pancakes

MicrostockGroup Sponsors


Author Topic: Artificial Intelligence killing the whole industry  (Read 81489 times)

0 Members and 1 Guest are viewing this topic.

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #375 on: January 14, 2023, 14:05 »
0
AI: friend or foe? (3 of 5) on my blog

A very brief history of art: today

https://luisafumi-digitalart.com/blog/2023/01/13/ai-friend-or-foe-3-of-5/


have fun  :D

OK so how does that table stand on those legs?  :o But that's the kind of thing that AI creates that's lacking science or physical logic. Not just picking, I think the things you have shown are very well done and attractive. Wait, does the right hand chair only have three legs? Or is that 5 or 6 legs?

Yeah, people in some of mine only have three fingers and a thumb.


« Reply #376 on: January 14, 2023, 21:33 »
0
What is the image size and resolution like on some of these AI generators? I one better than the other when it comes to that? DALLE vs Midjourney vs ??

No experience with Midjourney I'm not sure how to get in or do anything, so I just took a pass. Maybe someone can share the details?

DALL-E2 1024 x 1024 PNG with a little color sample bar in the lower right. I don't know if it's just me, but the lines in illustrations are not sharp and will get "quality" rejections at most places. People who know more than I do, say, 4x with Topaz Gigapixel AI or some other Topaz, edit and reduce to Microstock size.

I believe them, but I'm not spending $?? or full price $99.99 to make AI images that will sell for fractions of a dollar. If I thought I could get 400 DLs in a year from AI (1,000 on SSTK) images, I could justify that? 👍
That $99,99 is a plan for a year?

Uncle Pete

  • Great Place by a Great Lake - My Home Port
« Reply #377 on: January 15, 2023, 11:30 »
+1
What is the image size and resolution like on some of these AI generators? I one better than the other when it comes to that? DALLE vs Midjourney vs ??

No experience with Midjourney I'm not sure how to get in or do anything, so I just took a pass. Maybe someone can share the details?

DALL-E2 1024 x 1024 PNG with a little color sample bar in the lower right. I don't know if it's just me, but the lines in illustrations are not sharp and will get "quality" rejections at most places. People who know more than I do, say, 4x with Topaz Gigapixel AI or some other Topaz, edit and reduce to Microstock size.

I believe them, but I'm not spending $?? or full price $99.99 to make AI images that will sell for fractions of a dollar. If I thought I could get 400 DLs in a year from AI (1,000 on SSTK) images, I could justify that? 👍
That $99,99 is a plan for a year?

DALL-E2 is 15 images a month for free, you need a phone number to verify your registration. One account only, virtual numbers don't work. (Google Voice for example) For those who pay, 115 credits for $15 = free is fun, that's as far as I go.

Someone else who pays will have to give you the accurate facts about Midjourney or others. I see ads for some that start at $49 a month.

« Reply #378 on: January 16, 2023, 15:21 »
0
AI: friend or foe? (3 of 5) on my blog

A very brief history of art: today

https://luisafumi-digitalart.com/blog/2023/01/13/ai-friend-or-foe-3-of-5/



have fun  :D
keep 'em coming!  welcome balanced approach to luddite negativity!





Justanotherphotographer

« Reply #383 on: January 18, 2023, 06:09 »
+1
First lawsuits against AI: https://www.prnewswire.com/news-releases/class-action-filed-against-stability-ai-midjourney-and-deviantart-for-dmca-violations-right-of-publicity-violations-unlawful-competition-breach-of-tos-301721869.html
Have you seen the source?
If you read here
https://www.theverge.com/2023/1/16/23557098/generative-ai-art-copyright-legal-lawsuit-stable-diffusion-midjourney-deviantart
the law firm has no idea how AI works  ;D

Yeah weird. I would have gone for arguing that the AI company has made commercial use of images without a commercial licence or permission from the copyright owner (which they have, even if no storage or image retention is involved).

Whether storage of information obtained through studying peoples images counts as infringement is, I guess, more iffy. Surely theres no doubt the model the AI has of what an object or person looks like is derived from the study of other peoples copyright work, regardless of the level of processing involved, hence 21st-century collage tool (yes, I do understand the original image isn't used in its original form but it IS process into information to create the "new" work).

