[AI Generative Art]

Water Guardian

 

 

[AI Generative Art]

Krea enhanced, to varying degrees of success.

 

 

 

 

 

[AI Generative Art]

1 hour ago, Player2 said:

The first few images look like they lean more toward anime stuff, while the last three look more comic booky (or comic book cartoonish).  I've never heard of Fooocus.  I will have to see what else you show us, but this definitely looks like it has potential.

‘It was part of the Free AI thread: 

 

[AI Generative Art]

1 hour ago, Player2 said:

Second set of smaller breasts? Interesting that it gave her boots but then put toes on the left one.

[AI Generative Art]

1 hour ago, Player2 said:

Comparison time:  OpenArt vs Krea

 

Pretty Cool has the Spines powerset, and I chose the crystalline form to look like ice.

Instead I got a mix of mutated feathers and possibly a tarantula limb...

Fingers all over the place, clothing is some kind of post-apocalyptic fashion trend I think...

Just weird all around.

 

Krea gives me sunglasses, costume looks spot on, and spines do kind of look like crystalline ice shards.

 

Bonus:  I like what it did with the clock face on her belt buckle.

In that top pic she looks Liza Weil who played Paris Gellar on Gilmore Girls.

[AI Generative Art]

8 hours ago, Player2 said:

 

Well that's a terrific upgrade.

 

It did the same thing with the open hand that it did to one of mine, Gato Blanco, turning it around to be fingers in a pseudo-fist. I forgot to post the ones I did last night after the server crashed. I’ll pop over to my PC in a bit.

[AI Generative Art]

6 hours ago, Player2 said:

Krea finally hit a wall with one of my designs...

 

 

First one I didn't change settings.  The second one I specifically added "mirrored sunglasses"; didn't help and changed her hairstyle.  Third one I replaced sunglasses with goggles. 

Had to happen eventually I suppose.

[AI Generative Art]

3 hours ago, Player2 said:

 

In the auto-generated prompt, Krea identified my Little Trickster as "A green alien with a red hat is jumping in the air on a space ship."  I figured goblin would have been the obvious term, but what do I know.  Still, it worked fantastically.

 

This has to be the single best transformation of all. Synthesizing the necessary kludges we have to do to approximate the look we’re going for into a cohesive whole that just looks right is amazing. This with your “flayed” Stalker really shows the program’s abilities.

 

Are you changing the parameters on Krea’s sliders?

[AI Generative Art]

The revamped Dread Vilda.

 

Bing AI version and Krea enhanced version. Krea made her chain mail green for some reason, even after I told it to use silver.

 

[AI Generative Art]

Enhancing villains.

 

Krea could not understand Dr. Vahzilok, even after I added that he was wearing a surgical mask and a head mirror.

 

Clamor was cleaned up, but when I added "maniacal grin" it made a creepy version that's pretty close.

 

[AI Generative Art]

Lady Vampyre in Steel Canyon

 

Krea Enhanced photo which OH MY GOD WHAT HAPPENED. Apparently Player2 used up all the talent because this is very very wrong. Apparently it was channeling that Spanish woman who ruined the fresco of Jesus a few years back.

 

 

[AI Generative Art]

Also did a few more Dread Vilda shots. (I think I’m going to remake her to be less Red Sonja and more her own thing. Maybe white or very pale blonde hair to start.)

 

Prompt: “Epic fantasy, an ancient tomb lit by torches, action scene, dynamic pose,  a warrior woman like Red Sonja wielding a sword and fighting off mummies”

 

DeviantArt just completely ignored the “fighting mummies” part, but the photo realistic renderings are freakishly good. Once again Bing gave me two Caucasian and two African versions of the character, and I really like the Nubian Warrior Princess version of Vilda. I’m definitely going to make a character based on that, especially since there were real-world all-female warriors in West Africa: https://en.m.wikipedia.org/wiki/Dahomey_Amazons Real life Dora Milaje from Black Panther!

 

Bing AI:

 

DA AI:

[AI Generative Art]

I used Bing AI and DeviantArt’s AI to render a mouse character using the exact same prompt and the results were sub-optimal, but DA’s was especially bad. Unlike Bing where you can still render scenes after your credits run out, DA cuts you off. Bing is 15 credits a day, DA is 10 credits a week.

