

 

[daddyCzapo]

My whole secondary drive got corrupted so i'm starting fresh Some underboob for the boys

 

[Lorthard]