Loading...
> >

AI is going to be our biggest game changer

1…1112141617…48

Comments

  • PendraiaPendraia Posts: 3,598

    Byrdie said:

    Question is, why does U.S. Copyright Office get to decide for anywhere that's NOT the U.S.? I think this is the sort of thing that individual countries should make rules about under their own copyright laws, then if need be negotiate differences via treaties and trade talks or whatever. The AI does not make anything entirely on it's own, there must be human involvement. If I can get copyright for an image by pressing the button on my camera, I should be able to do so by performing what is roughly the equivalent action with this new technology. 

    The difference is that other artworks have originally been fed into the ai program to allow it to learn. Some of the images are quite derivative in terms of the way they look.

  • PendraiaPendraia Posts: 3,598

    Mada said:

    WendyLuvsCatz said:

    barbult said:

    WendyLuvsCatz said:

    one can use it to texture 3D stuff too, I made some seamless in Gimp, others like the framed ones I can use as is

    I like your floral design with gold wires.

    feel free to click and download, no copyright on AI art wink 

    AI art is not copyrighted because its done by a machine - however once there's human input and reworked into a new art piece then it becomes copyrighted. Similar to how original William Morris textures are copyright free - but wallpaper and materials that's been updated and retouched are not, that particular image belongs to the artist/company who worked on the project. :)

    Isn't William Morris work dependent on the age of the work? I thought there was a timeline after which stuff became in the public domain...

  • MadaMada Posts: 2,016

    Yes. But artwork derived from it is not. Which is why Mehndi ran into a problem long ago. She didnt use the original textures, but images created by another company that was based on the Morris textures - and those images were copyrighted.

    Blast from the past :)

  • ByrdieByrdie Posts: 1,783

    Pendraia said:

    The difference is that other artworks have originally been fed into the ai program to allow it to learn. Some of the images are quite derivative in terms of the way they look.

     Humans also learn by looking at things. Nothing was ever created in a vacuum, 100% totally free of outside influences. We all learn by studying the works of the masters, even imitating their style until, hopefully, we grow into our own. And we are all inspired by the culture around us. Yes, there will very likely need to be new copyright legislation to cope with the new technology -- it shouldn't be a total Wild West -- but common sense ought to prevail. It's not, afaik, illegal for anyone to paint like Picasso. But it is indeed illegal to offer their Picasso-styled paintings for sale as if they were the original artist's work. 

  • PixelSploitingPixelSploiting Posts: 898

    Some fields of art are so heavily formalized you could probably use AI for making them. Just feed it right rules.  Generic cliparts and stock images also might end replaced by AI generation.

     

    Concept arts and sequential storytelling, on the other hand, require someone making conscious decisions on every level. I can't imagine art directors camping entering text after text and waiting for the machine to eventually (maybe) generating something in the graphics direction they wanted when they can tell a concept artists what kind of design they need.

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    https://stability.ai/stablediffusion-terms-of-service?fbclid=IwAR0dBjWghZwC6KU5vgSISXAwzEhkoXfaSzGi2goQZqt9czdItUSeFJIip-U#:~:text=Note%2C%20that%20while,including%20commercial%20purposes.

  • PendraiaPendraia Posts: 3,598

    Mada said:

    Yes. But artwork derived from it is not. Which is why Mehndi ran into a problem long ago. She didnt use the original textures, but images created by another company that was based on the Morris textures - and those images were copyrighted.

    Blast from the past :)

    Interesting. ..

  • PendraiaPendraia Posts: 3,598

    Byrdie said:

    Pendraia said:

    The difference is that other artworks have originally been fed into the ai program to allow it to learn. Some of the images are quite derivative in terms of the way they look.

     Humans also learn by looking at things. Nothing was ever created in a vacuum, 100% totally free of outside influences. We all learn by studying the works of the masters, even imitating their style until, hopefully, we grow into our own. And we are all inspired by the culture around us. Yes, there will very likely need to be new copyright legislation to cope with the new technology -- it shouldn't be a total Wild West -- but common sense ought to prevail. It's not, afaik, illegal for anyone to paint like Picasso. But it is indeed illegal to offer their Picasso-styled paintings for sale as if they were the original artist's work. 

    The problem is the image I saw looked like his and it would be easy to fake and scam people. I'm not saying anything other than it needs to be thought about and discussed. I don't have the answers I'm just seeing a lot of grey here. 

  • MelissaGTMelissaGT Posts: 2,611
    edited September 2022

    Byrdie said:

    Question is, why does U.S. Copyright Office get to decide for anywhere that's NOT the U.S.? I think this is the sort of thing that individual countries should make rules about under their own copyright laws, then if need be negotiate differences via treaties and trade talks or whatever. The AI does not make anything entirely on it's own, there must be human involvement. If I can get copyright for an image by pressing the button on my camera, I should be able to do so by performing what is roughly the equivalent action with this new technology. 

