Associate
I wonder if there are people creating AI art who are then selling it off as art they did themselves ? Is there a way to trace AI art back to its source ?
Last edited:
Midjourney - AI art
- Thread starter tang0
- Start date
I'd imagine so for the people selling, Adobe and others have been developing software which can attempt to detect image editing and AI generated images but it is far from definitive.
The give away as things stand is that the stuff AI produces is often a interpretation of something which is lacking cohesion like the text in those images is mostly just random rubbish which looks something like text but mostly isn't.
Last edited:
Soldato
- Joined
- 13 May 2003
- Posts
- 8,933
Hi Peeps
Do any of these AI Art generators accept photos as part of the process. I'd love to surprise the kids with an AI generated picture with them as their favourite *insert IP here*.
Any recommendations greatly appreciated.
Do any of these AI Art generators accept photos as part of the process. I'd love to surprise the kids with an AI generated picture with them as their favourite *insert IP here*.
Any recommendations greatly appreciated.
Soldato
I'm also interested in this.
About a year ago (June 2022), there was Craiyon. I didn't know what AI art was back then. You could type any topic into Craiyon to generate wacky images and it became some sort of 'meme' in my WhatsApp and Signal group chats. I didn't think much of it.
Then I discovered Stable Diffusion. Again, type in any topic and it creates weird stuff for you.
stablediffusionweb.com
My best shot at it was probably "bird in the milky way".
Now the thing is, that these images are only 512x512, yet people on deviantART are uploading images of 2000x3000 (6MP). How is this being achieved? I messaged a few of these deviantART users on how they're doing it, but none of them ever respond to me. It's like they're trying to protect some secret recipe, while not hiding the fact tha they're using AI art because they put "aiart" and "stablediffusion" etc in the image's meta tags, and with "Created using AI tools" written underneath. The AI tools surely got to be public domain?
About a year ago (June 2022), there was Craiyon. I didn't know what AI art was back then. You could type any topic into Craiyon to generate wacky images and it became some sort of 'meme' in my WhatsApp and Signal group chats. I didn't think much of it.
Then I discovered Stable Diffusion. Again, type in any topic and it creates weird stuff for you.
Stable Diffusion Online
Stable Diffusion is a deep learning model that generates images from text descriptions. Use Stable Diffusion online for free.
My best shot at it was probably "bird in the milky way".
Now the thing is, that these images are only 512x512, yet people on deviantART are uploading images of 2000x3000 (6MP). How is this being achieved? I messaged a few of these deviantART users on how they're doing it, but none of them ever respond to me. It's like they're trying to protect some secret recipe, while not hiding the fact tha they're using AI art because they put "aiart" and "stablediffusion" etc in the image's meta tags, and with "Created using AI tools" written underneath. The AI tools surely got to be public domain?
![[FnG]magnolia](/styles/overclockers/avatars/9.png){kind=avatar}
Caporegime
^ Rob, we have our differences but that is rad as the proverbial. Well played.
Associate
Was this done with freeware, i need to have a play with this groovy tech.So, what would happen if you ask AI to envision Star Wars as a Blaxploitation pic from the 70's?
If you have a Microsoft account you can use the Bing image creator for free. It is based on Dall-E. Go to Bing->Images->Image Creator in top right
Just put the word "Xenomorph" in front of any random sentence for hours of fun.
E.g. "Xenomorph scoring the winning goal in the 1978 FA Cup Final"
Last edited:
So, what would happen if you ask AI to envision Star Wars as a Blaxploitation pic from the 70's?
I wonder how far you could go with this, could you get it recreate scenes in the same format, and create a comic script!
Yes, people are doing art commisions using AI art.
People are also selling reference/inspiration packs using AI art, those are a lot easier to tell.
In both instances, they don't explicitly state it is AI art.
Soldato
So, what would happen if you ask AI to envision Star Wars as a Blaxploitation pic from the 70's?
Middle top row there is a bit... ahem... (I know it's supposed to be chewbacca, but I'm sure that could easily be taken the wrong way by some people
).I think the paid version allows for larger res picturesI'm also interested in this.
About a year ago (June 2022), there was Craiyon. I didn't know what AI art was back then. You could type any topic into Craiyon to generate wacky images and it became some sort of 'meme' in my WhatsApp and Signal group chats. I didn't think much of it.
Then I discovered Stable Diffusion. Again, type in any topic and it creates weird stuff for you.
Stable Diffusion Online
Stable Diffusion is a deep learning model that generates images from text descriptions. Use Stable Diffusion online for free.
My best shot at it was probably "bird in the milky way".
Now the thing is, that these images are only 512x512, yet people on deviantART are uploading images of 2000x3000 (6MP). How is this being achieved? I messaged a few of these deviantART users on how they're doing it, but none of them ever respond to me. It's like they're trying to protect some secret recipe, while not hiding the fact tha they're using AI art because they put "aiart" and "stablediffusion" etc in the image's meta tags, and with "Created using AI tools" written underneath. The AI tools surely got to be public domain?
Soldato
That’s Throwbakka, he’s one of those retro brothers from the planet Harlem.Middle top row there is a bit... ahem... (I know it's supposed to be chewbacca, but I'm sure that could easily be taken the wrong way by some people
Don
I wonder if anyone can do better than Amazon's AI generated Fallout image...
(Look closely at the red car amongst other things)
kotaku.com
(Look closely at the red car amongst other things)
Amazon's First Official Fallout TV Show Artwork Is An AI-Looking Eyesore
Amazon's Fallout adaptation streams in 2024, but its first promotional image is full of bizarre errors
kotaku.com
Last edited:
Soldato
Is anything that's computer generated Art?
Its a select group though that makes the big bucks, most artists or graphic designers are doing pretty mundane stuff day to day but it provides an income. Those are the jobs that are at risk of being replaced by AI image generators, some will embrace it and others will struggle to stand out on marketplaces that are already flooded with AI art.
AI is also good at writing the prompts, so I'm not sure if an AI artist will really become a thing. You can already ask an AI tool to look at an image and break it down into prompts to create new content, you can also use AI to reference another image to add elements to a generated scene like adding a real face for example. You might need to tidy things up but its not really a skilled job.
There are also a lot of questions about who owns the copyright to AI generated art, what sources have been referenced (Getty Images for example are fighting that one) and can you protect generated content that you use for your business. I believe everything on midjourney is public and accessible unless you pay for the expensive tiers.
Its a very interesting tool though, especially for those of us who can't draw
Edit: This thread is old, AI art has moved on a lot since this thread started.
Last edited:
Soldato
Does anyone here have a layman's guide to AI art? Someone up-thread said that you do have to pay to get the 2000x3000 results and I don't mind paying purely out of curiosity. There are plenty of results on deviantART at that resolution and some of it is in breathtaking detail. I want to get started but they never say how they did it or what program or site they're using. If it's relevant, I have used ChatGPT a few times, asking it questions and getting it to write poems or texts in the style of a famous person.
Yes, because it still needs a creator with a vision and knowledge on how to put it together. If you just tell AI what you want it usually stuffs it up big time. In my eyes it's still an art to making these things
Honestly, there isn't a layman's guide to AI art, it's hard graft and loads of reading to learn how to prompt. A good starting point is joining the Discord of the art tool you want to use like Mid-Journey or Stable Defutions and start reading what everyone is doing. It's like reading Egyptian hieroglyphs at first put at some point you start to get it, you're basically learning to code with a more English based language code. I barely get it myself but I'm slowly making progress
Last edited: