Vintage photo's copyrights in this new age of AI.

[slave1]

Warning to the mods, this post is meant to explore and debate what I believe to be a fundamental issue. It's all meant to be taken at face value, and when I use a "?", I am really asking a question, and not implying anything other than that.

Thanks to @effCup, and their magnificent thread on photographers and their aliases, it is clear that there wasn't that many photographers who created the bulk of the material found here, and elsewhere. Most of them have likely passed on and the thread makes it clear that even around the time of production, the owner of the copyrights for a particular set cannot be ascertained at 100%, and looking for it today is likely an impossible task to achieve.

Fictional case study

For the sake of this discussion, I will use Warren Tang as an example. His website is still up and running here, and although the copyright is set in 2021, it looks like it was created a long time ago and no one seems to have updated or upgraded it. To wit, the copyright notice points to an organization called APIC, whose website can only be found using the Wayback Machine and was taken down decades ago.

Perhaps Warren is still alive, but according to his bio on the defunct Pixotna website, he arrived in New York City as a “young man” in 1962, which puts him near or above 70 years old. According to his website, he seems to have retained the rights to some of his work, but if he has indeed passed, then who holds the rights to his work? His Descendants, if he had any? An Estate? (D/E, onward) Or perhaps no one…

------

When you create an account on Instagram or Twitter, you agree to the following: "By posting content, users grant Instagram/Twitter a non-exclusive, fully paid, and royalty-free, transferable, sub-licensable, worldwide license to use, modify, display, reproduce, and distribute the content."

The German non-profit LAION was given permission by the biggest social media platforms there are to scrape their platforms, and its latest scraping yielded 5 billion image/caption pairs. That scrape was fed into Stable Diffusion and Google's Imagen to train their Ai-generator model. The LAION is only one of a great many entities that scraped the web, which has been fed into a growing list of learning models, and no doubt, more scraping is on the way.

Even if Warren's photos were somehow directly scraped from his website by LAION, how will he know? Given social media's terms, if you have a fan account dedicated to, say, Alyssa Alps, and you posted one of Warren's shots of her, then that photo, not the one from Warren’s website may have been snagged by any scraper.

Now, today, anyone can be using an Ai generator and generate a photo-realistic picture using only prompts. In fact, there is no need for human interaction anymore as bots can now be programmed to generate images while you sleep.

Let’s consider the two following scenarios:

1.
Suppose someone, as a hobby, loves to generate portraits using an Ai Generator, and although profit may not be his main drive, one can purchase a high-resolution copy of his generated material by "buying him a coffee". One morning, this person looks at the portraits that were generated by his bots during the night, and one of them is a photo-realistic duplicate of one of Warren’s shots of Alyssa Alps. This person doesn’t know who either of them is, thus is unaware of the existence of such a photo, and the copyrights on it that may still apply. This person puts it on his site and sells it for pennies from time to time.

What are the chances that Warren or his D/E will ever find out? It doesn’t look like Warren has any social media presence, and perhaps so does his D/E. Even if they did, how can they be made aware of it if there are no hashtags or names attached to the post?

2.
Now suppose this person works for a film production company. He is so smitten by this portrait that he frames it and put it on his desk at work. One day, the director of one of their upcoming films steps into his office, and upon looking at the frame, is also instantly smitten by Alyssa’s beauty. After learning that this portrait was generated, and not a photo of a real person in this man’s life, he asks him if he could use it in his film, and the guy agrees because, well, why not. Three years later, the film goes on to become an international hit, and it turns out the director used the portrait as a prop that drove the entire narrative of his film, and the photo becomes viral.

Now the chance that Warren or his D/E become aware of it is significantly higher, not to mention how many of us vintage lovers will spot it. Surely, that is grounds for filing a copyright violation, but the production company cannot remove it from the film without incurring significant costs, and frankly, the damage is done, this photo has become part of pop culture and may be recreated or used in memes onward by countless people.

Legal actions could be launched but at what costs, and who is to blame? LAION for scraping the photo? Facebook for letting them do it? Midjourney for creating the means to generate that photo? The user who generated it? The film production company? All of the above? Any of these actions will require a tremendous amount of time, money, and resources, and the damages suffered by Warren or D/E are rather small. It’s not like libel or slander are involved, and it’s not like Warren is a famous National Geographic photographer, and that photo won him awards the world over. If that were the case, he’d probably have enough clout to create a PR nightmare for the parties involved, which would give him a chance to at least mediate the matter with them or settle out of court.

The big problem is that intent cannot be proven, and under these kinds of circumstances such a case becomes extremely complex, and again, that means a vast amount of money, time, and resources will be needed to pursue legal actions. Even worse, if the verdict does not favor Warren, he or his D/E may be open to a countersuit.

