Reading this in the context of Sora, this is contingent on:
1. movie or tv show length generation needs to be reasonably affordable,
2. the generated output must look real beyond first look (text is readable, physics is correct, no dream-like morphing),
3. continuity between scenes
4. prompts can accurately produce the artistic vision from directors and producers faster than a human just doing it
5. the public will accept it
I am reminded of a scene in Westworld, where basically a someone can sit a desk and describe in natural language a scene to a computer for instant rendering. The user can now concern himself exclusively on the details of the story. Perhaps roles will transition to becoming full time story tellers, and the minutia of vfx, animating, rigging, 3d models, actors, cameras, set designers, trucks, food, lighting, electrical, location, licensing, permits, etc. all go away.
I think you're looking at what's needed to replace a lot more than 90% of the jobs.
The comparison was against 500 artists working for 5 years.
So you don't need to get to "write a script and have it render" but get it to the point where a top class animated film can be made with 50 artists working for 5 years.
Also at that scale it can cost an enormous amount and still be totally fine.
It starts with AI generating a short scene. Maybe a transition scene. The public doesn't have to accept it if they don't know it's there.
It's interesting you mention Westworld, the first movie to use CGI. It was extremely primitive, 2D, and only for a few seconds. See now CGI in modern movies.
Even if progress halts at Sora's level (the absolute worst case scenario), I think it's good enough to replace some 1-5% of scenes (meaning 1-5% of animators' jobs are ticking down since last Thursday).
Before I would have said we have plenty more to do before A.I takes over. And a.i ( Assisted Intelligence ) would aid those creative process. So it will be 50% job cut in the near 10-15 years, while the other 50% being basically double as effective if not more.
After Sora I think future is quite bleak for a lot of artist. Sora is like 80% there, I think we would be 90% there in 2 years time with 5-10x more GPGPU power training. And in 6 years time 95% there by 2030. Even if it didn't eliminate 90% of job but 80%. The redundancy would mean the job market is flooded with animators and their salary will likely stagnant if not fall.
At the same time I finally understand why Nvidia is worth so much.
> Sora is like 80% there, I think we would be 90% there in 2 years time with 5-10x more GPGPU power training.
Don’t underestimate the last 20%! I don’t think this is just a matter of more training or more data. But yeah, 2-5 more years will not change the outcome you’re describing!
Speaking as someone who has done VFX professionally, I think Sora is closer about 30% there. It's really not constituted in a way that makes it easy to combine with existing techniques. You either need to come up with a way to make them play nicely together or you have to recreate the entire production pipeline in the model.
It'll get better, but I don't think they've been aiming in the right direction if the goal is to replace current production methods.
Yeah text-prompting only won't replace production work, but it can surely accelerate and boost pre-prod work, maybe doing prelim shots for a storyboard, creative ideas for the costume dept... lots of cool things.
I think for prod work it needs what the auto1111 plugins have given SD - ControlNet etc, where you have a lot more control over the generation and the text prompt is just a small part of it.
Yeah, I don't think people understand how much pre-prod work is done. For the next 5 to 10 years this what it'll be used for. In the next 20 years, 90% are out of work.
The problem is that Hollywood doesn't really care. They live on making compromises during shoots, and if they can save hundreds of thousands as well as dozens of hours and manpower on a scene, they will do it (and they do all the time with horrible CGI, poorly shot scenes, etc...), and they aren't the only industry that will do this.
Right away, the stock footage market is in danger because content creators will opt for a free (good enough) option, over having to not just pay for, and but look for exactly what they want and need.
There's plenty of other industries that will be affected, I can't say that 80% of people will lose their jobs, these things aren't a switch you flip, it's a gradual process, but over time we'll definitely see jobs made obsolete/optimized.
This technology is the worst it'll ever be, and we've seen how far it got in just a single year with Will Smith eating spaghetti and now Sora.
