Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

stablediffusionlitigation.com

Post reply on HN
Print view View original

121–130 of 473 posts

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#121

Post by AlotOfReading »

Salgat wrote: post #26
akjetma wrote: post #20

Earlier quoted context omitted.

I don't think you have to reproduce an entire original work to demonstrate copyright violation. Think about sampling in hip hop for example. A 2 second sample, distorted, re-pitched, etc. can be grounds for a copyright violation.

The difference here is that the images aren't stored, but rather an extremely abstract description of the image was used to very slightly adjust a network of millions of nodes in a tiny direction. No semblance of the original image even remotely exists in the model.

This is very much a 'color of your bits' topic, but I'm not sure why the internal representation matters. It's pretty trivial to recreate famous works like the Mona Lisa or Starry Night or Monet's Water Lily Pond. Obviously some representation of the originals exist inside the model+prompt. Why wouldn't that apply to other images in the training sets?

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#122

Post by hutzlibu »

chii wrote: post #102
riversflow wrote: post #76

Earlier quoted context omitted.

> Are image generators giving exact (or very similar) copies of existing works? um, yes.[1][2] What else would they be trained on? According to the model card: [1] https://github.com/CompVis/stable-diffusion/blob/main/Stable... it was trained on this data set(which has hyperlinks to images, so feel free to peruse): [2] https://huggingface.co/datasets/laion/laion2B-en

> What else would they be trained on? why does it matter how it was trained? The question is, does the generative AI _output_ copyrighted images? Training is not a right that the copyright holder owns exclusively. Reproducing the works _is_, but if the AI only reproduces a style, but not a copy, then it isn't breaking any copyright.

Yes, because real artists are also allowed to learn from other paintings. No problem there, unless they recreate the exact work of others.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#123

Post by LudwigNagasena »

myrryr wrote: post #7

I think they will ask them to use the tool to reproduce a known piece of art - which they will most certainly fail to do. I think this has an almost 0 chance of success.

I just took the first few sentences from Wiki that describe Mona Lisa [1], pasted it on HuggingFace [2] and got pretty similar results [3].

[1] The Mona Lisa (/ˌmoʊnə ˈliːsə/ MOH-nə LEE-sə; Italian: Gioconda [dʒoˈkonda] or Monna Lisa [ˈmɔnna ˈliːza]; French: Joconde [ʒɔkɔ̃d]) is a half-length portrait painting by Italian artist Leonardo da Vinci. Considered an archetypal masterpiece of the Italian Renaissance,[4][5] it has been described as "the best known, the most visited, the most written about, the most sung about, the most parodied work of art in the world".

[2] https://huggingface.co/spaces/stabilityai/stable-diffusion

[3] https://imgur.com/a/L2LDOS4

EDIT: With the Starry Night it worked even better. But it failed to reproduce the Bathing of a Red Horse (that one doesn't have a page on English wiki, so I took the description from elsewhere).

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#124

Post by groestl »

dr_dshiv wrote: post #2

“Sta­ble Dif­fu­sion con­tains unau­tho­rized copies of mil­lions—and pos­si­bly bil­lions—of copy­righted images.” That’s going to be hard to argue. Where are the copies? “Hav­ing copied the five bil­lion images—with­out the con­sent of the orig­i­nal artists—Sta­ble Dif­fu­sion relies on a math­e­mat­i­cal process called dif­fu­sion to store com­pressed copies of these train­ing images, which in turn are recom­bine…

> It is, in short, a 21st-cen­tury col­lage tool.

Interesting that they mention collages. IANAL but it was my impression that collages are derivative work if they incorporate many different pieces and only small parts of the original. Their compression argument seems more convincing.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#125

Post by realusername »

manholio wrote: post #84
visarga wrote: post #42

Earlier quoted context omitted.

> 90%ish of a single input image Oh, one image is enough to apply copyright as if it were a patent, to ban a process that makes original works most of the time? The article authors say it works as a "collage tool" trying to minimise the composition and layout of the image as unimportant elements. At the same time forgetting that SD is changing textures as well, so it's a collage minus textures and composition? Is the…

But they are not original works, they are wholly derived works of the training data set. Take that data set away and the algorithm is unable to produce a single original pixel. The fact that the derivation involves millions of works as opposed to a single one is immaterial for the copyright issue.

The training data set is indeed mandatory but that doesn't make the resulting model a derivative in itself. In fact the training is specifically made to remove derivatives.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#126

Post by vldmrs »

I find this lawsuit very important both for AI community as well as for artists since it can let everyone know where the boundaries are.

All mentioned models were trained on a large batch of works taken from Internet - no one is going to argue that.

All of those models cannot exists without content created by people.

The models allow to mimic the styles of particular artists making it a perfect tool to kill creativity and artists diversity in the long term.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#127

Post by TheMiddleMan »

limitedsupply wrote: post #56
TheMiddleMan wrote: post #40

Earlier quoted context omitted.

I believe Copilot was giving exact copies of large parts open source projects, without the license. Are image generators giving exact (or very similar) copies of existing works? I feel like this is the main distinction.

These models produce a lot of “in the style of” content, which is different from an exact copy. Is that different enough? I guess that’s what this lawsuit is going to be about.

Yeah what's considered a copy or not is a grey area. Here's a good example of that: https://news.ycombinator.com/item?id=34378300

But artists have been making "in the style of" works for probably millennia. Fan art is a common example.

I suppose the advent of software that makes it easy to make "in the style of" works will force us to get much more clear on what is and isn't a copy. How exciting.

However, I don't see how the software tool is directly at fault, just the person using it.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#128

Post by jimnotgym »

limitedsupply wrote: post #70
jimnotgym wrote: post #58

Earlier quoted context omitted.

The owner of that Banksy painting certainly thinks so.

The painting that has several cuts in about 25% of the surface area? Don’t think that constitutes as a shapeless pile of dust.

So what % does?

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#129

Post by visarga »

8n4vidtmkvmk wrote: post #77
TheMiddleMan wrote: post #31

An artist can look at images for reference, and draw something new inspired by them. Why does it matter if a software tool can do this much faster? If the artist makes the image very similar to one of the reference photos, it may be a copyright violation. It doesn't matter if the artist used a pencil or software to create the new work. Current AI image generation does, however, make it easy to unknowingly violate cop…

I don't know if you're right or wrong, but it seems plausible that we could create a database of copyrighted images to check against.

Only the training set should suffice.

Re: We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion

#130

Post by 8n4vidtmkvmk »

realusername wrote: post #19
fruit2020 wrote: post #16

Earlier quoted context omitted.

How does it work then? :)

The opposite way, the training images are there to support the model to generalize features. Reproducing parts of existing images in the dataset is called overfitting and is considered a failure of the model.

how do you measure success?

i wrote an OCR program in college. we split the data set in half. you train it on one half then test it against the other half.

you can train stable diffusion on half the images, but then what? you use the image descriptions of the other half and measure how similar they are? in essence, attempting to reproduce exact replicas. but i guess even then it wouldn't be copyright if those images weren't used in the model. more like me describing something vividly to you and asking you to paint it and then getting angry at you because its too accurate

Post reply on HN