    The difference is that with photography (digital or film), you still need to understand the concepts of composition, lighting, shadow, etc etc...versus with things like Midjourney, you type words into discord and the AI spits out an image for you. In the time I had before my Midjourney demo expired, I did not see any control over things like the rule of thirds, or how to frame my subject. Words like "heroic lighting" and "ultra-realism" or "Unreal render" are what people use as prompts for the AI to guess at what to do for lighting, etc. 

    The argument that images made by an artifical intelligence is just like the advent of the digital camera is like comparing apples and buffalo wings (in my opinon). With a digitial camera, the laws of photography still apply and are no different than those used by film photographers. The only difference is that the image is exposed onto a digital sensor rather than a piece of light-sensitive negative. Post work is just as important with film photography as it is with digitial photography...the only difference is that it is done with chemicals in a dark room rather than Photoshop. Yes, a person holding a camera can hit an auto button (even later film cameras had/have auto buttons), but if that person doesn't understand the concepts of composition, lighting, etc...they are at a disadvantage from someone who does and uses that skill to ply their trade. Understanding what f/3.5-5.6 and a shutterspeed of 1/50s means and being able to utelize that knowledge only adds to the photographer's skillset (regardless of digital or film). Most of these same concepts can even be applied to digital artwork, even pieces made with Daz (think of the differences between dropping a few props into a scene with preset lighting and hitting the render button, vs a user who makes careful consideration of form and subject down to the very last detail). With things like Midjourney, the AI is using a best guess to spit out something the viewer wants to see and it's a roll of the dice as to what comes out. It's like telling a computer that I want it to write me a story about elves and dwarves and out pops The Hobbit. Did I write it, or did the computer write it? And should I be able to claim ownership of it?  

    Post edited by MelissaGT on
  • ByrdieByrdie Posts: 1,783

    The Discord I am in has had much discussion around this and how to improve their renders. Those who understand composition, lighting, etc. have been using it in their prompts and teaching others. Sure, there'll be people for whom whatever the AI throws up that's closest to their text will do, but many more are hitting the art studies books and the photography websites in search of actual knowledge. 

  • Ghosty12Ghosty12 Posts: 2,060
    edited September 2022

    There was that story going around how a AI created piece of art was entered into an art competition and won, and a lot of people were very miffed because of it.

    Post edited by Ghosty12 on
  • juvesatrianijuvesatriani Posts: 556
    edited September 2022

    if anyone really concern about the ethical those AI tools  ,just  use (ONLY) IMG2IMG  features with your own Images or Renders ( lowest INT_stregth )

    In theory it should be speeding up our correction/refining works using prompt definition .  In my case , I`m using it to add clean outline or cel shading style - FOG-Smoke and sometimes to check if having different color schemes maybe can give me better results .And yes sometimes you still need doing manual correction in 2D editor to make those FX seamless with your own original works 

    In my own perspective , thats depend about what your intent or  how you will use those tech . Abuse or use it with wise mindset ?

    In the end no matter how great those AI tools will become , High Skilled artist will be still needed or get hired  

    Post edited by juvesatriani on
  • ServantServant Posts: 759

    Pendraia said:

    outrider42 said:

    There is already a push by unionized artists in various fields to stop AI from taking their roles. This will only grow as AI works get more attention.

    https://www.bbc.com/news/technology-61166272

    This next link does not mention unions, but a recent news article that used an AI generated picture drew instant backlash from artists.

    https://mixed-news.com/en/artists-and-designers-protest-against-ai-generated-graphics/

    So I believe it is only a matter of time that their efforts will result in new copyright laws and usage restrictions that will curtail a lot of AI generated works from gaining too much ground. I think it will come down to how the AI was created. If you create an AI yourself using materials you have ownership of or permission to use, then you will be fine. But if you use one of these online AI generators in this thread, there will be no way to copyright your work.

    The catch is obvious, not many people or companies are going to make their own AI. There will be exceptions, like I can see Disney creating an AI to replicate deceased actors, but even here they still need permission from the estate to use the likeness in the first place.

    I know the Stable Diffusion says you own the results you get, but they also make it clear in the EULA that the laws of your country take priority:

    "You agree not to use the Model or Derivatives of the Model:

    - In any way that violates any applicable national, federal, state, local or international law or regulation"

    Obviously this would include any copyright laws.

    Additionally, the EULA for Stable sounds restrictive enough to limit naughty boys and girls from using it for their naughty purposes. So you naughty Daz artists out there have little to fear, LOL.

    There will be all kinds of cool things you can do with this stuff, no doubt. But do not expect to be able to use them commercially so easily, and they will not replace real artists that easily, either, no more than a stock image seller has.

