Levi Strauss uses AI models now

[PSS]

We're doomed. So now I can make an ID with fake pictures. ID photos are usually made by an uninterested employee with a fixed, standard set-up (think drivers license or passport photos). The majority of those who generally employ photographers to make head shots are actors and models (although AI might soon eliminate those jobs) and the casting director or photographer will be bound to be upset and disappointed when the model/actor shows up and doesn't look quite like the perfect AI image that was submitted for the job.

all anyone needs from you is any photo, a voice recording (phone conversation, YouTube video,....) to produce a pretty convincing looking (especially if the audience does not know you) clip of you talking about anything THEY want you to talk about.
so headshots are easy. and most people working with talent these days checks their social media first, plenty of over retouched (and AI improved) images there, but it hopefully gives some indication of what these people actually do look like.
the trend toward analog photography has been there for a while, all this might bring back a need for less retouched/plastic look. but for less then perfect not so young anymore people who hate to have their picture taken and a lot of e commerce, AI is the future.

 

[PSS]

LS&Co. Partners with Lalaland.ai - Levi Strauss & Co

The partnership with the Amsterdam-based company will enable us to experiment with AI-generated models across our products.

www.levistrauss.com

Levi is paddling back a bit based on the public backlash. The reality is though that this technology, given how convincing it is, will advance in the fashion industry.

Imagine a DTC brand that is bootstrapped - they would ofc save model / photographer costs...

Lines will be blurred and unclear what is real and what not ...

I wonder how many actual fashion designs already live in 3D anyway, so it makes sense to pay the models (Kylie Jenner,...) a fee to use her AI/3D face and body (she will never age and never has to do more enhancements) and the AD can finally realize that shoot in the desert of Kylie riding the polar bear wearing the ...... design. It would probably take me a few minutes in mid journey now but I of course cant legally use Kylie's face and I of course don't have whatever she is wearing but that is actually the easy part.
the car industry has been doing this for years. people shoot locations and details that are mapped in.

 

[Paul Spinnler]

I wonder how many actual fashion designs already live in 3D anyway, so it makes sense to pay the models (Kylie Jenner,...) a fee to use her AI/3D face and body (she will never age and never has to do more enhancements) and the AD can finally realize that shoot in the desert of Kylie riding the polar bear wearing the ...... design. It would probably take me a few minutes in mid journey now but I of course cant legally use Kylie's face and I of course don't have whatever she is wearing but that is actually the easy part.
the car industry has been doing this for years. people shoot locations and details that are mapped in.

Fully agree and your experience is best proof of this. This is just another big change for high end mfd as the car and fashion industries traditionally have been key customer segments for P1 and tech cam manufacturers.

 

[richardman]

The ignorance of reality, the naivety, combined with the absolute conviction of some people are very amusing to read

 

[Pieter 12]

I wonder how many actual fashion designs already live in 3D anyway, so it makes sense to pay the models (Kylie Jenner,...) a fee to use her AI/3D face and body (she will never age and never has to do more enhancements) and the AD can finally realize that shoot in the desert of Kylie riding the polar bear wearing the ...... design. It would probably take me a few minutes in mid journey now but I of course cant legally use Kylie's face and I of course don't have whatever she is wearing but that is actually the easy part.
the car industry has been doing this for years. people shoot locations and details that are mapped in.

Not sure the fashion industry is that computerized that they have 3D models of their designs. Most of the work is still manual today, witness the sweatshops around the world.

 

[PeterA]

 

[richardman]

Another week or so, and they will fix the hands.

 

[Paul Spinnler]

View attachment 202132

This is insane, but effectively it is a combination of thousands of similar stock images and in that way it is a sign of the times that we live in a world where everyone effectively has a camera in their pockets and nothing or almost nothing has not yet been done before and people thinking they are creating some kind of rare art most of the time are just walking on shoulders of others, sometimes giants and are merely remixing what exists already.

For me landscape photography is the prime example of this. You can go to the Yosemite and whip out your P1 and make a shot of El Capitan or some trees when Spring comes and think you are a great photographer but effectively you are just copying what thousands have done before and catering to your vanity.

If now tools like Firefly allow you to skip the whole part of flying there and doing it then it is just a reflection of precisely how boring everything has become in a world where everyone is a photographer.

Interesting technology.

 

[pegelli]

Despite all the black stories at the time photography didn't "kill" painting.
Similar to that it is my belief that AI will not kill good photography.
And for sure AI will not kill my pleasure of going somewhere, inhale the atmosphere, trying to find good light and make a photo that reflects my mood of being there.
That photo might be similar to what others (or an AI bot) have produced at the same location but it won't be my photo that reflects what I saw and felt there.

 

[Paul Spinnler]

Despite all the black stories at the time photography didn't "kill" painting.
Similar to that it is my belief that AI will not kill good photography.
And for sure AI will not kill my pleasure of going somewhere, inhale the atmosphere, trying to find good light and make a photo that reflects my mood of being there.
That photo might be similar to what others (or an AI bot) have produced at the same location but it won't be my photo that reflects what I saw and felt there.

