Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

[The Mass Shooter Ron Soye]

China bans deepfakes created without permission or for evil

'Deep synthesis service providers' otherwise free to create AI-generated humans in line with socialist values

www.theregister.com

China bans deepfakes created without permission or for evil • The Reg…

archived 13 Dec 2022 00:11:16 UTC

archive.ph

China bans AI-generated media without watermarks

China regulates generative AI tech with rules that aim to spur growth and ban deception.

arstechnica.com

China bans AI-generated media without watermarks | Ars Technica

archived 12 Dec 2022 23:05:46 UTC

archive.ph

 

[AnOminous]

China bans deepfakes created without permission or for evil

'Deep synthesis service providers' otherwise free to create AI-generated humans in line with socialist values

www.theregister.com

China bans deepfakes created without permission or for evil • The Reg…

archived 13 Dec 2022 00:11:16 UTC

archive.ph

China bans AI-generated media without watermarks

China regulates generative AI tech with rules that aim to spur growth and ban deception.

arstechnica.com

China bans AI-generated media without watermarks | Ars Technica

archived 12 Dec 2022 23:05:46 UTC

archive.ph

People need to make a bunch of deepfakes of Xi sodomizing Winnie the Pooh.

 

[Puff]

People need to make a bunch of deepfakes of Xi sodomizing Winnie the Pooh.

This but the other way around.

 

[Pocket Dragoon]

Spoiler: "2001, A Clockwork Odyssey"

 

[DeadwastePrime]

China bans deepfakes created without permission or for evil

'Deep synthesis service providers' otherwise free to create AI-generated humans in line with socialist values

www.theregister.com

China bans deepfakes created without permission or for evil • The Reg…

archived 13 Dec 2022 00:11:16 UTC

archive.ph

China bans AI-generated media without watermarks

China regulates generative AI tech with rules that aim to spur growth and ban deception.

arstechnica.com

China bans AI-generated media without watermarks | Ars Technica

archived 12 Dec 2022 23:05:46 UTC

archive.ph

they'll totally do it themselves tho

 

[Smug]

Some of the things I've made while lurking. Nothing too special.

Spoiler: attempted to show two dota2 heroes fighting using inpainting

Spoiler: pretentious open-to-interpretation artpiece made by accident

Spoiler: generic beautiful women in various outfits

Spoiler: kinda lewd?

Spoiler: lewd, one happy, one disappointed in her lack of pants

Spoiler: some H.R. Giger inspired stuff

 

[Pocket Dragoon]

My wife started playing with MJ too, this was her first attempt & variation.

Kawaii pastel version of Gustav Klimt's The Kiss

 

[Ahriman]

CG3DANKFUN on Instagram: "No to AI generated Images. #SupportHumanArtists #notoaiart Support us ° @cg3dankfun Want to get featured? Register in our website www.cg3dankfun.com and make your complete portfolio. We selected deserving artist everyday a

8,524 likes, 525 comments - CG3DANKFUN (@cg3dankfun) on Instagram: "No to AI generated Images. #SupportHumanArtists #notoaiart Support us ° @cg3dankfun Want to ge..."

www.instagram.com

 

[Pocket Dragoon]

Jawas

Tuskan Raiders
Han
Chewie
Luke
Vader
Boba Fett
Darth Maul

 

[The Mass Shooter Ron Soye]

Dezgo has added the ability to choose the model you want to use, from:

Stable Diffusion 1.4
Stable Diffusion 1.5 (default)
Stable Diffusion 2.1-b
Anything 3.0 (anime)
Waifu Diffusion 1.3 (anime)

As well as negative prompts and sampling method (DPM, Euler, Euler Ancestral, or DDIM). I don't know shit about the sampling method.

Spoiler: The last luggage thief you'll ever meet

1.5

2.1-b

 

[Pocket Dragoon]

Dezgo has added the ability to choose the model you want to use, from:

Stable Diffusion 1.4
Stable Diffusion 1.5 (default)
Stable Diffusion 2.1-b
Anything 3.0 (anime)
Waifu Diffusion 1.3 (anime)

As well as negative prompts and sampling method (DPM, Euler, Euler Ancestral, or DDIM). I don't know shit about the sampling method.

Gave this a shot: used SD 2.1b, DDIM sampler.

Negative prompts are:

Extra fingers, extra eyes, blurry lines, extra ears, missing limbs, missing fingers, unnatural skin color, blurred face, unsymmetrical faces, blurred background.

Cinematic screen grab of Zardoz designed by Ron Howard, with Sean Connery as Zed, directed by Ron Howard, in the style of Willow, 1987.

With DPM sampling

Cinematic screen grab of Sean Connery as Zed designed by Ridley Scott, in Zardoz directed by Ridley Scott, in the style of Blade Runner, 1979.

Without mentioning Sean Connery, and used Euler sampling:


Back to DDIM sampling, landscape.

(Edit: this is the AI sampling of a scene with scientist Charlotte Rampling & Connery, where she's showing him violent images to elicit emotional & physical responses. But it filled in Rampling's spot with another Connery)

Cinematic screen grab of Sean Connery as Zed designed by Ridley Scott, in Zardoz directed by Ridley Scott, in the style of Blade Runner, 1979.