    This has been my concern all along. Who will own the ai. I see some beautiful ai images recently obviously derived from a particular artists work. This is such a grey area.

    No one will own the AI generated image (or AI generated anything). Not even the company that created the AI program (such as Midjourney). No matter what legalese they put in their user agreement, as long as laws do not change, A.I. produced art, voice, programs, etc. are effectively PUBLIC DOMAIN as the key element in copyright and trademarks are HUMAN AUTHORSHIP. The law is clear in that it can be machine aided, but not produced by the machine.

    Typing in prompts that the AI will search for existing content (art,photos, etc.) is not authorship... it's rolling the dice and hoping for a good result.

    It is unlikely the U.S. laws will change as this is an area where even Democrats and Republicans agree with. Unions and voting public will be very vocal in deflecting A.I. to be considered actual authorship, as it would lead to further potential labor threats down the road.

    So while A.I. "art" can be nice as a tool for conceptualizing and for other things, an artist would still need to put in the work to make it anything of value.

    https://www.copyright.gov/comp3/chap300/ch300-copyrightable-authorship.pdf

  • ByrdieByrdie Posts: 1,783

    I see it as an extra, a helper not a replacement. And if everybody agrees the raw output is public domain but the stuff that somebody adds to and builds on -- be it with Photoshop or Daz or whatever -- is copyrightable to the artist that puts in the work on it, that seems reasonable and fair to me.

  • PendraiaPendraia Posts: 3,598

    Servant said:

    Pendraia said:

    outrider42 said:

    There is already a push by unionized artists in various fields to stop AI from taking their roles. This will only grow as AI works get more attention.

    https://www.bbc.com/news/technology-61166272

    This next link does not mention unions, but a recent news article that used an AI generated picture drew instant backlash from artists.

    https://mixed-news.com/en/artists-and-designers-protest-against-ai-generated-graphics/

    So I believe it is only a matter of time that their efforts will result in new copyright laws and usage restrictions that will curtail a lot of AI generated works from gaining too much ground. I think it will come down to how the AI was created. If you create an AI yourself using materials you have ownership of or permission to use, then you will be fine. But if you use one of these online AI generators in this thread, there will be no way to copyright your work.

    The catch is obvious, not many people or companies are going to make their own AI. There will be exceptions, like I can see Disney creating an AI to replicate deceased actors, but even here they still need permission from the estate to use the likeness in the first place.

    I know the Stable Diffusion says you own the results you get, but they also make it clear in the EULA that the laws of your country take priority:

    "You agree not to use the Model or Derivatives of the Model:

    - In any way that violates any applicable national, federal, state, local or international law or regulation"

    Obviously this would include any copyright laws.

    Additionally, the EULA for Stable sounds restrictive enough to limit naughty boys and girls from using it for their naughty purposes. So you naughty Daz artists out there have little to fear, LOL.

    There will be all kinds of cool things you can do with this stuff, no doubt. But do not expect to be able to use them commercially so easily, and they will not replace real artists that easily, either, no more than a stock image seller has.

    This has been my concern all along. Who will own the ai. I see some beautiful ai images recently obviously derived from a particular artists work. This is such a grey area.

    No one will own the AI generated image (or AI generated anything). Not even the company that created the AI program (such as Midjourney). No matter what legalese they put in their user agreement, as long as laws do not change, A.I. produced art, voice, programs, etc. are effectively PUBLIC DOMAIN as the key element in copyright and trademarks are HUMAN AUTHORSHIP. The law is clear in that it can be machine aided, but not produced by the machine.

    Typing in prompts that the AI will search for existing content (art,photos, etc.) is not authorship... it's rolling the dice and hoping for a good result.

    It is unlikely the U.S. laws will change as this is an area where even Democrats and Republicans agree with. Unions and voting public will be very vocal in deflecting A.I. to be considered actual authorship, as it would lead to further potential labor threats down the road.

    So while A.I. "art" can be nice as a tool for conceptualizing and for other things, an artist would still need to put in the work to make it anything of value.

    https://www.copyright.gov/comp3/chap300/ch300-copyrightable-authorship.pdf

    But given that it is based on the work of other artists...should it be public domain? If I was one of the artists whose work it was based on, I wouldn't be very happy about it. Would you?

  • PendraiaPendraia Posts: 3,598

    Byrdie said:

    I see it as an extra, a helper not a replacement. And if everybody agrees the raw output is public domain but the stuff that somebody adds to and builds on -- be it with Photoshop or Daz or whatever -- is copyrightable to the artist that puts in the work on it, that seems reasonable and fair to me.

    this is different to what the Servant said just above you, which illustrates to me that there is confusion about whether this is the case or not.

