Hi, I thought I'd post
some of my AI generated images and hear what you guys think.
AI generated images have
become controversial.
Many artists are rightly
afraid of competition from the new technology.
And it's frighteningly
easy to copy someone else's style.
In addition, it is
believed that the AI companies' data collection is theft.
Technically, it probably
isn't, because neural networks work in a similar way to a brain.
Artists, as you know, imitate each other. But this technology did not
exist when the current laws were enacted. So it is probably fair to
change the legislation so that consent is needed to use images online
for machine learning.
Can AI make art then? Yes,
if a stolen urinal can be art, then AI generated images can also be
art.
Drawing or painting can be
a very relaxing or even meditative activity. There is nothing that
prevents you from continuing with it, but it will probably be more
difficult to get reasonable payment for it.
But despite all the
objections, AI will certainly become an important tool in the future.
Especially in the media and in the entertainment industry.
I personally want to test
the technology and try to keep up with the development.
Hope no one takes too much
offense from these pictures. Not because they are dirty but because
they are AI generated.
AI images become often
very wrong. They can look like something out of a John Carpenter
movie. My experience so far is
that once you learn how to write good prompts, it's relatively easy
to get pictures that look good. But it's still difficult to get poses
and composition the way you want them. The solution is to make lots
of images and select one or a few as input for more images. Then you
should edit the images in PS. Remove redundant fingers and other
errors. If you have several similar images, you can also take parts
of different images and recompose them in PS.
I use Stable Diffusion on
my own computer. It's free and uncensored and there are tons of
plugins (Checkpoints and Loras) that make it easier to get the images
the way you want them.
In my opinion, the image's
surface, textures and lighting are usually much better than in 3d. At
least if you want them pseudo photorealistic like I want. It is
possible to use AI as an additional rendering for 3D images.
There will also be
programs where you can do the same thing with film. Text to speech
programs with AI are getting so good that I'm considering using them
in my movies. Although now it will be difficult to have time to learn
everything.
First a self-portrait.
Another goat.
Light and texture turned
out well on this one.
Introduction to the
breeding act.
This one got pretty
talented down there.
Chandra.
A bad dragon.
Drows are hot.
This one was unusually
successful.
A drow goddess.
Moonlight.
(moonshine?)
An elf woman.
A gnoll warrior.
Puppy making.
A romantic couple.
A goblin.
A horny minotaur.
Minotaur with his cow.
Minotaur in armor.
A flying monkey.
An Ogre.
Shasira and Rurika.
Here I made two pictures. One with two Shasira and one with two Rurika.
Then I joined two parts in PS.
Rurika, as usual pregnant.
Shasira.
To make Shasira in 3D, I
modified a Poser model representing the virtual
singer Terai Yuki. So
in Ai, I was able to use "Terai Yuki" in the prompt and
got
the images decently similar to 3d Shasira.
First pregnancy.
Twelfth pregnancy.
Twenty-first pregnancy.
It's hard to get the right "time period" on the pictures.
Anachronisms creep in here and there.
Shasira is waiting to be
conceived.
Shasira with a different
style (different model).
Tentacle pit.
Hentai tentacles.
More tentacles.
A werewolf.
A temple relief.
This can be used as a texture in 3d.
And finally a goatman.
First the 3d image as input in "image to image". And
finally the finished image.
Original 3D image made in
C4D.
Image made with the former
as input in Stable diffusion
"image to image".
