Artwork Messing around with AI, A Art thread.

[charclone]

All of the art AIs I found so far want money to use them which sucks.

Starry.AI gives 5 credits free per day, and Automatic1111's web UI for Stable Diffusion is a free one if you can install and run it.

 

[Agent23]

All of the art AIs I found so far want money to use them which sucks.

There is

GitHub - Stability-AI/stablediffusion: High-Resolution Image Synthesis with Latent Diffusion Models

High-Resolution Image Synthesis with Latent Diffusion Models - Stability-AI/stablediffusion

github.com

I think this thing is free, even has some waifu/hentai datasets IIRC.

Stable Diffusion AI: Absolute beginner's guide & online demo - Stable Diffusion Art

Want to learn Stable Diffusion? This beginner's guide is for newbies with zero experience with Stable Diffusion or other AI image generators.

stable-diffusion-art.com

 

[TheRejectionist]

All of the art AIs I found so far want money to use them which sucks.

I am currently trying to find a way to use disposable emails to avoid paying for Midjourney.

 

[Culsu]

I am currently trying to find a way to use disposable emails to avoid paying for Midjourney.

It's probably less of a hassle to simply pay $30 for one month and spam the system in relax mode.

 

[Bear Ribs]

I am currently trying to find a way to use disposable emails to avoid paying for Midjourney.

Seriously, just throw down Automatic1111 and run it at home, you both get better quality (the Checkpoints are made by normal people and not hobbled due to corporate policy) and no stress about how many images you get to make a day.

 

[*THASF*]

I wanted to see if I could get Midjourney to do my Champions Online main from way back in the day, the alien wolf furry psychic supersoldier, Major Ulrica.





Oh. Holy shit. That turned out way better than I anticipated.

 

[Carrot of Truth]

I wanted to see if I could get Midjourney to do my Champions Online main from way back in the day, the alien wolf furry psychic supersoldier, Major Ulrica.

View attachment 1699

View attachment 1700

Oh. Holy shit. That turned out way better than I anticipated.

 

[*THASF*]

I don't, like, identify as a furry or anything monumentally stupid like that. Coming up with my Champions main was like, "Wait, they have animal heads and tails and shit? Lol, okay, whatever. Ironic wolf time."

If anything, this ended up generating more annoyance for me, because other players would sometimes approach me and be like "Are you into RP?" and I had to tell 'em, y'know. Fuck the fuck off.

 

[Carrot of Truth]

I don't, like, identify as a furry or anything monumentally stupid like that. Coming up with my Champions main was like, "Wait, they have animal heads and tails and shit? Lol, okay, whatever. Ironic wolf time."

If anything, this ended up generating more annoyance for me, because other players would sometimes approach me and be like "Are you into RP?" and I had to tell 'em, y'know. Fuck the fuck off.

I don't man that shit is still pretty sus.

 

[Agent23]

I wanted to see if I could get Midjourney to do my Champions Online main from way back in the day, the alien wolf furry psychic supersoldier, Major Ulrica.

View attachment 1699

View attachment 1700

Oh. Holy shit. That turned out way better than I anticipated.

Suffer not the furry to live!

 

[Allanon]

So thanks to @Baughn I got my hands on Disco Diffusion. And decided to make an art thread just like them.

Here's what I've generated so far.

That first one is stunningly beautiful.

Sadly I cannot display any of my artwork here.

 

[mspence]

Dark Moon Rising nightcafe?

 

[mspence]

Bruce Willis color portrait nightcafe:

 

[mspence]

River Landscape nightcafe from original work:

 

[*THASF*]

MJ Niji 5 does surprisingly good Sonics. So, anyway, I asked it to do Sonic as "the badass protagonist of a swords and sorcery graphic novel written by Alan Moore and illustrated by Brian Bolland", and, well, it did exactly that.

Spoiler: The REAL Knight of the Wind

Damn. That's pretty much exactly what that would look like.

I also had Niji do Twilight Sparkle as an absolutely terrifying vampire, for some reason, complete with a transformation sequence from cutesy mode to seriously-going-to-tear-out-and-eat-your-heart mode.

Spoiler: Vampire Twi

Bonus Widescreen:

I also had MJ V5.1 do "vintage Bonneville land speed record cars", and sure enough, it came through.

Spoiler: Old Bonneville Cars that Never Were

 

[Jormungandr]

Huh, those look pretty good. I could easily see the Vampire Twilight ones being the cover for some edgelord FiM Fiction site story, lol.

Hmmm... two experiments to test the bar?

"Star Trek Borg Twilight Sparkle" (or Princess Cadence) and "Doctor Who Cyberman".

AI collectively seem to have trouble with the Borg and Cybermen in prompts for some reason, so it's a good measure test, I think.

 

[*THASF*]

Huh, those look pretty good. I could easily see the Vampire Twilight ones being the cover for some edgelord FiM Fiction site story, lol.

Hmmm... two experiments to test the bar?

"Star Trek Borg Twilight Sparkle" (or Princess Cadence) and "Doctor Who Cyberman".

AI collectively seem to have trouble with the Borg and Cybermen in prompts for some reason, so it's a good measure test, I think.

Niji just spat out a generic Twilight Sparkle Cyborg:

MJ V5.1 puts out similar disappointing results to what it used to:

V5.1 still does off-brand Cybermen:

The accuracy of a generated image depends entirely on how many examples there are in the training data. The pics of Imperial Stormtroopers in the training data vastly outnumber Cybermen, which is why I can very easily have it do a stormtrooper eating pizza with his helmet on, and it comes out pretty much spot-on:

Another thing is that you really have to be specific with the prompts, because the models are trained on images from many different eras in science fiction, so, for instance, if you put in "Star Trek Borg" without anything else, you get something hyper-modern and disgustingly detailed:

Now, that looks really cool, but it doesn't look right. So, let's try that again. This time, the prompt is "star trek borg, 90s style costume with cheesy hoses and cabling done up to look like implants":

Still not perfect, but way closer to the actual thing.

Both Borg and Cybermen are pretty detailed character designs. They have lots of edges and visual noise. That's hard for AI image generators to replicate effectively. It's almost like camouflage.

 

[Jormungandr]

That makes sense.

At least the AI got the top of the Cybermen's helmets correct in the first image, lol. I wonder what "Dalek" would produce? I mean, Doctor Who has always been popular on the British areas of the net, but it's pretty niche in general.

But yeah, I guess AI isn't quite "there" yet.

Hmmm... does that mean that lesser-known vehicles with a lot of textures and greebles would be baffling to an AI? Also, "Xenomorph" would likely be the same, since the original design used everything from human and animals bones, to mechanical parts, to organic jelly for its design.

 

[King Krávoka]

At least the AI got the top of the Cybermen's helmets correct in the first image, lol. I wonder what "Dalek" would produce? I mean, Doctor Who has always been popular on the British areas of the net, but it's pretty niche in general.

Daleks have a thoroughly consistent and distinct shape. It should be easy mode for them.

 

[Jormungandr]

Daleks have a thoroughly consistent and distinct shape. It should be easy mode for them.

There are a lot of variants, though.

The lambasted New Paradigm Daleks, for example, look like giant sweets, while the Last Great Time War models look like bronzed death machines.

And then there are the Renegade and Imperial factions' models...