AI generated content.

FirstBastion

" Illustrator discovers her art was 'STOLEN' to train an AI art generator "

So TRAINING is the new word for copyright theft. Similar to the word SHARE on those warez sites. 

 

He did have the decency to attach her name to it atleast  while he was violating her copyright.

Griffin Avid

He did have the decency to attach her name to it atleast  while he was violating her copyright.

The name was attached to make it wasy to find and sort/show up in the search field, like a tag. 

Not to give her any sort of credit....lol (you knew that)

If you created a bunch of art, it's already in one of the search directories.

And/or - despite all the written claims- there is some live searching going on.

A friend used someone who is established, but not famous, and yep, their book covers were emulated...and we could find the exact covers that they were puilled from.

And since there weren't a thousand images to source, it looked a lot like the original - just with the scattered quality....

nonesuch00

All this training and sharing was the original premise behind Google to begin with. Come join the community! Wink, wink, nudge, nudge.  

You could always switch to only physical media but even that is not safe as here locally the police posted surveillance camera footage yesterday asking for help to nab the thief who stole 2 giant flower pots on someone's front door stoop.

wolf359

Shutterstock is embracing the reality and adjusting their business model it would seem.

I had just named shutterstock as one of the businesses who would be wise to adjust their business model,
in my recent editorial Video.

MelanieL

WendyLuvsCatz said:

I just have a thing for cats in armour

have posted others in other threads, did more today

these made using DeepDreamGenerator

Cats?! How about a dormouse in armour? (from Nightcafe)

WendyLuvsCatz

MelanieL said:

WendyLuvsCatz said:

I just have a thing for cats in armour

have posted others in other threads, did more today

these made using DeepDreamGenerator

Cats?! How about a dormouse in armour? (from Nightcafe)

awe he/she is cute 

Byrdie

Where do cases like this fit, I wonder? He had access to the original works, but no mention of who owns/owned the copyright? Also no idea if he plans to distribute his model, in any case is it art theft, tribute, or totally legit? Things aren't always as clear cut as one might wish. 

https://www.pcgamer.com/nvidia-rtx-4090-stable-diffusion-training-aharon-kahana/

On the other hand, yeah, there are a lot of jackasses out there ... and I don't mean donkeys.

ArtAngel

I thought I would share this. Some food for thought. It's old but relevant. At nineteen, Robbie Barrat was viewed as genius when it comes to AI.

He was approached by three french students to use his code. They did not disclose it was for commercial gain.

Here's his response on October 24, 2018 before auction was held.

https://twitter.com/videodrome/status/1055285640420483073?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1055285640420483073%7Ctwgr%5E5b634aeb3e8e56f56d535eb8922ceb409c52ddad%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Ftheconversation.com%2Fnew-ai-art-has-artists-collaborators-wondering-who-gets-the-credit-112661

Christies' Auction of 'devrivitive work' 'created by AI', expected to sell at 7k-10k actually sold for $435,000.00. The work was not derivitive work created by a machine. It was devrived from Barrat's code.

https://www.theverge.com/2018/10/23/18013190/ai-art-portrait-auction-christies-belamy-obvious-robbie-barrat-gans

Psyckosama

AI art is a tool. Nothing more, nothing less, and is not going to replace traditional art. Instead, it's going to supliment it and honestly, the artists who thrive are going to be the ones who know how to use it to streamline thier own workflow because at the end of the day, AI art is painting with a hammer. Great for banging out the general shape, but you still have to chisel in the details.

Ghosty12

Psyckosama said:

AI art is a tool. Nothing more, nothing less, and is not going to replace traditional art. Instead, it's going to supliment it and honestly, the artists who thrive are going to be the ones who know how to use it to streamline thier own workflow because at the end of the day, AI art is painting with a hammer. Great for banging out the general shape, but you still have to chisel in the details.

Exactly, those who know how to get the best from AI, will be wondering why they had not gotten onboard earlier. And the thing is there is more to AI than just creating art, there are options to enhance and more an existing piece. Also there is AI that can colorize remarkable well, B&W pictures also being able to animate pictures of people long past.

WendyLuvsCatz

I like that it can create seamless textures 

even of things like roses on a trellis, crowds of cats or scenery not just fabric, stone, wood etc

IceCrMn

Looks like these "AI" bot makers are going to get their day in court.

https://developers.slashdot.org/story/22/11/05/1257233/microsofts-github-copilot-sued-over-software-piracy-on-an-unprecedented-scale

This story is about one that scrapes other peoples code snipets instead of images.

They were trying to make a "Make Software" button.

