Jump to content

AI being used in SL product ads-your thoughts?


Recommended Posts

41 minutes ago, Stephanie Misfit said:

I'll just leave this here, from an actual ad for an SL shape.

 

Screenshot 2024-06-19 180656.png

Every time I see a smile enhanced avatar, I always see this movie poster in my head haha

image0.jpeg

  • Haha 1
Link to comment
Share on other sites

See, I'll be the first to say it: I'm a dum dum. So chances are I don't get a lot of things. Which perhaps gives me a bit of a special perspective. Question: How is this different from vendors using straight up renders from blender or pushing product pages through hours of post editing?

What's sold isn't what is shown on the ad. That was already true when I joined Second Life years ago by now. The product images back then showed an impossible quality that I chased before I realised they were just renders.

Ads should reflect the reality of the product but obviously don't. Set me on fire if you must but I don't hate AI - I dislike when something is passed off as something it is not. Which would also apply to renders and meticulous post edits that pose as "RAW straight outta SL, just tweak your windlights dummy, teehee!". Frankly, this also applies to photography in SL which increasingly sees people use AI while claiming it's all skill. The dishonesty annoys me, not the use of which.

5 hours ago, Katherine Heartsong said:

I'm curious if SL could ever look this good?

In theory? Maybe? Second Life is weird in that it has gone in a different direction from the industry. We do a lot of things by brute force that the games industry tricks and cheats. We pump more polygons into a tiny jewelry than entire hero characters in commercial state of the art videogames do for an entire scene. There are individual 1024x1024 textures on shoe soles no one bloody sees to begin with.

Videogames, they cheat. A lot of what they do is to cut details where no one would realistically expect them. This frees up GPU/CPU power to add additional tricks such as subsurface scattering, fake bounce lights, perhaps even full raytracing. The thing is, we've only just now gotten PBR. That's from over ten years ago.

Case in point, here's "Remember Me", according to wikipedia the first commercial application of partial PBR.

image.thumb.jpeg.f4f0b5d8a0fd39ac556bb729460e0fc3.jpeg

We can probably get *there* if LindenLab gets their act together. Using fixed baked lighting for the scenery and such. We can probably even punch a bit above the weight class of 2013 because we are just utterly insane about how we use textures and render budget. But to get to the level of that skin ad, we'd need a whole lot more from the video game bag of tricks. Not least of all, lighting. That said, I do feel like there is a lot of potential for Second Life by adding more shader support and less fubared alpha *ducks for cover*.

Case in point:

image.thumb.png.4fcb09ed77c9528cd38716f9f5414aed.png

Here you can see the difference between basic SL and what a bunch of cheap shaders do. Both, the basic image as well as the one I pasted over half of the face, came out of Second Life like that, but the latter had the support of reshade with some basic fake raytracing, fake subsurface scattering, clarity, tonemapper, different ambient occlusion and so on. So I'd say there's still room for Second Life to go but... hah. We've seen how the implementation of PBR went.

Thus in practice - nah we won't.

3 hours ago, Scylla Rhiadra said:

And that's quite a bit different from what AI can produce, at least now.

I am sorry to pipe up with an "actually". The public perception of AI has been coloured by the ease of use several image generation websites offer. I think an apt comparison would be photography, not photoshop. Because in essence what these websites do is allow you to make the equivalent of a random snapshot. As if you were to grab a camera, point it somewhere and click. The act of pointing and choosing the motive is the equivalent of the prompt.

However as you know very well yourself, there is a lot more to photography than just randomly pressing a button and hoping for the best. It's a deliberate dance of picking a scene, composition, adjusting the lighting and so on. Well - the same is true for AI because what you get on those image generation websites is the cheap toy camera that obfuscates all of the settings and options for ease of use.

However behind the scenes it can get very elaborate and deliberate very quickly. There are a lot more ways to influence the result than just the prompt. Here's an example of what the Krita AI Diffusion plugin by Acyl can do - and that's just the surface of things.

