Predictions about the impact of AI-generated imagery

There is a line from Carnival Row, "Chaos creates opportunity."

If you've been riding a gravy train your skill set may have gone stale. It's not just AI. Consider drones. Overnight high paid filmography from helicopters died. Yet an opportunity was created for almost every wedding that has or wants one.
 
And drone video services should follow the same cycle - start at $500/hr, go down to $100/hr within a decade, becoming a wedding side dish.

Then being replaced by the AI fake droning. With Mary flying like the Superwoman by herself, having been filmed against a green screen in a studio earlier.
 
I see this as a fascinating topic. So much of media is consumed without attribution of the creator and without desire to attribute to the creator. I came across a stunning photo listed on Facebook of two dancers, one seemingly floating on the air with nice lighting and nice backdrop etc.... I am sure this shoot took a while to plan and execute. Yet the viewing medium was Facebook. Was any money made off of this endeavour? No, of course not. The photographer probably was paid to take the photo but for the most part, it is just another image out there. So is it really a stretch to think if that photo was created by AI nobody would care?

It is an image. The nuts and bolts of how an image is created is past its due date. Photographers used to have the respect of the general public as craftspeople. That has gone away to a large extent imho. I think this will be a tidal wave of middle and lower level advertising and general imagery. Why? Because nobody cares to begin with. Us one man banders' want to focus so much on quality yet the general public does not care beyond price for the most part. These AI images will arrive without copyright issues and be a low cost alternative for many to have 'something' to put out there to sell their products or represent their business.

Yes, it would be great for us to head it off at the pass and try to make money off it but how? Do you know how to make these images? If you did, what would the margin be since this is probably going to be a lower cost alternative?

I think of movie actors. Once they get the AI stuff to really have personality does it really matter in the end if it is a human? Story is story. Many people dislike actors anyway. Most movies are loaded with CG already. It is not much of a stretch to have this carry over to the leads.

They will get better and there is no stopping it.
 
Lots of things I want to respond to in this thread, but for starters I'll share John Brawley's response to me about this on facebook -

Me:
Hey John - do you have predictions for the future of AI-generated imagery in filmmaking? It seems to me like a natural fit into CGI workflows and probably also stock photo/video. The still lifes and landscapes I've seen have been pretty impressive. But less so were the results of adjusting humans in a shot. What do you think?
Click to expand...

John:
I have no doubt it will happen quickly. I suspect it will be something that is for the masses and more prestigious content will still be real life. Rather than replace it will take over the heavy lifting of a lot of corporate / event work. It will be like stock footage or royalty free music. It will have its place but won’t replace doing it for real when the money and perceived value of authentic is there
Click to expand...
 
Bassman2003 said:
... These AI images will arrive without copyright issues ....

I think of movie actors. Once they get the AI stuff to really have personality does it really matter in the end if it is a human? Story is story. Many people dislike actors anyway. Most movies are loaded with CG already. It is not much of a stretch to have this carry over to the leads.

They will get better and there is no stopping it.
Click to expand...

Major features have to pay for their use of the CGI to the estates of the replicated actors. I would reckon that the Presley estate has the first dibs on any AI produced image of Elvis. There may be some debate about the "personal" use of someone's likeness sort of like a home tape copy of a song but any commercial rights will stay with the copyright holders.

Trickier will be the voice impressionists and deep fakes - this dude may sounds like Patrick Stewart but it's really Bob from Pacoima. And then Stewart might argue that clearly it's designed to sound like him but Bob might say, "Nah, it's a traditional British accent. Go away".

As to what JB said, I think a lot of people will want to see a young (ages 30-45) Harrison Ford again. And if you can create an AI/CGI Ford for a low sum, you can pay the real actor for his image while letting him relax by the beach in Wyoming.
 
NorBro said:
Anyone else miss the 90s? lol
Click to expand...

My wife and I have had 90s nostalgia for a year or 2. We say to each other, "we didn't know how good we had it back then." Mostly because it was pre-social media, pre-napster. Human connections were more personal. The many forms of internet piracy had not yet wrought havoc to the creative world. Seeing the way AI can rip off the style of digital artists, it seems like one more form of piracy. But copyright law could be changed to recognize the contribution of the images that trained the AI
 
ahalpert said:
My wife and I have had 90s nostalgia for a year or 2. We say to each other, "we didn't know how good we had it back then." Mostly because it was pre-social media, pre-napster. Human connections were more personal. The many forms of internet piracy had not yet wrought havoc to the creative world. Seeing the way AI can rip off the style of digital artists, it seems like one more form of piracy. But copyright law could be changed to recognize the contribution of the images that trained the AI
Click to expand...