The point was more about the business impact for people making a living in areas susceptible of being replaced. I agree it will not easily replace good fine art or have an impact on people shooting pictures for their own pleasure.

That stock photography pic above is uncannily good, though. If this is one prompt away ... why get stock photos or commission simple picture content by a photographer ...

 

[pegelli]

I agree Paul, some people will lose their jobs as technology develops, but it will provide opportunities for others.
For example after photography there was less work for commissioned portrait painters, when steam trains were replaced by diesel or electric traction firemen lost their jobs, when automation hit people were replaced by machines etc. etc. A lot of professional photography is already replaced by smartphone snaps and now AI is threatening it even more. However moaning, groaning and whining about it won't stop it getting a foothold, that has never happened and I believe will not happen now either. Desparately hanging on and trying to stop technological development because one feels it's "unethical" has never worked and will not work in this case either.

 

[Ray Harrison]

AI is possibly going to drive the cost of delivering repetitive, cookie cutter, high volume content of any sort to near zero across a lot of industries. People will often choose low cost at the expense of higher quality but more expensive items. They'll choose sitting on their couch rather than initiative and effort. I know it can be true with me . This is certainly true in software development where I spend my time. In the 90s I could get $500+/hr doing certain types of development that can be had for $30/hr today. Software development has traditionally been a relatively safe white collar job. Someone coming into the field today will need to broaden their tool kit because it's definitely a new world. If any of us do the sort of work that AI can replace, we're going to need to figure out the best way to learn about and work with new paradigms and tools to enhance our ability to do whatever it is we do. We're going to need to look for opportunities - and there will be plenty - and to problem solve and adapt. I've had to do that many times in my career and will continue to need to do so. All of this is anxiety invoking for sure, no two ways about it, but humans are pretty creative, I find.

 

[Paul Spinnler]

In my view the real economic impact of AI will be that property rights become more important again, ie real estate, enforceable patents of inventions (not art), etc.

If AI allows people to do knowledge work or creative work at near 0 marginal cost it will come down to who owns shares in profitable businesses and who owns real estate where people live on. In large corporates, it means bosses who sit at a certain seniority level and are good politicians will prevail because there it is all about protecting your position in a functioning economic enterprise.

Ie it is good for capitalists, and corporate captains with high political skills and bad for people trying to move up the ladder and generally average people.

By extension for all normal creative employees and self-employed which do not adapt quickly or are already at the top of their game.

 

[Pieter 12]

AI image generation relies on prompts. In the near future, those with the talent to accurately describe a scene, along with lighting and emotional cues will be the ones with jobs.

 

[Paul Spinnler]

AI image generation relies on prompts. In the near future, those with the talent to accurately describe a scene, along with lighting and emotional cues will be the ones with jobs.

“Prompt know how” will be so readily available via online tutorials and by virtue of the fact that you can just share the one or two good sentences needed incl some workflow tips, that it hardly will be a job …

Firefly already lets you piece together the underlying prompt in a modular manner …

You can easily learn already how to “prompt” any of the common models to get a great result - it is all on YouTube and not eaxactly rocket science. AI is a Great Equalizer by removing big part of the skill component

 

[Pieter 12]

Example, please. It depends if there is a specific image and look one is after or more like a search of a bunch of stock photos which can be tedious and a bit of a crap shoot if you are vague with the keywords used in the search.

 

[Paul Spinnler]

You can already now steer AI via control imagery - see stable diffusion’s control.net; this will evolve significantly and at one point it will be very easy to get the look you are after by combining sample images with prompt presets.

That’s the whole point about AI that it makes it very easy for everyone to create what they want. The skill component is vastly reduced when presets, prompts and sample input reference data gets you to great end results in seconds.

 

[Ray Harrison]

It’s about finding a place in any industry where skill hasn’t been completely automated away, if possible, and to be able to pivot when automation comes calling, which it inevitably will. If our skills get automated away, it can also be about finding the market - and there will be one generally - where people who do it in a less automated way are still valued. It’s going to be a smaller market, for sure, but it will often be there. Or be the expert of the tools that are doing the automating. I think, for me, the most valuable thing I do is add more tools to my tool set to meet my clients with what they value. Sometimes they value cheap-n-fast, sometimes its other things.

On the photography side, it’s interesting in so many ways (to me) that the proliferation of our images online over the last few decades has been poured into training image-based AI tools, both the ones we’re talking about here but all of the ones showing up in our image processing tools. We may have talked about this here on this thread, I don’t remember, but there are some sites that are now specifically preventing their hosted images being used in these AI training. Obviously, too late and difficult to enforce, but it’s still interesting how people are reacting, nonetheless.

 

[PSS]

AI image generation relies on prompts. In the near future, those with the talent to accurately describe a scene, along with lighting and emotional cues will be the ones with jobs.

a lot of people playing with AI image creation right now use chat GPT for prompts.

 

[Pieter 12]

What will be scary is when AI useless your device's camera to capture your image and then puts it in the situation being illustrated, for example, you visit a restaurant site and there you are, enjoying a meal. Or wearing the clothes, driving the car or whatever. It will come to that.