As for the ethics side of things, I get it. Many years ago, I was looking into doing a doctorate that would specialise on the cross section of culture and machine learning. This was... I want to say more than 10 years ago.

I was doing basic things. Stupid things. Such as creating a basic model learned on census and gender data. I then used that model to analyze amazon book releases with the intent to figure out whether the market skewed towards one or the other gender along the years.

Even back then, I wondered about the ethical side of this. In my case, I quite frankly had limited empathy for the juggernaut Amazon but I could also see the writing on the horizon for a lot of fields. Feeling conflicted about it, I remember writing my would be professor and asking about the ethics of it.

The answer was more or less: sucks, but the cat is out of the bag. People had given up their rights to social media platforms for ages at that point. Heck, I got bitten by that too. I once wrote a webnovel, wanting to stay ahead of publishing trends. I uploaded it to various webnovel hosting sites, maintaining a readership and community until - I got slammed and pummeled by legal, several of them claiming ownership over my work. My mistake. Was stupid, didn't fully understand the terms and paid for it. Or rather, they did because heck as hell I am gonna finish that story anytime soon. Enjoy your open ended *****show!

Sadly. I'm not alone in my idiocy and for decades at this point, people have pumped their creative works onto the internet for platforms to use and make use of as they see fit. I wish there was an ethical way to compensate artists but at the same time, the way the technology works, that's unrealistic. It's the billionth part of a greater whole that might share tensors with a dozen other artists. I get that it sucks for artists. Heck, my writing is stuffed and fed into those LLM too because there sure as well were enough pirated PUBs and PDFs flying around.

All I can do is get with the times, adapt, improve. There's no putting that cat back into the bag.

  • Like 6
Link to comment
Share on other sites

11 hours ago, Rowan Amore said:

I'm like you.  If it's obviously an AI vendor photo, I usually pass it by, especially with skins.  I tried on a few at last years Skin Fair that ended up not looking at all like the photo.  It was extremely deceptive and unnecessary, IMO.  Own your work, don't hide it behind fancy AI.

It's very rare I find a skin that looks like the ad, been happening for years. I know a lot of things factor into what the skin looks like, and you can duplicate those things and still look different. Though I am using an alternative look these days and not regular skins, I find these don't differ as much.

  • Like 1
Link to comment
Share on other sites

7 hours ago, Katherine Heartsong said:

I'm curious if SL could ever look this good?

Absolutely.

There are a few current examples of game engines that can look this good. That said given the outrage that the PBR update has caused amongst many maybe the technology is only half the battle...

 

Edited by AmeliaJ08
  • Like 2
Link to comment
Share on other sites

5 hours ago, Scylla Rhiadra said:

But access to the art will be increasingly restricted, and our culture's understanding of, and taste for art will coarsen.

I agree with some of what you said (as a practicing artist of 40+ years) and disagree with a bit, but this quoted part stood out.

Access and understanding of what art is is already restricted by the narcissistic gatekeepers and art experts who tell the unwashed public what "good" art is. It's an insular world of insider-only BS that already stifles how people think they should view art. Add in capitalism as a driving force that defines "value" and sadly, the world of "art" is already a politicized and ugly mess.

For both artists and those who want to experience art.

  • Like 5
Link to comment
Share on other sites

42 minutes ago, Katherine Heartsong said:

I agree with some of what you said (as a practicing artist of 40+ years) and disagree with a bit, but this quoted part stood out.

Access and understanding of what art is is already restricted by the narcissistic gatekeepers and art experts who tell the unwashed public what "good" art is. It's an insular world of insider-only BS that already stifles how people think they should view art. Add in capitalism as a driving force that defines "value" and sadly, the world of "art" is already a politicized and ugly mess.

For both artists and those who want to experience art.

Yes indeed. One of the reasons I'm a little leery about using the word "art," especially in reference to my own stuff (I generally prefer "images"), is precisely that the term has a lot of cultural baggage.