Napster!! Oh to be young with a healthy download queue again! Oh.. sorry, misread the room.
 
Yeah, growing up as a kid/young teen and living through Napster was nuts, along with many other major technical changes/transitions during that decade or two as the world drastically changed (AOL, video game systems, CDs, iPods, beepers, more advanced cell phones, YouTube a bit later...so much).

And I had a 90s meltdown around 2018/2019 and bought so much vintage stuff on eBay. lol
 
DLD said:
The reply to the DSLR --

https://www.dxomark.com/xiaomi-12s-ultra-camera-test/
Click to expand...

Cons
  • Some noticeable exposure and HDR rendering instabilities in various scenes
  • Lack of zero shutter lag for HDR conditions or low light, with a systematic delay measured at higher than 0.3 seconds
  • Focus and exposure instabilities when using the tele camera
  • Visible noise in high-contrast scenes even in good lighting conditions
  • Occasional low contrast and halo effects in backlit scenes, especially with deep or olive skin tones
  • No HDR on the preview image, which does not match the captured image
Click to expand...
this is the thing about my galaxy s22+ : portrait mode, for example, is very nice. But fails often enough that I wouldn't rely on it professionally

But please, this is an AI-generated imagery thread. Smartphones are more like AI-assisted capture
 
ahalpert said:
...

But please, this is an AI-generated imagery thread. Smartphones are more like AI-assisted capture
Click to expand...

All the stuff you see advertised as "artificial intelligence" by the aftermarket/third party vendors like Topaz is more likely to end up first - in fact, most of it is already present in some form - in smartphones. Your A7SIII has an "Clear Image" digital zoom but smartphones have been offering them for a long time. Samsung had it up to 100X ... which was crap ... but it worked reasonably well up to 30X. Likewise, a digital "portrait blur" to simulate large sensor shot wide open for a shallow depth-of-field has also been around a while. And there will be more AI enabled features in future models, obviously getting better with each release.
 
DLD said:
Major features have to pay for their use of the CGI to the estates of the replicated actors. I would reckon that the Presley estate has the first dibs on any AI produced image of Elvis. There may be some debate about the "personal" use of someone's likeness sort of like a home tape copy of a song but any commercial rights will stay with the copyright holders.

Trickier will be the voice impressionists and deep fakes - this dude may sounds like Patrick Stewart but it's really Bob from Pacoima. And then Stewart might argue that clearly it's designed to sound like him but Bob might say, "Nah, it's a traditional British accent. Go away".

As to what JB said, I think a lot of people will want to see a young (ages 30-45) Harrison Ford again. And if you can create an AI/CGI Ford for a low sum, you can pay the real actor for his image while letting him relax by the beach in Wyoming.
Click to expand...

True but I was speaking more about new AI actor leads, not representations of existing actors. Maybe stars need to be real people to become stars but with the younger generations, who knows? But imagine if major release movies were made with true to life computer creations. The industry would be turned upside down. No more sets, cameras crew etc...Would it be accepted by movie going audiences? I do not know but animated movies are pretty popular. (although they tie famous voices to the characters). Would budgets go down immensely? Interesting stuff. A little ways off but given how real it replaces/affects photography, moving images are not far behind.
 
My hunch is that the creators would "assemble" stars from many parts - male model face and bodies with professional voice actors providing the audio. In other words, it wouldn't be much different from, say, animated films but with less familiar voices because it's one thing to voice Buzz Lightyear, it's another to be Jim John Petrov. Overall, this is going to continue a CGI trend of the animated films becoming far less expensive per given quality. If some ran over $200M mark, then some now can be made for $50M. Scouring the internet, the current high end pricing is at about $1M per minute of animation - before bringing in the A-listers for voiceover - but there are efforts to bring that down to $200K. Non-US/Hollywood production could be there already.

PS. To render a high quality 3D still image may cost $5,000 in NYC and $500 in Sri Lanka.
 
"Model Parts" have taken over the still advertising world. Photo shoots are becoming less common with existing models selling digital versions of themselves. The ad agencies just put the parts together with the clothing images and there is your ad. Impersonal but it is an internet clothing ad, how much meaning does one need to draw from it :)

Movies have become quite fantastical in nature, at least many of the popular blockbusters. They are not too far away from being 100% contrived. Avatar is coming back around almost as if the industry realized it was too early the last time out. So try it again.
 
You must log in or register to reply here.
Back
Top