 

edit: Here's a quote from one of the replies...

by buck-yar ( 164658 ) on Saturday November 05, 2022 @11:27AM (#63026519)

"...I haven't read the lawsuit but I'd guess there's a lawsuit because someone made a system for scraping code from github, then changing a few things to obscure its origin, then passing it off as their own. If someone did that in college, they'd probably be accused by the professor of plagiarism...."

Torquinox

Psyckosama said:

AI art is a tool. Nothing more, nothing less, and is not going to replace traditional art. Instead, it's going to supliment it and honestly, the artists who thrive are going to be the ones who know how to use it to streamline thier own workflow because at the end of the day, AI art is painting with a hammer. Great for banging out the general shape, but you still have to chisel in the details.

Your viewpoint is inconsistent with current use, presentation and marketing of AI art.

lorddayradon

The day I could successfully sell an AI generated image as an original work of art, and thereby steal the profits of the artists hard work, is the day I can deem it stealing.  If I can't sell it as an orginal work of art, and noone believes it is the orginal, at best, it would be deemed a failed forgery.

 

edit: I don't think anyone could take any AI image no matter how long they worked on it, to an art appraiser and have them stamp it as an original.

FSMCDesigns

MelissaGT said:

outrider42 said:

wolf359 said:

AI does not "create" art.....It "processes" existing artwork into something different. There is no creativity to it, it is a processing algorythim

 

Back in the Glory days of  the CG society forums the “Pro” Character modelers said similar things about people who used canned models Like Poser/Zygote/Daz or even Mocap instead of hand keyframing everything.

 

 

Movies Like “Final Fantasy the spirits within” and “Beowulf” were highly divisive and controversial at one time.

 

People can’t hold back the tidal wave of technology with angry internet screeds and a push broom.

 

This is happening. https://waxy.org/2022/11/invasive-diffusion-how-one-unwilling-illustrator-found-herself-turned-into-an-ai-model/

...and now I'm depressed... :( 

Agreed! that article is just another example of creators mentalities verse users mentalities just like we have in the 3D communities. Creators see piracy as a huge problem and users don't care about it or copyright because it doesn't affect them directly

WendyLuvsCatz

I personally am more excited about ai assisted editing

the ability to restore, improve and enhance 

taking old vintage footage and creating colourisd HD 60fps video 

the upscaling and face enhancements while generative GAN trained assumptions and not necessarily accurate are nonetheless pretty awesome 

turning 2D images and particularly video/movie into 3D point cloud models if one has enough angles of architecture lost to time

Snow

deleted

PerttiA

Snow said:

https://avatarai.me/

I think the time has already come that people will start doubting if they still require 2D and 3D software to create their art, thats how fast this is evolving.

Why create/render for hours in 3D or put hours of retouching/compositing in 2D into something that can be created with AI in no time. I have put a lot hours into creating my own face in DAZ and with good result. Now you just train the AI with a few selfies and it will composite you into anything you desire and with good, sometimes even great and realistic quality.

...

...But Im not going to spend hours of work into something another person can create in a few seconds to achieve the same result either. I love the creative process and being able to use my talent (and will continue to do so for the time being) but Im not an fool. If people don't care if it's made by you or a bot then why still bother?

What you guys think, Am I talking rubbish?

Depends on what is it that one is interested in. If the goal is to get out 2D "art" to be sold or presented to masses somewhere, then AI may present a threat, although even then it doesn't make the artists obsolete, it just makes it more difficult to make money with the results.

But, I can't see AI replacing the current tools when for example creating 3D comics/novels, or making the current tools obsolete for us, who never show their work to anyone, to whom it is the world and the characters we 'create' that matters more than the occasional 2D renders.

WendyLuvsCatz

someone is making an Unreal plugin

I wonder when a DAZ studio plugin will pop up

wolf359

WendyLuvsCatz said:

someone is making an Unreal plugin

I wonder when a DAZ studio plugin will pop up

 

Looks like a workflow similar to the SD addon for Blender.

Gogger

Part 1:
UGH - edited to re-add images that were removed when I added a last image after posting. <sigh>

So, at what point will DAZ and A.I. Art collide?  I'm stoking an ember of an idea and wondering about the implications.  Take a DAZ figure and render them, let's say Alawa 8, in various angles and use those images to train an A.I. model on a home instance of, say, Stable Diffusion of some flavor or another.  Take that training and generate endless A.I. renders of Alawa 8 in various painterly styles, wardrobes, hairstyles, scenes, etc. All for the cost of a single figure.  

Question 1.  What are the useage implications?  Should one have an extended license for this sort of use?  Will DAZ need to come up with an A.I. Use License that you can buy?

Question 2. Why use a 3D model this way?

Question 3. Why NOT use a 3D model this way?