An analogy is in my own professional field of English lit: the term "literature" has the same kinds of connotations as "art," and tends to be read, especially by those with a more conservative bent, as meaning . . . well, a lot of things I'll get into trouble saying here, but generally "classics approved by the 'best' authorities." Most academics I know take a much more democratizing approach. I teach a lot of texts that would be defined by some as "popular literature," including graphic novels and children's lit. They are every bit as valid as anything on a prescribed list of "classics."

The same is true of the visual arts, I think. When I said "coarsen," I didn't mean that it would turn attention away from "The Old Masters," or infect "taste" with merely popular art (anime, manga, comic art, for instance), but rather that there is the danger that it will flood our screens with stuff that is formulaic, hackneyed and, well, really just not very good.

A simple example: doing a search for works by Edward Hopper a while back on Google, I noticed that a disturbing number of the results fetched up were AI works "in the style" of that artist. And indeed, they were recognizably "like" Hopper's aesthetic in some ways, and even seemed to sort of mimic his characteristic themes -- meaning that they were copying some of his paintings reasonably closely. And they were all singularly awful -- all surface, and zero depth.

Edited by Scylla Rhiadra
Grammmmmmmmar
  • Like 2
Link to comment
Share on other sites

4 hours ago, Stephanie Misfit said:

I'll just leave this here, from an actual ad for an SL shape.

 

Screenshot 2024-06-19 180656.png

Oof. That's such a badly done morph. The skin tone on the face doesn't even match the neck!

  • Like 2
Link to comment
Share on other sites

AI will become part of multiple tools in life, no doubt about that in my mind.
And just like a bike or a train were once seen as inventions from the devil, a lot of people don't know yet what to think of AI or how to use it properly. In the end products with AI integration will be tools in the toolbox in many fields.
Progress can't be stopped. People have to learn to use it wisely.

  • Like 3
Link to comment
Share on other sites

11 hours ago, Rowan Amore said:

Sadly, I didn't find this to be the case with the Skin Fair items I demoed.last year.  While the photos had depth and highlights, the actual skins were nothing like that no matter what lighting I tried.  They were simply NOT the skins depicted in the ad at all.  

A beautiful Ad will make me want to at least try the demo. If I try the demo and the product isn't recognizable as the one in the Ad, then I'm likely to put that creator in my "don't bother" list going forward. 

I see AI as just another tool in the artist's bag of tricks. Use it wisely and it can be a great addition to the artists toolset. Use it poorly and it can be the artist's downfall.

  • Like 3
Link to comment
Share on other sites

4 hours ago, ValKalAstra said:

I am sorry to pipe up with an "actually".

Don't be! Who the hell wants to be right all the time? Certainly not ME! 😏

4 hours ago, ValKalAstra said:

The public perception of AI has been coloured by the ease of use several image generation websites offer. I think an apt comparison would be photography, not photoshop. Because in essence what these websites do is allow you to make the equivalent of a random snapshot. As if you were to grab a camera, point it somewhere and click. The act of pointing and choosing the motive is the equivalent of the prompt.

However as you know very well yourself, there is a lot more to photography than just randomly pressing a button and hoping for the best. It's a deliberate dance of picking a scene, composition, adjusting the lighting and so on. Well - the same is true for AI because what you get on those image generation websites is the cheap toy camera that obfuscates all of the settings and options for ease of use.

This is, I think, a pretty good analogy. For the first half century or more of the history of photography, it was viewed primarily as a "documentary" form, rather than as "art." Then came the Pictorialists, who decided it could be art, and from them we get the true masters of the medium.

I sort of alluded to the idea that AI could produce tools that would be welcomed as legitimately empowering elements of the artist's toolbox. I was actually kind of thinking of a few existing platforms that I've heard about (but never tried myself) that are used now by photographers (including SL ones) as a means of enhancing images. My sense is that most SL people using these are employing them to get closer to what is still "The Holy Grail" of SL photography, shots that look like RL photography -- but I may be wrong. And I DO know a couple of SL artists/photographers who use AI well, and in really interesting ways. 