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    meanwhile in the Facebook groups they are arguing about people stealing their prompts cheeky

     

  • PendraiaPendraia Posts: 3,598

    lol...too funny Wendy!

  • ByrdieByrdie Posts: 1,783
    edited September 2022

    Dunno about Facebook AI groups, I have an account but I mostly use it for playing games. Have not checked out the scene there. Prompts are like recipes, anybody can use 'em but no guarantee the results any two users will get are gonna be the same or even remotely similar. I could provide examples here but well, not exactly safe for work. Right now I'm trying to figure out why I'm suddenly getting incorrectly assembled cats. And dogs. 

    (edited for misspelling)

    2022-09-08-17-20-42-1--1-scale-1.00--1.png
    512 x 512 - 433K
    2022-09-08-17-23-21-3--1-scale-1.00--3.png
    512 x 512 - 369K
    2022-09-07-23-23-04-4--1-scale-1.00--4.png
    512 x 512 - 464K
    Post edited by Byrdie on
  • evacynevacyn Posts: 975

    Byrdie said:

    I see it as an extra, a helper not a replacement. And if everybody agrees the raw output is public domain but the stuff that somebody adds to and builds on -- be it with Photoshop or Daz or whatever -- is copyrightable to the artist that puts in the work on it, that seems reasonable and fair to me.

    How about the artist gets 50% and the other 50% goes in a fund to prevent the AI from becoming sentient? :P

  • PangurBanPangurBan Posts: 28

    There are numerous interesting technologies on their way in this arena - for example, in relation to 3D, consider https://mrtornado24.github.io/IDE-3D/

    I get the feeling "we ain't seen nothing yet"...

  • ByrdieByrdie Posts: 1,783

    I would say "Wow!" but that doesn't quite do it justice. Got a feeling only the really big players will be able to afford or run the likes of that. The potential, for both good and bad, is just ... well, consider me mind-boggled. And just a wee bit envious of the kids who will get to play with those kind of toys.surprise

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484
    edited September 2022

    a video of beautiful AI art I prompted away at on my PC while eating pizza

    life as a creator is hard cheeky

    best viewed vertically on a phone or tablet in bed before going to sleep

    sweet dreams heart

    Post edited by WendyLuvsCatz on
  • ByrdieByrdie Posts: 1,783

    Interesting. Maybe got a Rapunzel vibe going on there. I did a couple last night, trying to get Aphrodite fantasy goddess type from a Discord prompt. Might post some later, when I check to see how Daz-safe the results are. 

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    lots of AI music videos popping up now

    most using song lyrics 

    for nightmares instead of sweet dreams you may inflict this one by me upon your eyeballs

     

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    https://pharmapsychotic.com/tools.html

    a collection of useful links

  • ByrdieByrdie Posts: 1,783
    edited September 2022

    Thanks! Have you used gradio or similar for Image2Image? I've been using it on some of my SD/Midjourney stuff and gotten interesting, amazing, horrible and downright ludicrous results. I also ran some of my own art through it. Here's a couple of samples. The first one I did in Poser many years ago. The others are Stable Diffusion Image Variations I made yesterday on the gradio app site.  Mind, they still need more work before they're gallery ready, but very promising so far.

    Beloved Prince_ALT.jpg
    750 x 743 - 658K
    download (13).png
    512 x 512 - 413K
    download (12).png
    512 x 512 - 494K
    Post edited by Byrdie on
  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    a corner of my loungeroom last picture

    2022-9-5-20-26-15-1-corner_of_a_luxurious_palatial_room-636456012-scale8.00-k_euler_a.png
    640 x 640 - 728K
    2022-9-5-20-26-15-2-corner_of_a_luxurious_palatial_room-636456013-scale8.00-k_euler_a.png
    640 x 640 - 710K
    2022-9-5-20-26-15-3-corner_of_a_luxurious_palatial_room-636456014-scale8.00-k_euler_a.png
    640 x 640 - 730K
    2022-9-5-20-26-15-4-corner_of_a_luxurious_palatial_room-636456015-scale8.00-k_euler_a.png
    640 x 640 - 746K
    IMG-1141.JPG
    3264 x 2448 - 2M
  • metasidemetaside Posts: 178

    WendyLuvsCatz said:

    https://pharmapsychotic.com/tools.html

    a collection of useful links

    I just wanted to say thanks for all the links you are posting in this thread! I'm still kinda on the sidelines about AI art generation in its current state (never prompted something I really liked and some of the stuff I see is impressive for sure but it usually doesn't really touch me emotionally, can't really describe it better than that), but I'm interested in machine learning in general and all the tools are super interesting to me!

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,484

    well this is embarrassing 

    I just mistook a DAZ render for a Midjourney generated image on Facebook blush

    to be fair the Friend had also been sharing their AI Art

This discussion has been closed.