 

The shield really confounded both of them. Bing kept sticking Coke bottles in the scene, so I leaned into that after a while.

 

Prompt: “art station style, watercolor, a tiny mouse warrior wearing an acorn helmet, wielding a sewing needle and holding a metal soda pop bottle top as a shield”

 

Altered prompt: “watercolor, a tiny mouse warrior wearing an acorn helmet, holding a needle as well as holding a metal Coca-Cola bottle-top lid as a shield”

 

After endless iterations I finally got something close to what I was after from Bing:

 

DeviantArt never even got near the neighborhood:

 

The best versions. I also later gave up on the shield idea and had the mouse riding a bird, first a house wren and the a cardinal, but it never really worked.

 

[AI Generative Art]

I was away from my PC yesterday and was goofing around with both Krea and Bing on my iPad. I wanted to try Krea on some of my pulp-inspired heroes, but all I had were quick photos I took of the monitor for use as costume references when making related characters, so the characters are distorted. Gave them a go anyway and they kind of turned out okay. Later today I’ll use actual in-game screenies.

 

Soul Sister

Masked Lark

Ace Club

[AI Generative Art]

5 hours ago, Player2 said:

That is the stuff of nightmares.  The Dark Armor soul noir effects in black added those black rags effect and the starburst aura radiating out of it is amazing

Looks super cool though.

[AI Generative Art]

8 hours ago, Krimson said:

New Lora running locally on vanilla SDXL. v3 is much better than v2.

Pushing it back towards manga really helps it. The exaggerated facial features stay out of the uncanny valley that way.

[AI Generative Art]

46 minutes ago, Player2 said:

I love this design...  and it enhanced beautifully.

 

This one really looks great.

[AI Generative Art]

4 minutes ago, Krimson said:

I adjusted the previous prompt to get something a bit more accurate. The first image was enhanced on Krea.

Curious how it sometimes adds the glowing line to the boots. Just interpreting the reflection, I guess.

[AI Generative Art]

Dread Vilda as interpreted by Bing AI.

 

I used the same simple prompt each time, just switching out the location and enemy.

 

”Art station style, epic fantasy, [location], a warrior woman similar to Red Sonja is wielding a sword and fighting [enemy]”

 

I did about a dozen variations, but here’s the gist. There’s a hint of Horizon: Forbidden Dawn in some, but overall I liked the results. It’ll be nice when we can get persistent characters in various scenes.

 

 

[AI Generative Art]

My character Dread Vilda with the standard setting in Krea.

 

The auto-generated description was half accurate and half not, but it came out fine.

 

“In this image, there are three panels featuring a Sim dressed as Red Sonja. The Sim is standing in a grassy field and is wearing a bikini, a corset, and knee-high boots. The bikini and corset are silver with metallic details, and the Sim is also wearing arm bracer armor on their arms. The Sim has short, orange hair and is holding a sword in each hand.‘

 

 

[AI Generative Art]

1 hour ago, Player2 said:

One magical mad scientist, please.  Here you go:

 

 

That hair is incredible. And I’m stunned the gun came out right. You’d think of all things that would really confound it.

[AI Generative Art]

1 hour ago, Player2 said:

find it weird that the thigh high boots on the left seem to have been interpreted as having some separate layer thing going on below the knee on the left where the original boots are smooth all the way up, whereas the middle and right actually have a separating line there but its been interpreted as little more than a wrinkle

It’s also odd how Krea insists on adding belly buttons. It seems to be interpreting tights as both clothes and skin simultaneously. However, a year from now this thing will be making these talk, so weirdness on the way is to be expected.

[AI Generative Art]

8 hours ago, Jawbreaker said:

What is everyone find to be the best AI platform for altering existing images?

I have to say, Player2 is getting some great results with Krea.

[AI Generative Art]

31 minutes ago, Player2 said:

On the other hand, the hero costume and civilian party dress came out perfect.

She ended up looking a lot like Aubrey Plaza.

[AI Generative Art]

44 minutes ago, Player2 said:

Wow, that came out great.