So yeah, I was being reductive.

/me hangs her head in shame

4 hours ago, ValKalAstra said:

I wish there was an ethical way to compensate artists but at the same time, the way the technology works, that's unrealistic. It's the billionth part of a greater whole that might share tensors with a dozen other artists. I get that it sucks for artists. Heck, my writing is stuffed and fed into those LLM too because there sure as well were enough pirated PUBs and PDFs flying around.

All I can do is get with the times, adapt, improve. There's no putting that cat back into the bag.

I don't know the answer to this either. And of course, there are artists who did not themselves upload their work to the web, but whose pieces appear here anyway.

I do think that at some point there will be guidelines. And that will happen when huge corporate content-creators (Hello Disney!) start worrying that they are being threatened. Those guidelines will inevitably be complicated, somewhat fuzzy, and hard to enforce, just as copyright is today. But if AI does represent a "democratization" of art -- well, you can be sure that the corporate world, with the assistance of their minions in government, will attempt to do something to curtail it. We can't have people producing their own stuff, can we?

Edited by Scylla Rhiadra
Clarity
  • Like 5
Link to comment
Share on other sites

Here an image of a photo that did $4.3 million on an auction at Christie's: "Rhein II" by Andreas Gursky. (it was once the most expensive photo in the world. Still top 5). Made in 1999.

Rhein_II.jpg

It was heavily photo shopped. A person with a dog and even the silhouette of a factory were taken away digitally by the photographer.
Gursky had previously photographed in 1996.[5] Dissatisfied with his earlier image, Gursky "thought about whether I ought perhaps to change my viewpoint ... In the end I decided to digitalise the pictures and leave out the elements that bothered me"

The image itself measures 73 by 143 inches (190 cm × 360 cm), while the frame measures 81 by 151 inches (210 cm × 380 cm).

Would it make any difference if he would have used an AI tool to do the job?



 

Edited by Sid Nagy
A few additions.
  • Like 3
Link to comment
Share on other sites

36 minutes ago, Scylla Rhiadra said:

I do think that at some point there will be guidelines. And that will happen when huge corporate content-creators (Hello Disney!) start worrying that they are being threatened. Those guidelines will inevitably be complicated, somewhat fuzzy, and hard to enforce, just as copyright is today. But if AI does represent a "democratization" of art -- well, you can be sure that the corporate world, with the assistance of their minions in government, will attempt to do something to curtail it. We can't have people producing their own stuff, can we?

Yah - fair point. I'm reminded of a supposedly leaked memo from a researcher at google a while ago. The idea was that Google (and OpenAI) had no valid defence against open source AI models. At the time, there were a lot of innovations coming from the open source vector that steamrolled the closed source variants.

https://www.semianalysis.com/p/google-we-have-no-moat-and-neither

I do find it fascinating how since then all the big AI corporations have begun to lobby extra hard for restrictions that would primarily hit open source models. More or less because the big corporations can just pay their way through the regulation while any open source endeavour is gonna fall apart in funding just to swat off the legalese. Basically, they're building their moat through legislation, pulling the ladder up behind them. They also got a lot of good PR, compared to the sheer ***** flinging contest the open source models got.

Funny how that goes.

In ways though I have seen some fascinating use of AI, for example using it as a dynamic virtualizer during concerts. To bounce that back onto Second Life:

I'm seeing some potential for creative expression using this. Say, a client that does nothing but point a camera towards a certain direction with HUD and UI off, a light weight local stable diffusion model that is fed the image output from the viewer (with hard locked negative prompts for obvious reasons), finally then a local server to loop it back into media on a prim.

That way, you could have people stand in front of the viewer, give their consent and then have a living art installation reflecting back how the AI sees their avatars and the world.

