https://x.com/Purring_Lynx/status/1884539859559186746

@Blomfilter is this real? The quality in the clip provided looks incredible

@ScipioFabius yep, you can try it now on their site

@Blomfilter also with 'subject reference'... a film can have a continuous actor appearing in multiple scenes
https://hailuoai.video/discover/333795669912666112

Would an animated film count? (Like WALL-E or Tintin)

@louis IIRC correctly, is was clarified that it has to be live-action -- but perhaps that was a related market for a shorter timeline. I think this is a mistake, because animated films are often better than live-action.

@Duncn Well okay then. You could probably just render a """live-action""" through good enough animation though haha

I assume that a feature length film based on The Booth at the End would qualify?

@Duncn It's immaterial, because it's never been (at least to me, which is the correct view on most things anyway) about the ability to generate video, but about the ability to maintain coherence over the long run.

Which is why this market is not significantly lower than this one: /Shump/in-2028-will-an-ai-be-able-to-gener-2103270803d9

@skibidist good question. I'd throw this one into the mix, which is similar to the one you linked, but with (possibly?) shorter timelines*, a bit more fleshed out criteria and crucially it does need to be available to the public (and cheaply). https://manifold.markets/Fion/will-good-quality-personalised-ai-n

Seems to me like the present market should be lower, my market should be lower, and the one you linked should maybe be higher?

Unless everybody is basically just trading on "will AGI be achieved" in which case maybe it's right that they're all quite similar...

*Close date suggests the same, but Shump doesn't explicitly say if they mean the start of 2028 or at any point in 2028...

@skibidist compute requirements being at least an order of magnitude higher will definitely limit the amount of research iterations on full movies vs novels

If each second of a film needs 6k tokens a 2 hr film will need 40 million tokens. RN google can do a 10 million context window and magic has a 100 context window, I assume input context is easier than output context but what are the keys that ac. Stopping this today?

@Mad2live you don't need to hold the whole film in a single context window. Generate a shotlist and some metadata about your actors, sets, etc. for continuity, and generate shots one by one. A ten second shot is long in modern cinema.

@robm NO bettors dont know about bash

@Blomfilter then I will keep taking their mana

https://x.com/npcollapse/status/1881001370729459964?s=46&t=D5sNUZS8uOg4FTcneuxVIg

9 days out of three years used up; nearly 1%.

Did we make 1% of the progress required?

@DavidBolin this is such a tired comment. The last month saw the release of sora and veo2.

@robm It needed to be the last 10 days to keep up the pace, assuming that was 1% of the way (which I don't assume.)

@DavidBolin I reject your premise of framing progress over just a few days. When this question opened almost 2 years ago, we were still holding our papers over videos like this

We're more than 40% of the way between that and Sonic 3.

I understand this market on this terms: three years from now, all entertainment will be on AI hands. Disney et all will either have bought OpenAI or have sold to them, given any customer can just type a prompt and get exactly what they want. Why bother paying Netflix, Disney+ or any other? If you can make a 120m movie, is it too difficult to make a game from the same or similar prompt? Please destroy my argument, thanks!

@JaimeSantaCruz Time delay, cost for individuals necessitating still paying a distributor and designer, limitations in creative direction, random quality facets, lobbying against it by companies or by workers or general social dislike making it infeasible to bring to market.
But yes, if this happens it does indicate a pretty big upheaval, but I do kinda expect that with AI.

@JaimeSantaCruz there would be a gap between generating a movie and generating the best movie in one query. Until then, Netflix can be an army of "curators" to sift through the piles of slop suggested by "producers".

@ICRainbow Or, an AI can generate it (o6 + Sora(Expensive) ) which costs $10_000 per prompt (so it's better than making a movie traditionally, but every youtuber doesn't do it for every video)