I was recently made aware that MetArt has the rights to Viv Thomas's work. If the scenario above implicated them instead of Warren, they certainly seem to have more resources on hand and can pursue legal actions. That said, the complexities of such a case remain, along with the expenses and the possible negative consequences.

Now what?

In the realm of Ai-generation, things are moving extremely fast, and a couple of major lawsuits are already underway, including Getty Images. It turns out learning models don’t need prompts to recreate a person, they just need their names. There is a fantastic Twitter thread on the matter if you are interested in learning more. It appears Twitter links cannot be embedded here, but if you Google the terms "Eric Wallace" and "Stable Diffusion" you will find it right away.

Some artists have joined forces and are also suing but their concern is over art, not photography. The problem here is style cannot be copyrighted. Warren certainly had one, but that can’t be grounds for copyright infringement.

The advent of AI-generating models will certainly have an impact on copyright laws, but the institutions being what they are, it will take them years to sort it out and amend those laws or perhaps create new ones. In the meantime, technology is moving forward at a breakneck pace and the ramifications for photography and art are immense. Perhaps down the road the very definition of “fair use” will drastically change, but which form will it be and when will it take effect is anyone’s guess.

Wealth of potential

I live in a big North American city and I know of two big used bookstores that happen to have a very large vintage erotica section with hundreds of mags and DVDs, though for the latter I have yet to find one produced before 2000. Last November, I paid them both a visit and purchased quite a few great mags, and I found some sets in them that I have never seen posted anywhere online, and I’ve been online since 1999.

I paid them a visit again last week, and I kid you not, it felt like both stores flipped their entire inventory in the meantime. I was stunned. These are just two used bookstores in one city. The amount of vintage stuff that’s out there is likely astronomical.

This post is in the photo section, but I could’ve made the same post in the video section. How many videos, films, and loops have long been discontinued? Just like photography, those who hold the rights to that material may have lost them, they may have passed and didn’t leave the rights to descendants or an estate. If they are still around, they may have left the adult industry altogether a long time ago, or simply just don’t bother at all.

Conclusion

As a photographer myself, of the non-adult variety, I’m glad my photos are protected, but today's tools are now widely available and are improving faster than most people can keep up with. Although not to that extent, it is entirely possible the scenario used above can apply to myself and my work. If it happens there is not much I can do about it. Nor can you, if you ever come across a generated copy of a photo you took during your 2015 summer vacation, which you posted on your social media then.

As vintage lovers, how do we act or proceed under this new paradigm? If tracking down the rights to photographers’ work is hard now, imagine how hard it will be in the coming years as the pool of photographers and/or people who hold the rights to their work is naturally thinning out.

Which mindset is needed now, one that strikes a balance between the will to respect copyrights while acknowledging today's and tomorrow’s reality?

I’m looking forward to seeing where this discussion goes.

S.

[deepsepia]

Originally Posted by slave1

Which mindset is needed now, one that strikes a balance between the will to respect copyrights while acknowledging today's and tomorrow’s reality?

I’m looking forward to seeing where this discussion goes.

S.

As someone who is deeply interested in the subject -- I started a thread devoted to the subject, and have posted a lot of my own material -- I'll say "it's complicated".

Photographers and cinematographers have often been "influenced" by the styles of others . . .does the Kubrick estate "own" his signature slow zoom? Similarly his dolly shot -- first seen in The Shining?

You can't own that.

Nor can you own a hairstyle. Or "the Annie Hall look"

At a practical level, the issue will addressed at licensing level. Adobe already owns a giant stock photography business -- they have rights to generate new content derived from these images, and you as an Adobe user can have assurance that the new Photoshop Generative Fill function was trained on IP to which Adobe has rights.

. . . but those of us who are both photography fan and tech savvy -- there's a lot of stuff we're going to do that's not going to be in the Getty Archives or other stock libraries.

Lots of folks have inquired about AI enhancing of poor quality footage. You've likely seen Topaz, which can do an OK job now with good footage, but can't do anything with bad.

Stable Diffusion, though -- train a model on good quality images of a performer so that you have a model or LORA that "knows" what they _should_ look like; with that in hand, you can recover/invent details that either weren't there in the first place, or were lost in the poor quality imaging

As an example of what "invention" looks like, consider

.

. . . on the left, a sketch by Brenot; on the right an image generated using that sketch as an input by Stable Diffusion . . . its hard to narrow down just who the "creator" of the image on the right is. There's the training on image databases, there's the prompt I wrote, there's my curation -- eg discarding zillions of examples that didn't work, enhancing the ones that did .. . if this were a commercial project, it would be difficult to assign rights. As it isn't commercial and Brenot is long deceased, I don't worry much