The problem is that Hollywood doesn't really care. They live on making compromises during shoots, and if they can save hundreds of thousands as well as dozens of hours and manpower on a scene, they will do it (and they do all the time with horrible CGI, poorly shot scenes, etc...), and they aren't the only industry that will do this.
Isn't this a problem for Hollywood itself? Isn't the strength of Hollywood the capital to hire good actors and effects people etc?
You seem to be assuming that Hollywood will just fire all the actors and save money, but to me it seems more like Hollywood is in the danger zone too?
The thing is, you run the same prompt twice and you get twice an unique result. Its trivial to keep asking for more renditions of given scene till you are happy with it, even kids can do this and certainly somebody in sweatshop in Vietnam can do that too. You literally need 1 person a bit of computer power. So instead of team of pixar and disney you have 1 guy with the product and some aws prepaid computing.
Now I don't know how long it took openai to put together that 10 minute demo video, if they ran 3x the resulting prompts or 100x more and had to cherry pick hard (ie that bird's head, I was expecting heavy morphing of those feathers but they kept their visual consistency, for sure this was not the first attempt with at-glance-perfect result).
sama sat on twitter during that day and rendered prompts from people on the fly, so those weren't heavily cherry-picked. I didn't watch most of them and they had some bugs for sure, but if they had had to cherry-pick like 1 of 100 for the blog demos he would never have sat and done live renders like that.
> Before I would have said we have plenty more to do before A.I takes over. And a.i ( Assisted Intelligence ) would aid those creative process.
My guess is in the short term it will still be assisted intelligence except that it will be humans assisting AI models as in humans in the loop correcting minor mistakes in the model generated video frames.
If we assume double productivity we could also foresee a future where we get double the amount of high quality movies produced for the same amount of money. Productions would be more viable even on smaller budgets. Could be nice for consumers if the curation keeps up.
If it becomes easier to produce a high end movie, maybe big studios won't be needed at all. The talent can make the movie then find a new means of distribution.
Demand for feature films is fairly inelastic though. Also, the TikTok brained generation can barely pay attention for fifteen minutes, let alone 90 minutes.
Demand for a video of a person reacting to another video was inelastic before youtube.
I don't understand TikTok, but isn't this same argument older people were making about kids playing Mario Brothers back in the 80s, and TV before that?
Either way, there's only so many hours in the day to watch video, and feature-length films are being displaced by 300 TikToks. Will that demand shift back to feature-length because of AI, or will AI rather change TikToks?
> the minutia of vfx, animating, rigging, 3d models, actors, cameras, set designers, trucks, food, lighting, electrical, location, licensing, permits, etc. all go away.
They won't go away, they'll be chosen by default by software. Analogous to the vast majority of powerpoint docs, which are shitty and created by people who don't know anything other than some vague ideas and bullet points they want to put down, but nothing about what it takes to make a quality presentation.
Similarly, some people like making ceramic coffee mugs, even tho the mass produced stuff is super cheap. And some people like buying the former, even tho the mass produced stuff is super cheap.
AI will also eliminate 90% of the audience. I certainly don't see myself queuing up to watch any AI-produced or -enhanced movie. TBH, they lost me much longer time ago, but I assume it will be a larger transition.
> I certainly don't see myself queuing up to watch any AI-produced or -enhanced movie.
When animated hand drawn cartoons (think Disney) were first introduced, film critics said they would never watch it, because didn't have actors in it.
When CGI was introduced, film critics said they would never watch it, because the acting wasn't real.
When 3D rendered movies (think Pixar) were first introduced, film critics said they would never watch it, because it wasn't hand drawn.
Just like cartoons, CGI and 3D rendered movies, AI generated movies has the potential to become a huge market. Maybe not in long format at the beginning, but you bet that short format video platforms such as TikTok and YouTube (shorts) will be flooded with AI generated content soon.
> TBH, they lost me much longer time ago, but I assume it will be a larger transition.
