Taronites

Verve, Artbreeder and 21st Century Art Education:

Hmmm quite a mouthful, isn’t it?

Art is changing with the advent of multiple A.I.s capable of taking words, images, Ideas and creating something off the beaten path. Looks at variations of a theme seems to be the best way to describe it at the moment.

I said it in Pilou’s thread before it got buried and I say it again. A.I.s are not a substitute for a good art education whether self-taught or schooled. I am fearful that a percentage of our younger generations might not pick up that pencil or crayon and piece of paper and just doodle, enjoying a subconscious freedom of expression instead of taking the easy way out and clicking seeds in an A.I. ad infinitum.

Art A.I.s are here. Walter Cronkite’s “21st Century is rocking along. Lo and Behold. We knew it was coming. How do we save Traditional Art? I mean that it is incredibly addictive sitting there trying seed after seed of iteration, wondering what it will do. How do we channel that productively and get the A.I. and Traditional Art to work together to create uniquely original art?
It all starts with a Doodle.
I have been going into my Verve thread starting with page one when I barely knew what I was doing and uploading to Artbreeder to try variations and also attempting to convey in words to the A.I. what my thoughts (if any) were at the time.

With Collage I can vary the intensity of the A.I. I want to see change, but I want to maintain the spirit of what I have done. The thought here is to see what the A.I. does with the guidelines, find a variation you like, go to an art program like Verve and try and recreate it, thereby creating a New and original variation without the A.I. This keeps us with our feet firmly planted, one foot in Tradition and the other foot dancing around new possibilities.

Possibilities, New Perspectives, More dice to roll with.

I will start this thread with a brief look at abstracts using my “The Devil is in the Details” The original is here in my much-neglected Verve thread http://www.taron.de/forum/viewtopic.php ... &start=210.
The Collage A.I. uses two inputs to create variations with: Image used for background and Text for description of what you want to see. You don’t have to have a background at all for some things but that’s not what I want here.
The text I used for description:
“Art Nouveau, Realistic, The Devil is in the Details, people, Halloween, Trick or Treat Gothic, Manga, highly detailed, vivid color tones, digital painting, concept art, smooth, pencil lines, expert shading, by John William Waterhouse and Alphonse Mucha.”

So in my description I have it set for art style, general setting and a couple of renowned artists. Because Details is a symmetrically based abstract and the A.I. seems to be randomly looking for shapes off the background to form images I used both of those artists to give me the intricate details I wanted present. Each seed in artbreeder leads to different results. I didn’t set it for full A.I. just 25% because I wanted to maintain a look closer to the original image. Full A.I. usually results in complete transformation and while it might be cool is not desired in this case. The description was previously used in several Full A.I. images.

And that’s it. Create in Verve or any decent paint tool, Concept in Artbreeder, Recreate or edit in Verve. Voila.

And because the description is set for it... Once again. Happy Halloween!

Tom

[

With some quasi all words of your prompt! (Inside Wombo)

That rocks Pilou!

Uh, yes, I haven't noticed that thread, yet, sorry about that!
Wow, well, yes, it does look quite beautiful, Tom! It certainly adds a different kind of sophistication. But it makes it all the more challenging to keep your original spirit and to keep pushing you forward, which has been happening beautifully already long before AI came onto the map.

The thing that worries me is the amount of detail the AI produces and the potential discouragement that comes with it, because few artists are ready to crank out that many elements. It's overwhelming. And the other side effect is the growing identifiability of AI produced artwork. That has the eerie other effect that already some people are being "accused" of using AI, while they did not, haha. It's almost funny if wasn't really sad.

Anyway, let it inspire you, sure, but above all, worry not and just enjoy! You still are giving the instructions and even a specific style by feeding it your verve creations. I've done that with Wombo and Vervette's Sandbox already and it works beautifully!

Yes, already Misinformation raises its ugly head when really talented artists get accused of using A.I. Figures and yes, it's sad. That's why if I use an A.I. image in my commercial work I will be totally upfront about it and if it's based off of my own 3d images and Verves I will definitely claim copyright

I agree about the many levels (and layers) that would be required to redo Details. Very complicated. I can handle the hands on my favorite one and the skulls and what look like thumbs? of the next favorite. The problem comes with the art nouveau angles and curves that I love. I have already been trying. I will use them as is.

A more easily done project would be Mouthless Cassandra. Remember her story from my Verve thread? http://www.taron.de/forum/viewtopic.php ... 4&start=10

I can work much more easily on this than Details Damn she has lips! I chose this one because
the hood is easy to fix and to let more hair through
I almost forgot. The Collage A.I. is set at 30% leaving me at 70% lol

@ Taron Yep Vervette inside Wombo is an infinite world of colorful possibilities!

That reminds me, it might actually be kind of clever to use AI as a teaching tool, like letting it suggest "improvements", so to say. It might be very interesting indeed!

Ahahaha, Pilou, yeah, isn't that great! You can truly design your composition in Vervette, colors, placement and even character to some extend. Nice stuff!

I think A.I. suggestions to make art better is a very cool idea It would require different parameters depending on what type of art you want to do.

Pilou that was a very smart way of doing color and placement. That's why some types of abstracts do so well in a good art A.I. I really should dig into Wombo more deeply to compare features.

if you put any image (or here the Vervette green left one above) inside Img to Music
IMG to Music https://huggingface.co/spaces/fffiloni/img-to-music
you will obtain that
https://moiscript.weebly.com/uploads/3/ ... jamhpi.wav

@ Tartan : there is quasi nothing to parametrize inside Wombo!
Only a prompt + a Style (or not) and just an reference image% (or not) if wanted!

It does a great job, but I do prefer more parameters. Sometimes to find an image from a artbreeder seed it helps to be able to increase or decrease A.I. strength. Their new controls work well towards getting consistency. I wish I could open up my private projects which contain spinoffs from Verve and 3d. Still Wombo is a fantastic tool. That series of Halloween images you did was illuminating and caused me to start digging

Taron, I have been thinking about your A.I. idea for education and I think it would be doable with early restrictions due to the A.I. taking baby steps as it learns. To be able to load an image and listen to art pointers would be cool. Made for people who hate Facebook

No project here just a wallpaper i generated for a graphic novel look. Inspired by Hurricane Ida which recently tore up South Florida. This is one of those impossible scenes for me to do without incredible patience unless I could trigger a state of hyperfocus. It would become as daunting as attempts to do those art noveau techniques that are required for details. Still, it makes an interesting wallpaper