In these MidJourney v4 examples of using the image as a base - in no way considered 'training the A.I.' but merely used as a reference, I used a render I did of Alawa 8 just to see what would happen and if I might want to pursue this actual training of an A.I. (an EPIC PITA for me, as I am not a programmer, but can usually manage to figure things out if I try hard enough, long enough.)  I got some images that were clearly based on my original image, then some others that were downright surprising.

Alawa 8 render (mine)

Then this prompt generated this:
/imagine prompt: https://door44.com/goodz/012a_Alawa_Orbital-Suit_1920x1920.jpg, beautiful comanche woman, sci-fi, space pilot, DAZ Studio, Iray, 8K --ar 3:2 --v 4

And I forgot to enter the url info and got this on this prompt (basically the name of my rendered image):
/imagine prompt: 012a_Alawa_Orbital-Suit_1920x1920.jpg, beautiful comanche woman, sci-fi, space pilot, Karol Bak, 8K --ar 3:2 --v 4

And I would pay a premium for this space suit in DAZ format!!!!


And using the same base image I rendered, with this prompt, it changes a lot!
Same original render, but different prompt: (I know this is weak, but with a properly trained A.I. it could be AMAZING!)

/imagine prompt: https://door44.com/goodz/012a_Alawa_Orbital-Suit_1920x1920.jpg, old crone in a nightgown, very elderly woman, DAZ Studio, Iray, 8K --ar 3:2 --v 4

Gogger

Part 2:

Another example was using Nineve 6 and a render I did:

 

And the prompt and resulting MidJourney A.I. image:
/imagine prompt: https://door44.com/goodz/010b_Ninive_Elf.jpg, Roman bathhouse, beautiful elven woman, romantic, flowers, Karol Bak, 8K --ar 3:2 --v 4

and

Gogger

Part 3:

Coral for Callie 6 - my render:



And a resulting A.I. version using the prompt:
/imagine prompt: https://door44.com/goodz/011a2_Coral_Mermaid_DreamLight_Waves_VT.png, Roman bathhouse, beautiful elven woman, romantic, mermaid surrounded by flowers in an undersea meadow, Karol Bak, 8K --ar 3:2 --v 4



And sometimes what you get makes no sense at all!  But it makes you think.
Based on promo for Ziva

Prompt:
/imagine prompt: https://gcdn.daz3d.com/p/19544/i/00-main-ziva-for-aiko-6-daz3d.jpg photograph of a beautiful woman, --ar 3:2 --v 4




And that same promo image generated these. <shrug>

Gogger

Part 4:

Using the main promo image for Aiko 7, I generated these test images. Honestly, I would have preferred to keep Aiko's look intact - I love her look so much, but this sort of illustrates getting a similar look and some variation on clothing.



Using this prompt generated all the images below with no changes:
/imagine prompt:  https://gcdn.daz3d.com/p/31487/i/00-daz3d_aiko-7_.jpg photograph of a beautiful woman, --ar 3:2 --v 4

 

 

Richard Haseltine

Gogger you have removed, or the forum removed as you posted, most of your atachemnts so the images are missing (only those with a cached copy will see them).

Gogger

Richard Haseltine said:

Gogger you have removed, or the forum removed as you posted, most of your attachments so the images are missing (only those with a cached copy will see them).

Thanks, I know. I just re-added them, like FOUR TIMES! Hopefully they stick this time! Can you see them now? They are all there for me after a page refresh. Sorry everyone!

WendyLuvsCatz

VOC has Dreambooth but I believe I read somewhere it needs a 24GB VRAM graphics card so no use to me or I would train a it with a few of my favourite DAZ characters

I know I don't have enough VRAM but it doesn't tell me much more than that

yes is here https://docs.google.com/document/d/1xHSHEohmCOLlhdCY0ox4EARFKKU29XbFd8ji8UgjGn4/edit

WendyLuvsCatz

well I decided to try it anyway with 5 resized selfies 

(my ugly mug)

and it's training 

using under 8GB VRAM so far

sharing are about a year apart pre pixie cut, after and regrowth (middle current look)

it ran out of RAM but there is a ckpt file I pruned and am now testing, so far the results look horrifying

will update post with images if get something that passes as human

it finds the WendyVain_Woman.ckpt under settings but these don't look like me

 

wolf359

@Gogger

Amazing images!!
Very inspriing to see this new tech advancing

Richard Haseltine

Gogger said:

Richard Haseltine said:

Gogger you have removed, or the forum removed as you posted, most of your attachments so the images are missing (only those with a cached copy will see them).

Thanks, I know. I just re-added them, like FOUR TIMES! Hopefully they stick this time! Can you see them now? They are all there for me after a page refresh. Sorry everyone!

Consecutive posts with attachments seem to be an issue, the previous attachments don't get properly cleared and so are lost on making the post. I have found that refreshing after the post, before making the next, usually works.