The 'great' thing about platforms like TikTok, is that it'll give a very fast feedback loop on whether the audience is willing to watch it or not. So I expect we'll know whether this transition you speak of will happen or not rather soon. My bet is that it won't, I expect people to watch it anyway.
People already don't care for job-killing products and processes that have way more negative impact than AI-generated movies. For instance, they don't care that they are eating food that is mass produced through a process that eliminated 90% of the jobs previously required for producing it, and which now contains high-fructose sugar, artificial preservatives, etc.
There will be a short-lived, knee jerk reaction against AI-created content. But ultimately, if the content looks good, people will wholeheartedly embrace it.
Are any of these things true at scale? The two I can remember -- CGI and 3D rendered movies, were all lauded by audiences and critics as well as the industry when they emerged. The two films which brought CGI to mainstream audiences were Jurassic Park and Terminator 2, both critical hits. Toy Story was also huge with critics and no one claimed it wasn't 'real' animation.
I think you are just making things up because they make a good argument.
At scale? I don't think so. Most didn't care, I remember 80's magazines questioning CGI and predicting another death of true art (no soul! computers do the job for you!!1), but that's it. Others were fascinated by the novel gimmick. But the same is true for AI-generated imagery - nobody cares in general how the sausage is made, people care about the message and execution. As long as it's good, people like it.
And unlike with Toy Story/Final Fantasy: The Spirits Within/The Phantom Menace, this time there might not even be a novel gimmick/look to sell, as you'll be free to choose any style you like.
GP is actually responding to @throw_pm23, and their number (losing 90% of the audience) seems arbitrary and implausible.
> Maybe not in long format at the beginning, but you bet that short format video platforms such as TikTok and YouTube (shorts) will be flooded with AI generated content soon.
This is already happening to some degree, not usually 100% AI-produced, but AI as a labor saving device or to get some kind of output you couldn’t get before. Presidents playing Yu-Gi-Oh! were practically a genre on YouTube last year, and that’s not all that’s been done with their voices.
Same here. And it's starting to look worse and worse over time. Before CGI was used to enhance a scene. Now no one shoots a scene in camera and adds CGI later (think Transformers / Jurassic Park), the whole scene is just CGI and it looks like crap.
I would wager you mostly dislike bad CGI – that's any CGI you actively notice being CGI – like most of us do. Of course, the disdain varies, but I have yet to meet someone who thought CGI was cool, because it was noticeable.
Even today, there's good and bad uses of CG, usually coming down to schedule and creative direction. The average quality has gone up substantially, though. As you pointed out, there's all kinds of ways to use CGI and VFX more subtly that most people would never realize. https://www.youtube.com/watch?v=clnozSXyF4k is a great example of green screen set replacement being used for things as ordinary as a city street, 15 years ago on a TV series budget.
I expect the use of AI will be similarly varied in the beginning. It could be used just for background details. It could be used in a hybrid fashion where artists design the characters with reference poses, a storyboard, and maybe even some keyframes, and the AI fills in the gaps. And of course some studios will just be sloppy and use AI to generate larger pieces without fixing the glitches and inconsistencies.
People are sometimes dismissive of AI based on the "fire and forget" prompting modes but don't consider AI-assisted tools or human-assisted AI like inpainting.
But that's the thing, I am completely alienated from films already and watch maybe one a year or so when flying long distance. Whether it will be a mass transition, though, I agree that is hard to tell.
Fair enough. I'm not a big film buff either, but for me that's due to the simplifications necessary for film as a medium rather than the acting, plot, or effects. Those simplifications are less present in a series, miniseries, or a multi-part film.
I find art to be a communication (however indirect) between people. I have no interest in it if there is no other person on the other side. It could be that production gets technically so good that I cannot recognize it, but I will check out much earlier when the suspicion will already be there.
CGI jobs are already boring and repetitive industrial "cog in the machine" type jobs though. The grunts doing the actual work don't have any say what their output should look like, there is no "personal touch" or creativity involved. It's all been professionalised to death long ago.
