johnnypie said: As a producer, I have vowed to never use AI for my productions and will always hire models. I always want to support them. But that could also mean that I'm vowing to keep my "Fotomat business running" upon the advent of digital cameras. Over time, AI video creators making the same fetish videos at a fraction of my costs, could put me out of business.
There are some big assumptions here:
that AI video will inevitably get good enough to produce a ten minute long scene with good continuity. Will this even be possible? If so, what's your timescale for this? 6 months? A year? 5 years?
that AI video will be cheaper to produce than real video. I know that people can pay OpenAI a few bucks and make a few 8 second long clips, but to do AI even now as a hobby is expensive if you want to own your own gear. A GPU that can do wan2.2 decently costs about $8k. The GPUs to run the newer wan2.5 would cost about $60k. And hiring GPU time is cheaper but by no means free. When I briefly sold a few AI scenes, I sold them at a net loss. Why do you assume this will get cheaper?
that the AI video in the future will be better than what can be produced with real people. Why do you assume this? What could AI do that could be better than a video featuring a real person?
that the two can't co-exist. Why wouldn't the same people buy both AI and real scenes if both of them appealed? Why would it have to be one or the other?
johnnypie said: As a producer, I have vowed to never use AI for my productions and will always hire models. I always want to support them. But that could also mean that I'm vowing to keep my "Fotomat business running" upon the advent of digital cameras. Over time, AI video creators making the same fetish videos at a fraction of my costs, could put me out of business.
There are some big assumptions here:
that AI video will inevitably get good enough to produce a ten minute long scene with good continuity. Will this even be possible? If so, what's your timescale for this? 6 months? A year? 5 years?
that AI video will be cheaper to produce than real video. I know that people can pay OpenAI a few bucks and make a few 8 second long clips, but to do AI even now as a hobby is expensive if you want to own your own gear. A GPU that can do wan2.2 decently costs about $8k. The GPUs to run the newer wan2.5 would cost about $60k. And hiring GPU time is cheaper but by no means free. When I briefly sold a few AI scenes, I sold them at a net loss. Why do you assume this will get cheaper?
that the AI video in the future will be better than what can be produced with real people. Why do you assume this? What could AI do that could be better than a video featuring a real person?
that the two can't co-exist. Why wouldn't the same people buy both AI and real scenes if both of them appealed? Why would it have to be one or the other?
With everything tech, wide adoption drives the prices way, way down. Look the the cost of the first personal computers...first digital cameras...first large screen TV's. They were thousands and thousands of dollars. What's the cost now? Exactly.
With everything tech, wide adoption drives the prices way, way down. Look the the cost of the first personal computers...first digital cameras...first large screen TV's. They were thousands and thousands of dollars. What's the cost now? Exactly.
My time-frame - 5 years sounds about right.
What about the other stuff?
I don't think AI will put you out of business for one simple reason: people hate AI. There are lots of people who need it to have really happened, they need video or photos of real people. You will always be able to sell to them.
But anyway, what do you think about non-realistic AI cartoons?
Love you, too
that AI video will inevitably get good enough to produce a ten minute long scene with good continuity. Will this even be possible? If so, what's your timescale for this? 6 months? A year? 5 years? 