« Reply #384 on: January 18, 2023, 06:36 »
+2
First lawsuits against AI: https://www.prnewswire.com/news-releases/class-action-filed-against-stability-ai-midjourney-and-deviantart-for-dmca-violations-right-of-publicity-violations-unlawful-competition-breach-of-tos-301721869.html
Have you seen the source?
If you read here
https://www.theverge.com/2023/1/16/23557098/generative-ai-art-copyright-legal-lawsuit-stable-diffusion-midjourney-deviantart
the law firm has no idea how AI works  ;D

Yeah weird. I would have gone for arguing that the AI company has made commercial use of images without a commercial licence or permission from the copyright owner (which they have, even if no storage or image retention is involved).

Whether storage of information obtained through studying peoples images counts as infringement is, I guess, more iffy. Surely theres no doubt the model the AI has of what an object or person looks like is derived from the study of other peoples copyright work, regardless of the level of processing involved, hence 21st-century collage tool (yes, I do understand the original image isn't used in its original form but it IS process into information to create the "new" work).
we human make the same (in a smaller scale... ) but we call it inspiration or learning or whatever  ;)

« Reply #385 on: January 18, 2023, 06:39 »
+1
That will be a really interesting and great law war. Most of points are questionable.

Surely theres no doubt the model the AI has of what an object or person looks like is derived from the study of other peoples copyright work, regardless of the level of processing involved, hence 21st-century collage tool (yes, I do understand the original image isn't used in its original form but it IS process into information to create the "new" work).

This, for example, seems a description of the process that any artist do before creating his own art: the study of other people copyright works. And it's perfectly legit.

It seems to me that the only point really against AI is:
Does it really USE section (or pixels) of other copyrighted works? Are there in the AI generated images any "PART" of other copyrighted works?

In last weeks I was sure that the correct answer should be NO, THERE ARE'NT ANY PIXEL OF COPYRIGHTED WORKS IN GENERATIVE AI IMAGES

... but I read a lot about AI engines in these last weeks, and I have to say that it's not really so clear to define what AI "understand and recreate" and what, at the contrary, AI "copy, even only partly"

The legal war probably will need to clarify first of all this point. And, probably, the answer would not the same for ANY AI engine
« Last Edit: January 18, 2023, 06:41 by derby »

« Reply #386 on: January 18, 2023, 07:06 »
+4

we human make the same (in a smaller scale... ) but we call it inspiration or learning or whatever  ;)

No, we humans do not make the same and we do not call it inspiration.

In your constant defense of AI (understandable, because you never took the time and effort to learn and master an artistic skill yourself. All you ever did was vectorize other people's illustrations from old book, so you always just leeched off other people's work, without understanding what it really takes to create  artwork. Now you see an easy way to make money, still without having to learn the skill of creating art yourself) you are again misunderstanding how human art and AI art works.
 Inspiration means that we humans see, read or hear someone else's work and then we are mentally stimulated by what we see or hear.

We then might take influence from someone else's work, but we add a significant portion of our own creativity to it. AI can't do that. AI has no own creativity. 100% of what it has learned is based off human work.

Never confuse what an AI does with "inspiration". An AI is not capable of that.  https://en.wikipedia.org/wiki/Artistic_inspiration
« Last Edit: January 18, 2023, 13:44 by Her Ugliness »

Justanotherphotographer

« Reply #387 on: January 18, 2023, 07:35 »
+1
Yup, it's what's called "weak AI". I said my bit about the tendency of some on here to want to anthropomorphise it.

« Reply #388 on: January 18, 2023, 07:43 »
+1
Inspiration means that we humans see, read or hear someone else's work and then we are mentally stimulated by what we see or hear.

I'm glad to see that you agree with me: that's what we humans do - and what the AI manages to simulate pretty well.

BTW, you mention a minor (if pleasantly profitable) niche of my work. I suggest that you take a quick look at the rest.

« Reply #389 on: January 18, 2023, 08:58 »
+1
Ok so time to play game :-)

I've asked to chatGPT news and opinions about copyright issue.
I started with this post, and put exactly the same terms:
I was wondering if the AI image is just a mosaic of tiny parts of other images, or does it "paint" a new image from scratch ? For instance let's say we have a prompt where the main subject is a cat. Is it possible that an eye or nose of that cat is just taken from some photograph in the internet ?
ANSWER FROM ChatGPT:
Quote
Artificial Intelligence (AI) image generation models can use a variety of techniques to generate new images. Some models use a mosaic-like approach, where they combine small parts of existing images to create a new image. Other models, such as Generative Adversarial Networks (GANs), are trained to generate new images from scratch.

