Same as the main market but the AI must only be at the level of the worst filmmakers, rather than the best. So instead of a Marvel movie level of quality, think The Room level of quality. That's the yardstick that I will be using. Alternatively it can be any other well-known bad movie, like Birdemic or Who Killed Captain Alex. It doesn't need to have the "so bad it's good" qualities of The Room, but it needs to have a minimally coherent plot structure and cinematography. It must not be too painful for humans to watch the entire thing, and it needs to have the basics like characters say and do things that advance the plot, at least usually.
Main market:
1,000
3.00This is probably close to state-of-the-art for a script that generates a complete film with no human intervention. Do any The Room enjoyers want to comment on whether we're almost there yet?
https://www.youtube.com/watch?v=lc4JH3Q9vJ8
Cost is $0.20 and about 2 minutes to generate a 5 second clip. So a full 1.5 hour film created this way would cost $200 and take over a day to render.
Shots lacking character consistency are due to Google Gemini being censored so I had to use a weaker model.
@LoganZoellner This is unwatchable slop.
Kudos on at least the attempt if you actually spent $8 working on it but I don't think you even needed to generate 3 minutes of video... 30 seconds was enough time to see it lacked any coherence or consistency with plot/characters.
@bens Yes, that's the whole point.
Usually when you see "ai generated" films, they have been carefully edited by a human with 10's of re-rolls for every single shot and things like voice-acting, sound effect, music done by a human.
This is the output of a single program whose only input was
"a woman crash lands on an island filled with zombies and is saved by a billionaire who is the only other survivor"
This is a task I periodically re-visit, and while this is obviously unwatchably bad, this is what I was producing 7 months ago:
https://www.youtube.com/watch?v=foh-QWyu2QE
The biggest advances have been:
-video models, there are a bunch of open-source ones now (this is with Wan)_
-Multimodal out models (this is Gemini 2.5 flash-exp) mean it's theoretically possible to have consistent characters now. The main problem is that Gemini 2.5 is ridiculously overcensored, so in about 1/2 of the shots I had to use Flux, which is a pure text-to-image model.
I also have never seen The Room, but I have seen art-films that are almost this bad, so I wasn't quite sure what the standard is.
Relevant: Higgsfield's "ReelMagic". Summary from therundown.ai:
a multi-agent platform that transforms story concepts into complete 10-minute videos, claiming to streamline the entire production process into a single workflow.
The tool uses specialized AI agents for production roles like scriptwriting and editing, creating cohesive long-form outputs in under 10 minutes.
ReelMagic starts with a short synopsis, and then AI agents handle script refinement, virtual actor casting, filming, sound/music, and editing.
ReelMagic's smart reasoning engine automatically selects optimal AI models for each shot, and it has partnerships with Kling, Minimax, ElevenLabs, and more.
LTX Studio created an entire storyboard from not just a single prompt, a single sentence. And in a couple more clicks it stitched everything together into this video. I did add music from Suno, but movies on demand seem very close
@benshindel everything you listed is a solved problem, the technologies need to be put together but they exist
@benshindel Solvable with IPAdapter integration, ElevenLabs voice generation, and an LLM writing the plot and prompting Suno for fitting music.
I think it'll happen in 2026. The workflow for creating AI generated videos is already being streamlined.https://www.reddit.com/r/singularity/comments/1gse2lk/making_the_coca_cola_ai_ad_in_5_minutes_speedrun/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
@Shump i’m assuming this has the same resolution criteria apart from quality as the main market (needs to be generated by one software/agent/whatever with a single prompt, not strung together by a person from a number of AI programs)
This market is so mispriced. Getting rid of loans has been terrible for the epistemics of these 5-year markets in general. Basically everyone who is savvy just avoids betting on them altogether; the only reason I haven’t sold out of my position is because it’s mostly still old loans from pre-pivot manifold.
no one has put it into a single application, but you could:
1. use claude 3.5 to write a movie script
2. have claude convert that movie script into a list of shots
3. generate images for each of the shots using flux
4. animate each of the shots using runwayml
5. generate sound using FoleyCrafter
6. generate spoken dialogue using elevenlabs
7. animate the dialogue using hedra
8. generate background music using udio
the whole process would take a couple weeks and cost several hundred dollars, but the question says "able to" not "able to quickly and cheaply"
@LoganZoellner No, that’s explicitly false. Read the resolution criteria for the main market (and clarifications in the comments).
It would not count if you did these things separately.