You are right of course, which is partly why I find almost everything produced in the last decade totally uninteresting, except maybe small independent stuff.
I think we'll see these types of films going the way of books and recorded music. We'll have an explosion of films released to hyper-niche audiences (at best). The middle of the market goes away. 10-20 traditional films still get produced and marketed to large audiences. This evolution makes more sense, since to your point — people like you have already stopped watching the product.
I suspect there is a market for adult-focused genre-based animated films that where never economically feasible before.
I feel the same way, but time after time it's shown that 90% of people stick with the heavily enshittified product if it's what they're used to. Most people are passive and docile.
The training data still has to come from somewhere. This just means churning out cookie cutter filmography can even more shortcuts than green screen CGI.
Most films don't reinvent cinematography, but follow established patterns, maybe with a twist. Sure, there are the Roger Deakins of the world, but I'd wager that most cinematographers aren't that.
I find this to be the real strange bit of the "AI art debate". All the management and publishing are very bearish on AI, even though its legality relies on wide-scale copyright infringement. Actual artists are far more skeptical, begging for copyright to become a de-facto AI ban, even though they're way more likely to infringe copyright than to own it. And then you have the Free Software people - i.e. the people opposed to software copyright entirely - who were the first to actually point out that AI training was infringement and trying to sue over it.
more people will be able to create their own animated films without requiring studios, execs and all the gatekeeping and bs that comes with that.
that's why i think the future really will be in content curation.
unless these companies can find a way to block the average person from having access to the same level tools. which i think they will try and they probably can to some extent.
I think everything is getting like this though, music, art, everything? Even trying to find something worth watching on Netflix is already so hard for me I just don't bother. I usually just default to watching classics because I can trust they're worth watching.
Once there is unlimited content, maybe I'll just go outside and watch the stars or something.
Maybe we'll come full circle back to the 80's and 90's movie scene where you had movie stars and directors who people would go to the movies to see, and for each of these you have about 100,000 2 hour AI generated movies with bland content that will have 100 views each. If that's the outcome, it's not as dystopic I think...
I mean already now you have an incredibly large fraction of sh*t on spotify, netflix etc. Will it matter if it goes up another factor of 1000x?
Meanwhile hopefully the really creative and interesting content creators will leverage the AI stuff to further take the art forward. They have much more ways to test out alternatives and scenes before committing to a $250M production shoot for example..
Yep. Maybe there will be a net reduction in jobs but I can't help but feel like catastrophic 90% reductions involve "lump of labor"-type fallacies where we just imagine taking the given set of work being done and imagine how much of that will be handled via AI vs. humans in 5 or 10 years or whatever.
When the reality is that good AI tools will cause lots of projects to add video content where it wasn't before, or to spend more time creating high-quality video content. And this will undoubtedly involve some new labor.
2. Disney can train its own models and make its (oh how I dislike this metaphor) "lobotomy" match the corporate policies the were otherwise going to force on human writers and artists.
There will still be a job for animators, at the very least. They'll flesh out the armature (skeleton) animations, and AI will paint over it, like a style transfer. Imo human-driven animation will stay around for a bit longer because of the nuance of movement and connection to storytelling.
90% of the existing jobs may be eliminated, but that will free up more artists to work in more animation studios. Therefore, we will see a resurgence in animated films.
When the assembly line came into being, we didn't end up losing all of the workers to the factory - we built more factories instead.
Hyped? Yes; Major flaws? Minor, these days; Impossible to fix? No, not even at the pure AI level, though at the consumer level I could believe any percentage impact on how many artists it would take to fix vs. having the artists create in the fist place.
I saw two errors, both of which are trivial to hide.
One of them was only visible after someone else pointed it out to me.
The other was just about in the realm of things I've seen in the pre-AI era where single fixed camera perspectives create odd visuals despite capturing reality.
Forget Animated movies we are fast moving towards a world where acting jobs can be eliminated from movies. I think the latest sora shows that all background acting jobs can soon be eliminated how long after will they come for the main parts.