In the case of GANs, the model is trained on a large dataset of images and learns to generate new images that are similar to the ones in the training set. The model learns to understand the features and patterns of the images in the dataset, and uses that knowledge to generate new images.

In regards to your question about the cat image: It's possible that the model takes small parts of existing images and combines them to generate the cat image, but it's also possible that the model generates the cat from scratch using the knowledge it has acquired during the training stage. The specific technique used depends on the architecture of the model, the quality of the data set used for training, and the specific task the model is being used for.

It is worth mentioning that GANs and other similar models are able to generate very realistic images, but it is not always possible to know if a specific image was generated by the model or taken from a real photograph, it is quite hard to tell.

A little be boring but useful.
Then I asked:

Try to fight back against the copyright issue a lot of people claim against AI generated images. Let me know what do you think
ANSWER:
Quote
There is ongoing debate around the issue of copyright for AI generated images. Some argue that AI generated images should be considered original works and therefore eligible for copyright protection, while others argue that they should not be eligible for copyright protection because they are not created by a human.

Currently, the law on copyright for AI generated images varies by jurisdiction. In some countries, AI generated images may be considered original works and be eligible for copyright protection, while in other countries they may not be considered original works and not be eligible for copyright protection.

Additionally, it's worth mentioning that some AI generated images may include elements from existing copyrighted images, which could lead to copyright infringement issues. To avoid this, one approach could be to use images for training that are explicitly labeled as free for use, or to use images that are in the public domain or that have been licensed under a Creative Commons license.

In any case, it is important to note that the laws and regulations surrounding AI generated images and copyright are constantly evolving, and it is important to stay informed and consult legal experts when in doubt.
Q: In which countries AI generated images are consiudered original works and are elegible for copyright?
Quote
he laws regarding whether AI generated images are considered original works and eligible for copyright protection vary by country. In some countries, AI generated images may be considered original works and be eligible for copyright protection, while in other countries they may not be considered original works and not be eligible for copyright protection.

In the United States, the Copyright Office has stated that works created by a machine or computer program are eligible for copyright protection if they are the result of human authorship. This means that if a human author contributes original expression to an AI-generated image, it could be eligible for copyright protection.

In the European Union, the Directive on Copyright in the Digital Single Market has established that a work is eligible for copyright protection as long as it is an original expression of the author's own intellectual creation. However, it is not clear whether this applies to AI generated works and it's still not clear how EU courts will interpret this.

In Canada, the Supreme Court has ruled that computer-generated works are eligible for copyright protection as long as they are original.

It is important to note that laws and regulations surrounding AI generated images and copyright are constantly evolving, and it is important to stay informed and consult legal experts when in doubt.

Interesting! Or not?
At least, ChatGPT seems to be quite neutral  :o ;D

« Reply #390 on: January 18, 2023, 11:48 »
0
Ok so time to play game :-)

I've asked to chatGPT news and opinions about copyright issue.
I started with this post, and put exactly the same terms:
I was wondering if the AI image is just a mosaic of tiny parts of other images, or does it "paint" a new image from scratch ? For instance let's say we have a prompt where the main subject is a cat. Is it possible that an eye or nose of that cat is just taken from some photograph in the internet ?
ANSWER FROM ChatGPT:
Quote
Artificial Intelligence (AI) image generation models can use a variety of techniques to generate new images. Some models use a mosaic-like approach, where they combine small parts of existing images to create a new image. Other models, such as Generative Adversarial Networks (GANs), are trained to generate new images from scratch.

In the case of GANs, the model is trained on a large dataset of images and learns to generate new images that are similar to the ones in the training set. The model learns to understand the features and patterns of the images in the dataset, and uses that knowledge to generate new images.

In regards to your question about the cat image: It's possible that the model takes small parts of existing images and combines them to generate the cat image, but it's also possible that the model generates the cat from scratch using the knowledge it has acquired during the training stage. The specific technique used depends on the architecture of the model, the quality of the data set used for training, and the specific task the model is being used for.