Edited by ValKalAstra
  • Like 1
Link to comment
Share on other sites

24 minutes ago, MadameGloom Ravenwood said:

I joined another grid out of curiosity but never even bothered logging in as I decided I couldn't be arsed. I still get emails form them, they now use AI images, which doesn't give me any inclination to want to try it image.thumb.png.cd0ee50c76af09bd8c251612b02bf546.png

I wouldn't log in either, his hair is so big I'd be too scared of what may be lurking in there!

  • Haha 1
Link to comment
Share on other sites

28 minutes ago, MadameGloom Ravenwood said:

I joined another grid out of curiosity but never even bothered logging in as I decided I couldn't be arsed. I still get emails form them, they now use AI images, which doesn't give me any inclination to want to try it image.thumb.png.cd0ee50c76af09bd8c251612b02bf546.png

Did they fall face first into a pot of Charlotte Tilbury highlighter, or a vat of oil? 

  • Haha 1
Link to comment
Share on other sites

There's a few stores which consistently use renders of products instead of in world images, and I refuse to even take their gifts now because the difference in what the item actually looks like vs the picture is so incredibly vast that it's hardly even the same item. 

Same feeling applies to stores using AI. If the quality or appearance of the item I'm buying is utterly alien to the ad, I'm going to stop shopping at that store. If they use renders or AI but the item still looks good/like the ad? No issue. I won't stop shopping there just because they used renders or AI. Product quality is the final say. 

  • Like 2
Link to comment
Share on other sites

2 hours ago, Sid Nagy said:

Here an image of a photo that did $4.3 million on an auction at Christie's: "Rhein II" by Andreas Gursky. (it was once the most expensive photo in the world. Still top 5). Made in 1999.

Rhein_II.jpg

It was heavily photo shopped. A person with a dog and even the silhouette of a factory were taken away digitally by the photographer.
Gursky had previously photographed in 1996.[5] Dissatisfied with his earlier image, Gursky "thought about whether I ought perhaps to change my viewpoint ... In the end I decided to digitalise the pictures and leave out the elements that bothered me"

The image itself measures 73 by 143 inches (190 cm × 360 cm), while the frame measures 81 by 151 inches (210 cm × 380 cm).

Would it make any difference if he would have used an AI tool to do the job?



 

Yeah but the person is trying to sell you just a photo of landscape, not use that photo as an ad to get you to buy the actual land that had all those things in there that were removed. That is a huge difference.

Link to comment
Share on other sites

I save image editing for trying to be fancy on flickr. Every product image I ever made for a vendor or marketplace came directly from the viewer, usually at ctrl-shift-Y midday just cropped. Enabling shadows is as fancy as it gets. If that makes the listing look bad then that's fine, I'd rather not lie to potential customers, and I want them to see what it will look like in world, since I'm not set up really to give demos.

  • Like 3
Link to comment
Share on other sites

38 minutes ago, Rowan Amore said:

Is that how odd the SIMS looks?

Only if you mod the hell out of it with some really bad and really strange mods trawled up from the deepest Hells on the interweb.

This is custom content, but closer to how Sims 4 looks than that vendor pic.

This sim was a 1.2 GB download, complete with custom skin, hair, all her outfits, and accessories.

04-22-24_6-22-53AM.thumb.png.b273e8fff796c706b77a5d591b4ab92e.png

See, nothing like those deformed plastic whatever the hell those are supposed to be.

 

 

  • Like 1
Link to comment
Share on other sites

1 hour ago, SabrinaCooke said:

Did they fall face first into a pot of Charlotte Tilbury highlighter, or a vat of oil? 

Mould Release coating, as both their faces were moulded in the same jig, for the cheekbones and jawline, before being custom detailed with cheap power tools, like adding his chin cleft with some kind of dremel tool.

 

  • Haha 1
Link to comment
Share on other sites

20 minutes ago, Zalificent Corvinus said:

No, it would still be over rated rubbish.

AI is just at its beginning and it will not be one mega program for all solutions. What Google and Microsoft (amongst others) do at the moment is just showcasing.  AI will be at it's best when designed for specific tasks IMHO.

 

  • Like 1
Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...