I'll keep playing with this, it seems to be going somewhere.

 

[Steinercard]

I'm getting pissed off at the misinfo about AI generated pictures from salty artists doing constant "pixar style" or "anime style" self portraits. This stuff is super fun and I'm having a lot of fun tinkering with my own custom version of SD.

 

[Hokuriku]

View attachment 4067592

CG3DANKFUN on Instagram: "No to AI generated Images. #SupportHumanArtists #notoaiart Support us ° @cg3dankfun Want to get featured? Register in our website www.cg3dankfun.com and make your complete portfolio. We selected deserving artist everyday a

8,524 likes, 525 comments - CG3DANKFUN (@cg3dankfun) on Instagram: "No to AI generated Images. #SupportHumanArtists #notoaiart Support us ° @cg3dankfun Want to ge..."

www.instagram.com

If these people had any awareness about their supposed passion, they'd know that Duchamp already literally pissed on their archconservative gatekeeping ideas on what defines 'art' over a century ago. There are literally no more barriers anymore and it's not even anything close to a 'new' concept, sorry guys.
In any case this crusade is by extension essentially denigrating photography as an art form so I'm inclined to tell them to get fucked regardless.

 

[Pocket Dragoon]

Further testing of Midjourney prompts with Stable Diffusion, all with DPM sampling:

style of alex maleev, realistic, Civil War cavalry soldier on horse, dramatic background, cloudy, strong shadows, winter colors, detailed realistic, single panel.

Modified prompt:

style of alex maleev, realistic, detailed face, American Civil War, Confederate cavalry soldier on horse, dramatic background, cloudy, strong shadows, winter colors, detailed realistic, single panel.

Seed: 4292906816

Interestingly, the AI added frames to these:



Negative prompts:

Extra fingers, extra eyes, extra limbs, extra tails, extra legs, extra ears, missing face, missing legs, missing feet, missing arms, unnatural skin color, blurred face, unsymmetrical faces, blurred background.

 

[Muu]

This looks amazing the way it creates scenes from films and directors the absolute seething over this is ridiculous.

Saw this on my twitter feed today doxed the guy too, prompt engineer is better than calling them artists or ai artists is it not?
https://twitter.com/CaraidArt/status/1603008804328976384?s=20&t=C83gWTTuWU1NyiKa5--p6A

 

[Starved Artist]

The amount of loathing and seething towards Ai at the moment is at an absolute premium.

>ZOMG CALL DISNEY LOOK AT WHAT AI IS DOING.


Meanwhile, on Etsy.


Nobody gives a fuck, in fact. If I didn't want to remain in the shadows of Twitter, I could easily direct this faggot to the decades of Fair Use coats and jackets people have been wearing since I was a kid in the hood.

 

[Starved Artist]

Anyway. Enough loathing.

Ai art is so much more work than regular art, if you have any semblance of standards. The compositions are off in a lot of images, the focal points are wrong, the lighting is wrong. In fact, There's always a bit of work to be done, like denoising and sharpening.

Now, Man Made Horrors beyond our Comprehension is one of the BEST brainstorming prompts I've ever fucking seen in my life. Almost everything that comes out of that tag is useless, as is, but if you can actually draw or compose art, wow.

Also, mockery is another great thing for Ai. Can't find the right reaction image? Make one.


Now if only ai could solve my money problems. I asked it and um... I'm not sure what to make of this.

 

[Pocket Dragoon]

From Midjourney Official:
Nervous Elizabethan Raccoon Assassins


Edit: more messing around with other prompts

Cinematic silent film screen grab of Blade Runner directed Fritz Lang, background designed by H.R. Geiger, 1922

 

[Viscious Seviper]

Yes, we should totally install multiple programs recommended from Nu-4chan. They're totally not some kind of fed backdoor into your computer.

 

[ZazietheBeast]

Japanese art community is passing around a Google form which asks you if you can tell the difference between ai-generated art from this tool and human made art. The results may surprise you. Some prompts are NSFW.

The quiz is here.

https://docs.google.com/forms/d/e/1FAIpQLSdhEpBRnOwiFI-ieNKKu3Y0KcoFbd_ZWod1LeyoV6EEfY78HA/viewform

Human or AI

archived 17 Oct 2022 02:30:36 UTC

archive.ph

Spoiler: results

Results archive:

Human or AI result - Google Slides

archived 18 Oct 2022 13:34:42 UTC

archive.ph

View attachment 3746640

View attachment 3746641

View attachment 3746642

View attachment 3746643

View attachment 3746644

View attachment 3746645
slides which show what people voted for example:
View attachment 3746654

In conclusion, no, ai is not replacing experienced artists at the moment. It IS replacing artists with poor skill and abstract or minimalistic style, however. Ai has poor understanding of three-dimensional space around characters that create awkward scenes and poses. Experienced artists are able to tell which are ai and which are human, but some ai work is bewilderingly good.

39/45. I'd make an ok Bladerunner.