It is worth mentioning that GANs and other similar models are able to generate very realistic images, but it is not always possible to know if a specific image was generated by the model or taken from a real photograph, it is quite hard to tell.

A little be boring but useful.
Then I asked:

Try to fight back against the copyright issue a lot of people claim against AI generated images. Let me know what do you think
ANSWER:
Quote
There is ongoing debate around the issue of copyright for AI generated images. Some argue that AI generated images should be considered original works and therefore eligible for copyright protection, while others argue that they should not be eligible for copyright protection because they are not created by a human.

Currently, the law on copyright for AI generated images varies by jurisdiction. In some countries, AI generated images may be considered original works and be eligible for copyright protection, while in other countries they may not be considered original works and not be eligible for copyright protection.

Additionally, it's worth mentioning that some AI generated images may include elements from existing copyrighted images, which could lead to copyright infringement issues. To avoid this, one approach could be to use images for training that are explicitly labeled as free for use, or to use images that are in the public domain or that have been licensed under a Creative Commons license.

In any case, it is important to note that the laws and regulations surrounding AI generated images and copyright are constantly evolving, and it is important to stay informed and consult legal experts when in doubt.
Q: In which countries AI generated images are consiudered original works and are elegible for copyright?
Quote
he laws regarding whether AI generated images are considered original works and eligible for copyright protection vary by country. In some countries, AI generated images may be considered original works and be eligible for copyright protection, while in other countries they may not be considered original works and not be eligible for copyright protection.

In the United States, the Copyright Office has stated that works created by a machine or computer program are eligible for copyright protection if they are the result of human authorship. This means that if a human author contributes original expression to an AI-generated image, it could be eligible for copyright protection.

In the European Union, the Directive on Copyright in the Digital Single Market has established that a work is eligible for copyright protection as long as it is an original expression of the author's own intellectual creation. However, it is not clear whether this applies to AI generated works and it's still not clear how EU courts will interpret this.

In Canada, the Supreme Court has ruled that computer-generated works are eligible for copyright protection as long as they are original.

It is important to note that laws and regulations surrounding AI generated images and copyright are constantly evolving, and it is important to stay informed and consult legal experts when in doubt.

Interesting! Or not?
At least, ChatGPT seems to be quite neutral  :o ;D

Astonishing!

Not so much that the answers remind pretty closely of the good old "My answer is maybe, and that's final!", as because they suggest that an AI can be way less biased and more honest than a human...  ;D ;D ;D

ADH

« Reply #391 on: January 18, 2023, 12:37 »
+2
Adobe, over 500k AI images for sale, 99% will never be sold because there is little or no use for them. By the end of the year, following the actual trend of uploads, it will be 5 millions for sale. When do you think Adobe will start charging for uploading AI generated garbage into their portal?


« Reply #392 on: January 18, 2023, 13:12 »
+1
AI: friend or foe? (4 of 5) on my blog

A very brief history of art: fight them or join them?

https://luisafumi-digitalart.com/blog/2023/01/17/ai-friend-or-foe-4-of-5/



a propos...

« Reply #393 on: January 18, 2023, 15:07 »
+1
Adobe, over 500k AI images for sale, 99% will never be sold because there is little or no use for them. By the end of the year, following the actual trend of uploads, it will be 5 millions for sale. When do you think Adobe will start charging for uploading AI generated garbage into their portal?

your source for any of those claims?

guess i'm finally part of the 1% since my AI images have sold on AS, SS and DT

« Reply #394 on: January 18, 2023, 17:16 »
+1
Adobe, over 500k AI images for sale, 99% will never be sold because there is little or no use for them. By the end of the year, following the actual trend of uploads, it will be 5 millions for sale. When do you think Adobe will start charging for uploading AI generated garbage into their portal?

your source for any of those claims?

guess i'm finally part of the 1% since my AI images have sold on AS, SS and DT
me too (in the 1%)!!!

« Reply #395 on: January 18, 2023, 18:09 »
0
What is the image size and resolution like on some of these AI generators? I one better than the other when it comes to that? DALLE vs Midjourney vs ??

