EG "make me a 120 minute Star Trek / Star Wars crossover". It should be more or less comparable to a big-budget studio film, although it doesn't have to pass a full Turing Test as long as it's pretty good. The AI doesn't have to be available to the public, as long as it's confirmed to exist.
1,000People are also trading
From the Midjourney discord:
As you know, our focus for the past few years has been images. What you might not know, is that we believe the inevitable destination of this technology are models capable of real-time open-world simulations.
What’s that? Basically; imagine an AI system that generates imagery in real-time. You can command it to move around in 3D space, the environments and characters also move, and you can interact with everything.
In order to do this, we need building blocks. We need visuals (our first image models). We need to make those images move (video models). We need to be able to move ourselves through space (3D models) and we need to be able to do this all fast (real-time models).
The next year involves building these pieces individually, releasing them, and then slowly, putting it all together into a single unified system. It might be expensive at first, but sooner than you’d think, it’s something everyone will be able to use.
So what about today? Today, we’re taking the next step forward. We’re releasing Version 1 of our Video Model to the entire community.
@robm https://x.com/emollick/status/1935504703023899096
at this point a midjourney 2028 video will be better than a Love, Death, Robotics Short on Netflix
https://blog.google/technology/google-deepmind/ancestra-behind-the-scenes/
"To create “ANCESTRA”, Google DeepMind assembled a multidisciplinary creative team of animators, art directors, designers, writers, technologists and researchers who worked closely with more than 200 experts in traditional filmmaking and production, a live-action crew and cast, plus an editorial team, visual effects (VFX) artists, sound designers and music composers."
It's pretty cool that they did this, but it's a long way from generating a movie from a prompt. And the final result is only eight minutes long!
https://youtu.be/US2gO7UYEfY?si=vl5KCshO7cog4WUL
The WSJ made an AI video. But every shot required a lot of human effort to stitch outputs from different tools together and select a final clip that actually worked.
The script was also not AI generated, though I don't think that's the hardest part.
Total compute cost: ~$1000 for a three minute video.
@TimothyJohnson5c16 ...and the end result is way worse than a mediocre filmmaking student project. Can we have the market go to 10% already
@TimothyJohnson5c16 we're a month or two away from the halfway point of this market. I think this is a good marker of where things are now.
@skibidist when you figure out that you're wrong about this market in the second half of 2027 will you go quiet in the comments or will you try and shift the goalposts to something nonsensical?
@NebulaByte Probably the former since I would be 1) mana-bankrupt and 2) busy watching stargate/serenity cross-overs .. however I doubt there is a single timeline in the multiverse where that is realized
Sparkify https://sparkify.withgoogle.com/explore is another step in this direction
@ScottAlexander
What does "comparable to a big-budget movie" mean? It's possible for an AI to construct a compelling movie with strategic stylistic choices to overcome its limitations.
ie: A movie like Waking Life (2001) seems doable
Also to be extra pedantic - what makes a big budget movie? Like if I were to deepfake tom cruise into a few scenes would that automatically count?
Additionally, would it count if the AI took in additional data during the generation process
ie: The movie generates as people are are watching . While it generates, it simultaneously measures audience sentiment and updates the script as it is being made to calibrate on human preference.
As far as the UX setup flow goes, all it took was a prompt
Additionally+1: What is the domain breadth of possible generated movies for this to count?
ie: I could overfit an auto encoder to act as a a video player for Star Trek - where any prompt will output the same full high quality movie.
Additionally+2: How distinct does the movie have to be from existing media?
ie: It's possible to make a system that maps any input prompt and outputs the most similar existing high budget movie.
Additionally+3: How well does the movie have to adhere to the input prompt.
@VitorBosshard Yeah but no system (AI or otherwise) will be able to make a good movie out of any possible prompt, no matter how bad. If the prompt is "a movie that is just 90 minutes of waves splashing on a beach, with no dialogue" well then that's not a high quality movie unless you go way off prompt. Still, one way or the other I doubt the choice of prompt will be make-or-break for resolving this.
@ErickBall even the example is a little weird. I imagine a Star Trek / Star Wars crossover would piss off fans of both, no matter how "good" it is.
@robm Well if were actually a theatrical release, sure. A while back I had GPT-4 make a few concept plots for the crossover and they sounded fun to me. It was a whole thing where a Borg ship goes through a wormhole into the Star Wars universe, crazy antics ensue, and by the end the main characters from both universes collaborate to figure out how to send it back. Would watch.