You are assuming that's what people want to see. We still have a long way to go before any human can be eliminated from the big screen. Issues with details such as hands and eyes are still way to obvious.
Also, hand and face issues are rapidly improving, I can easily believe models like Sora getting used for replacing extras despite the flaws it still has.
It takes hundreds of man years of various to produce even cheap looking b-movie. A-level one have easily tens of thousands of man years overall, just watch the credits on some Marvel movie and put 3-18 months of effort there.
If you want to achieve almost the same for 0.1% of the cost with 1 very talented person and some computing power (or small team of talented writers/artists), it won't take long and Hollywood's business will collapse. You don't need to pay a cent to that old guy in the video.
Current A-listers will slowly fade while milking their residual fame, but more importantly there won't be a next generation in same way it was till now. A major shift indeed, bigger than Hollywood ever saw. If the quality is there it will be unstoppable and unions won't make any difference, next country with more liberal laws can render very same stuff.
The point still stands though - if people won’t pay money to see AI generated movies but will to see human made ones, then the market doesn’t collapse.
As the cost of producing it drops, AI content may come to be seen as worthless background noise to culture and verified human-made media valued.
How do you reconcile "issues with details such as hands and eyes" and "a long way to go"? Seems to me that "details" remaining means we're almost there.
There already was a discussion somewhere about no more new main actors.
Why would studio executives bet on some new talent when they can generate Bruce Willis, John Depp or already put their faces on some less known actors in convincing way.
Then they make a reboot of a reboot of previous generation movies and it rolls ad infitum.
> Why would studio executives bet on some new talent when they can generate Bruce Willis, John Depp or already put their faces on some less known actors in convincing way
Because AI Willis can't sign autographs, do press tours, and get randomly sighted by fans. Hollywood is about more than just content, it's about cult of personality, and to have that, you need a living person.
Why couldn't it? Imagine doing some Comic Con 45 min special with your virtual actor. Yes its not the same fame, but people will adjust quickly. If they mix together good voice recognition in real time they can do press conferences rather easily, its just about gluing well together already existing pieces. There may be hilarious glitches from time to time, but who cares?
No real fame was achieved about stupid stuff like opening restaurants or random sights in the town, I think we as society can easily live without that.
We already have that, though - the Gorillaz, Hatsune Miku, etc. They've had success, but not nearly enough to drive any kind of mass-market adoption of the concept that would displace real-life celebs from the talk show marketing circuit.
I would argue that for 90% of people on earth this does not make a difference. For me any big movie star is exactly the same as some virtual entity, never seen one, never been close to any places where I could meet one.
There is difference where I could make effort to fly to places where they give autographs and meet with fans but I also don't care enough. I imagine people that do care about that either live close by where they can have easy access or are weirdos.
There’s already too many movies and people get bored watching more of the same. If thanks to AI there’s a flood of cheaper mediocre movies content it could kill the remaining excitement.
You can choose your favorite actors/characters/avatars and have the AI re-render the story featuring these people. Watch Lord of the Rings with all the hobbits replaced by Bratz dolls or your favorite K-pop girl band.
AI will also eliminate any and all user interface design/construction work (UX). So if you're a designer, prepare yourself for professional reorientation now.
Reading this in the context of Sora, this is contingent on:
1. movie or tv show length generation needs to be reasonably affordable,
2. the generated output must look real beyond first look (text is readable, physics is correct, no dream-like morphing),
3. continuity between scenes
4. prompts can accurately produce the artistic vision from directors and producers faster than a human just doing it
5. the public will accept it
I am reminded of a scene in Westworld, where basically a someone can sit a desk and describe in natural language a scene to a computer for instant rendering. The user can now concern himself exclusively on the details of the story. Perhaps roles will transition to becoming full time story tellers, and the minutia of vfx, animating, rigging, 3d models, actors, cameras, set designers, trucks, food, lighting, electrical, location, licensing, permits, etc. all go away.