No experience with Midjourney I'm not sure how to get in or do anything, so I just took a pass. Maybe someone can share the details?

DALL-E2 1024 x 1024 PNG with a little color sample bar in the lower right. I don't know if it's just me, but the lines in illustrations are not sharp and will get "quality" rejections at most places. People who know more than I do, say, 4x with Topaz Gigapixel AI or some other Topaz, edit and reduce to Microstock size.

I believe them, but I'm not spending $?? or full price $99.99 to make AI images that will sell for fractions of a dollar. If I thought I could get 400 DLs in a year from AI (1,000 on SSTK) images, I could justify that? 👍
That $99,99 is a plan for a year?

DALL-E2 is 15 images a month for free, you need a phone number to verify your registration. One account only, virtual numbers don't work. (Google Voice for example) For those who pay, 115 credits for $15 = free is fun, that's as far as I go.

Someone else who pays will have to give you the accurate facts about Midjourney or others. I see ads for some that start at $49 a month.

I have not used DALL-E2, because it ask for a phone number and in my country is not listed (i live in Venezuela). I have used MidJourney but in the Discord chat, because i don't know if that app is in another place. Also i have used a app called craiyon.com, but the image don't have size for microstock. Do you know some program that could be use for free for microstock image?

« Reply #396 on: January 18, 2023, 18:29 »
0
Adobe, over 500k AI images for sale, 99% will never be sold because there is little or no use for them. By the end of the year, following the actual trend of uploads, it will be 5 millions for sale. When do you think Adobe will start charging for uploading AI generated garbage into their portal?

your source for any of those claims?

guess i'm finally part of the 1% since my AI images have sold on AS, SS and DT
me too (in the 1%)!!!

Me too, only in Adobe stock, several dls of AI images in few days from start  ;D

« Reply #397 on: January 19, 2023, 07:16 »
+2
Definitely "not a collage".

Quite interesting and extremely clear first analysis of the class-action lawsuit against Stability AI, DeviantArt, and Midjourney by Dr. Andres Guadamuz, a Senior Lecturer in Intellectual Property Law at the University of Sussex.

https://www.technollama.co.uk/artists-file-class-action-lawsuit-against-stability-ai-deviantart-and-midjourney

« Reply #398 on: January 19, 2023, 12:03 »
+4

we human make the same (in a smaller scale... ) but we call it inspiration or learning or whatever  ;)

No, we humans do not make the same and we do not call it inspiration.

In your constant defense of AI (understandable, because you never took the time and effort to learn and master an artistic skill yourself. All you ever did was vectorize other people's illustrations from old book, so you always just leeched off other people's work, without understanding what it really takes to create  artwork. Now you see an easy way to make money, still without having to learn the skill of creating art yourself) you are again misunderstanding how human art and AI art works.
 Inspiration means that we humans see, read or hear someone else's work and then we are mentally stimulated by what we see or hear.

We then might take influence from someone else's work, but we add a significant portion of our own creativity to it. AI can't do that. AI has no own creativity. 100% of what it has learned is based off human work.

Never confuse what an AI does with "inspiration". An AI is not capable of that.  https://en.wikipedia.org/wiki/Artistic_inspiration

It's not nice to personally attack others on the forum and insult their work as leeching or unskilled.

« Reply #399 on: January 19, 2023, 12:05 »
0
Definitely "not a collage".

Quite interesting and extremely clear first analysis of the class-action lawsuit against Stability AI, DeviantArt, and Midjourney by Dr. Andres Guadamuz, a Senior Lecturer in Intellectual Property Law at the University of Sussex.

https://www.technollama.co.uk/artists-file-class-action-lawsuit-against-stability-ai-deviantart-and-midjourney
This is good, we need a lawsuit to bring this to the public and have some court decision if AI art is legal or not.


 

Related Topics

  Subject / Started by Replies Last post
63 Replies
40131 Views
Last post May 25, 2010, 05:52
by youralleffingnuts
8 Replies
10493 Views
Last post March 15, 2011, 05:28
by Microbius
42 Replies
15948 Views
Last post February 26, 2013, 01:09
by Xanox
6 Replies
6465 Views
Last post April 03, 2015, 01:36
by fmarsicano
22 Replies
7324 Views
Last post May 30, 2023, 17:08
by cobalt

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors