Stable Diffusion is a big deal • Hacker News Board

Re: Stable Diffusion is a big deal

#441

Post by michaelt » Wed, Aug 31, 2022, 7:07 PM UTC

abigail95 wrote: post #176

michaelt wrote: post #129

Earlier quoted context omitted.

The ethics partly depends on the analogy you choose. I can write a book about a boy wizard's adventures wizard school and that's legal, but if I call them Harry Potter it isn't. I can create Harry Potter fanart and distribute it online pretty freely - but slap it on a mug and sell it, and that's illegal. I can record an audio description of a painting that's as detailed as I like and it's legal to distribute - but ta…

Harry Potter is a trademark. It protects and distinguishes identity and authority. It is much harder to get a trademark, and is also harder to fairly use one. Different issue. If your fan art is infringing, it was infringing whether or not it was on a mug or on dropbox. Photograph one is not true. For commentary it can be by audio, or printed on a mug or whatever, commentary is transformative. Go out and take a photo…

> Harry Potter is a trademark. It protects and distinguishes identity and authority.

Perhaps I should have said Darth Vader, then - the point is you can copyright a character independent of the copyright on a book's text, and the trademark on the series name, and the fact that broad concepts like "black-clad masked evil overlord" are uncopyrightable. And that copyright can persist even if you transform a book character into an engraved coffee mug.

> I'm not seeing an ethical difference from any of this. Or did I miss the point?

The difference is:

If a person says "Stable diffusion is to its copyrighted training data as an audio description is to a painting" or "Stable diffusion is to its copyrighted training data as the concept of boy wizards is to harry potter" they would probably say it's ethically fine.

If a person says "Stable diffusion is to its copyrighted training data as a photograph of a painting is to the painting" or "Stable diffusion is to its copyrighted training data as video lecture is to a single image in its slides" they might well say it's not ethical.

Re: Stable Diffusion is a big deal

#442

Post by galdosdi » Wed, Aug 31, 2022, 7:07 PM UTC

chucksmash wrote: post #407

galdosdi wrote: post #400

Earlier quoted context omitted.

It's a nitpick, but a little bigger than that. It's as bad as including "bus driver" in the list. Cobblers just shouldn't be included in the category at all. Cobblers are in just as much demand in most of the world as they always have been. They only fell out of demand in car-dependent areas, which is a small minority of the world population (but a vast majority of the HN commenting population since most of the USA o…

> I don't know if it has anything to do with construction but doubt it. If you actually walk everywhere shoes don't last very long these days, especially shoes under $100. By construction, I mean the material and design of shoes people tend to wear. I can't say I've ever met someone who takes sneakers or running shoes to a cobbler and these shoes are more common nowadays.

Those kinds of shoes you mention tend not to last very long at all and are not resoleable. If you walk a lot, you find yourself throwing away the $80 shoe after just 6 months.

These kinds of shoes are most of the market because most people don't walk much in the USA. If you walk a lot, you might still not change anything and keep buying the disposable sneakers, throwing away $160 a year.

But if you walk a lot AND are disposed to think critically about the situation, you find that if you pay a bit more for shoes you can make them last many years as long as you resole them periodically. And as I recall, a good $50 sole on a good $150 shoe costs half as much and lasts three times as long as a disposable $80 sneaker.

Not only do you save money (not really a ton) but it actually is more convenient, since even counting resolings, you get more miles between having to go repair or replace your shoe. And you don't have to wear in your leather uppers again. It is truly a luxurious feeling when you come back from the cobbler and have shoes that are worn in and fit your foot just perfectly like a glove... yet the soles are brand new and strong and comfortable and ready for another thousand miles.

Why do you think there's a stereotype of leather boots being popular in NYC? I'm sure the resoleability and longevity in the face of large amounts of daily walking have a lot to do with it.

Re: Stable Diffusion is a big deal

#443

Post by jannes » Wed, Aug 31, 2022, 7:30 PM UTC

zaptheimpaler wrote: post #395

_gabe_ wrote: post #307

Is it just me, or do the comments in this thread seem to be the exact opposite of the sentiment in the comments on similar Github Copilot threads? I just find it a bit ironic that programmers are irate about Github Copilot using their copyrighted material to train. However, if it's an ML model training off of copyrighted artists material, clearly its a transformative work. I just find the opposing sentiments for thes…

Copilot is paid and closed source. It takes freely from the commons but doesn't give back as freely. There is a big difference.

Your brain is also closed source and you can train it on copyrighted material as well.

Re: Stable Diffusion is a big deal

#444

Post by peoplefromibiza » Wed, Aug 31, 2022, 7:39 PM UTC

thegrimmest wrote: post #426

peoplefromibiza wrote: post #309

Earlier quoted context omitted.

> My brain has been trained on even more copyrighted material That's in fact a problem: plagiarism is considered cheating/intellectual dishonesty and can ruin a career, copyright infringement is punished by civil law (and fined), counterfeit is a crime, etc. etc. You must retain yourself from copying too much, but only the original author (and eventually a judge) can decide if that happened. If stable diffusion is us…

Authoring any work, no matter how closely copied, is not an infringement. The infringement is commercial distribution of copyrighted work. And yes, I do think you should be liable for whatever you distribute commercially, regardless of how you obtain it.

> Authoring any work, no matter how closely copied, is not an infringement

Nitpicking, but only unless it's not counterfeit material.

Claiming painting is from "famous painter X" while it's instead a copy it's an infringement.

How long before "famous painter X's new painting found in the attic of old lady"?

Re: Stable Diffusion is a big deal

#445

Post by ElijahLynn » Wed, Aug 31, 2022, 7:46 PM UTC

burlesona wrote: post #297

dylan604 wrote: post #288

Earlier quoted context omitted.

>Code that's 98% correct is completely useless. Whoa! Really? Seems like modern software releases would be excited to achieve 98% with the incredible amount of bug fixes/patches released very quickly after the massive beta test known as release day.

Code has to be 100% correct or else it’s considered a bug (assuming it’s syntactically valid). Code that is 98% correct is actually much worse than no code at all. That’s the kind of code that will introduce subtle, systemic faults, and years later result in catastrophic failure when the company realizes they’ve been calculating millions of payments without sales tax, or a clever hacker discovers they can get escalat…

If the 98% correct code works, and provides value then it is better than no code at all, yah?

Re: Stable Diffusion is a big deal

#446

Post by stdbrouw » Wed, Aug 31, 2022, 8:01 PM UTC

wruza wrote: post #259

stdbrouw wrote: post #233

Earlier quoted context omitted.

Well, but that's hardly a universal rule either. If we have a magic cure for bad eyesight, that'll be pretty terrible for optometrists, lensmakers and so on, but nobody would think it unethical that we're taking away their jobs.

To me it’s a different situation, because no previous hard, unique, “ownership” work of these people was used without consent to make this cure. Think of this instead: a doctor collects a big volume of symptoms and analyses and creates a statistical way to cure people more easily. They publish many examples of their work without licensing anyone (legally and morally) to use it freely. Now some algorithm collects thei…

But that's what makes this so tough. Whether you consider the ways in which these AI models repurpose existing artistic works to be mere technical details or central to decide on the ethics of the matter depends largely on the analogies you reason by, as michaelt noted.

Re: Stable Diffusion is a big deal

#447

Post by peoplefromibiza » Wed, Aug 31, 2022, 8:08 PM UTC

thegrimmest wrote: post #428

peoplefromibiza wrote: post #339

Earlier quoted context omitted.

> I would venture to guess that the author doesn't have a full understanding of what the AI is doing or how it works. Actually he does. > There are a gazillion artist who take clear inspiration from other artists Inspiration is not the same thing Stable Diffusion does. We can't even define inspiration in a proper manner, but for sure we can say that if someone wants to draw comics in the way Tezuka made them, they ha…

If the model physically does not have enough information to create a work that would be found to be infringing on your friend's rights, then why is your friend upset? The model viewed his work and learned from it in the same way that humans do. If your friend published images on the internet, he clearly intended them to be viewed by (meat) neural networks. Why are silicon ones operated by humans any different? Edit:…

[flagged]

Re: Stable Diffusion is a big deal

#448

Post by thegrimmest » Wed, Aug 31, 2022, 8:32 PM UTC

peoplefromibiza wrote: post #444

thegrimmest wrote: post #426

Earlier quoted context omitted.

Authoring any work, no matter how closely copied, is not an infringement. The infringement is commercial distribution of copyrighted work. And yes, I do think you should be liable for whatever you distribute commercially, regardless of how you obtain it.

> Authoring any work, no matter how closely copied, is not an infringement Nitpicking, but only unless it's not counterfeit material. Claiming painting is from "famous painter X" while it's instead a copy it's an infringement. How long before "famous painter X's new painting found in the attic of old lady"?

But in that case it's not the work that's the problem, but the misrepresentation (read lie) about the provenance.

Re: Stable Diffusion is a big deal

#449

Post by peoplefromibiza » Wed, Aug 31, 2022, 8:41 PM UTC

thegrimmest wrote: post #448

peoplefromibiza wrote: post #444

Earlier quoted context omitted.

> Authoring any work, no matter how closely copied, is not an infringement Nitpicking, but only unless it's not counterfeit material. Claiming painting is from "famous painter X" while it's instead a copy it's an infringement. How long before "famous painter X's new painting found in the attic of old lady"?

But in that case it's not the work that's the problem, but the misrepresentation (read lie) about the provenance.

and the fact that it is a replica

anyway, copying money is a crime, almost anywhere in the World, the simple act of making a copy is enough.

Money usually contain artworks

I would also not generate pictures of child pornography

There's a reason why SD apply censorship filters to generated images

Re: Stable Diffusion is a big deal

#450

Post by thegrimmest » Wed, Aug 31, 2022, 8:57 PM UTC

peoplefromibiza wrote: post #449

thegrimmest wrote: post #448

Earlier quoted context omitted.

But in that case it's not the work that's the problem, but the misrepresentation (read lie) about the provenance.

and the fact that it is a replica anyway, copying money is a crime, almost anywhere in the World, the simple act of making a copy is enough. Money usually contain artworks I would also not generate pictures of child pornography There's a reason why SD apply censorship filters to generated images

Yes, selling an exact copy of a copyrighted work is an infringement (giving it away generally isn't). Creating a derivative work, or work in the same style, is fair use.

Generating a digital image of money isn't the same thing as counterfeiting currency.

> I would also not generate pictures of child pornography

Legality aside, why not? Who is harmed?

> There's a reason why SD apply censorship filters to generated images

I'm not sure there is. I don't think any one group of people is uniquely equipped to limit what images another group of people can generate with ML.