Skip to main content
MANIFOLD
In 2028, will an AI be able to generate a full low-quality movie to a prompt?
395
Ṁ1.6kṀ360k
2028
80%
chance

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:

Get
Ṁ1,000
to start trading!
Sort by:

Thanks for bringing me over here @logan . Looking through some of these videos , I think every single one(here and what I've seen elsewhere) seems to fail the bar in my mind for meeting the minimum visual consistency, plot advancement, basic cinematography, and "not painful to watch" criteria. AND they seem to be at most a couple minutes.

Some questions I'm asking myself when I watch these:

  1. Does the world actually work this way(or within the world being displayed, are the inputs yielding outputs that make sense)? I.e. if someone puts water in their hands, you should then see water in their hands when they move them and not nothing or some other substance. If something falls in one direction, it should continue falling until something causes it not to. An object shouldn't randomly spawn extra appendages, something magically appear or disappear from someone's hands, etc

  2. Does how we are viewing the world make sense in a way that isn't jarring(cinematography)? Are cuts/camera zooms/pans/etc. done in a way where you could at least say "this kinda sucks, but at least I'm seeing what the story is trying to convey to me"? As an example if we are looking at someone singing a song; the camera shouldn't randomly focus on a shelf and tilt and pan back and forth away from the focal point. Even very bad movies/low quality don't do that

  3. Plot coherence/set coherence: When we are looking at one place, does it look the same within the shot and from shot to shot? Do people/objects look the same/talk the same/react the same/stay the same size/etc.? Do things actually move forward within the story?

  4. Do the combination of those things keep at least some level of interest and not make you uncomfortable from the obvious AI-ness of it?(I know the AI-ness criteria seems debatable, but I think that does connect to the discomfort criteria)

I see mentions of micro dramas at an hour long being able to clear the bar, but the key one I'm aware of is the hour-long Duanja series out of China which is (From what I understand) 60 to 100 separate 1-to-2-minute vertical episodes stitched together by human teams with human scripts and manual shot selection, rather than a continuous zero-shot film. But even counting that(which honestly I wouldn't answer yes to on a single of the questions I mentioned); I haven't seen a five-minute clip that meets the bar yet maybe not even a two minute clip. (I could be judging too hard) .

The best steelman counterpoint, which I noted in a comment on the high quality market, for video at 5-10 minutes is Google's 10-minute agentic demo or some of the 2D cartoon tools, but even Google's paper was celebrating just stopping two musicians from swapping instruments across a single camera cut; which I think is below the bar of something like the room and to me puts cold water on moving past the time frame(exponentially harder to keep consistency the longer you get) . The mollick essay(posted in the other high quality market) added more cold water to these markets for me with the challenges of agentic coordination

In any case whether I am just very uniformed(probably most likely case haha) or too judgy(second most likely case lol) I do think after reading through the comments that it would be very helpful to crowdsource some very clear yes/no questions to be added to the resolution criteria(similar but probably not exactly my list, For example I do think I need to do a better job of defining question set two) so that we could all agree on what counts.

Comment disclaimer: I learn so much from you guys here. Lots of awesome debate, I always pick up something new reading through the comments. Manifold to me has been great way to incentivize exploring new ways to think about topics that interest me and to make me question the opinions/knowledge/lack of knowledge/skill/ lack of skill I have. Hoping this overly long comment can be a positive addition and I'll get corrected, be less judgy, or maybe make the resolution of this market less combative with some criteria suggestions to prevent the coming rage lol

@ShaneBo

>As an example if we are looking at someone singing a song; the camera shouldn't randomly focus on a shelf and tilt and pan back and forth away from the focal point. Even very bad movies/low quality don't do that

I'm not sure how much time you've spent watching actually bad movies, but "the framing on this shot doesn't make sense" is absolutely par for the course.

>Do the combination of those things keep at least some level of interest and not make you uncomfortable from the obvious AI-ness of it?

Again, I think you might be apply a double-standard where "cheesy bad movie effects" don't count against them, but "obviously AI" does.

@logan fair pushback! I'll totally concede that #4(the "obvious AI-ness" discomfort) is the softest criteria on my list(and #2 the second weakest as I need to define better), and I definitely don't want to hold AI to a double standard against cheesy practical effects or bad CGI. That said those counters are not really reflective of the points I'm trying to distinguish; I'll try to rehash with some more clarity:

  • On #2(bad framing vs. drift): You're 100% right that terrible movies have awful framing(cut-off heads, out-of-focus shots, weird angles in The Room). What I mean by the shelf example isn't just "bad composition," it's that models often lose the focal subject entirely mid-generation and pan off into a wall or morph the background because they can't hold the scene past 4 seconds. Even an amateur holding a camcorder points the lens generally toward the action for a dialogue scene. The example you shared is a perfect supporting example of this as even this obviously terrible movie doesnt have the camera oddly tilting back and forth for no reason, shifting to the floor during a dialogue scene, etc. There are some weird cuts, bad framing, etc but thats not what I'm getting at here

  • On #4(cheesy effects vs. AI morphing): If an AI movie has cheesy-looking explosions or stiff acting, I wouldn't count that against it at all for a low-quality bar. The "painful to watch" factor for me is when the entire physical world has zero permanence(faces, shirt collars, lighting, and room geometry shifting every 3 seconds across every cut), which makes it exhausting for your brain to track in a way a guy in a cheap rubber monster suit standing in a real room isn't. And this is simplifying it as the combination of drift, lack of consistency, lack of casual physics, feedback loops, etc etc blends into something that makes even what you shared look appealing/comforting in comparison. When you watch some of these even if you don't quite understand why you feel this way the lack of permanence is jarrring. Watching some of the music videos from AI models is maybe the best example of this effect I see in the comments it. Its just so discomforting for your eyes. In a way that even the most poorly made by human movies aren't

I'm still very impressed by the video content generation and you can get really cool things and expect we will get even cooler/complex things soon, but I think there seems to general underrating of the complexity of visual storytelling/narrative and how we interpret them even at the bad quality level. The bad movie video you shared is to me so far above anything I ever seen done by AI to this point. I fully believe that AI can write a fully believable story board far better than The Room, Fully believe that it can generate images that are far more visually interesting and compelling than anything in the room. That it could generate dialogue much better, etc. BUT its the combinatorial element of all of those together and maintaining consistency, continuity, internal logic, permanence, etc; where it is just to me very below the worst quality human made movies right now even for a few minutes without significant harnessing, human intervention/pruning, etc

All that said, even if we throw #2 and #4 completely out the window and just stick to #1(basic causal physics) and #3(set/character permanence and plot moving forward across cuts); have you seen a zero-shot 5-minute clip that clears just those two?

Edit Note: I should be clearer as my argument does overstate the level of video generation wonkiness that occurs when applied to short videos(we don't necessarily see obvious issues like these at 30 second clips unless you're looking very closely), I should have been clear these problems occur very obviously for the low quality bar as soon as you get past a pretty low time marker. And the longer you extend the more you see this; where I'm far from certain a "doubling of the doubling" of performance will get us there in a year. Should also say that the few minutes around the 1 hour mark of the video logan shared; that is purely animated is probably below AI quality haha, but on the whole still standby that statement that it is far above current outputs across any 5 to 10 minute section and certainly when considered as a single product/output

Question for those saying someone might try to solve this in a couple weeks:

Has anyone yet made a 30- or 60- or 90- minute one-shot AI video that’s NOT minimally coherent / is still painful to watch?

we got lots of unwatchable/painful 15-second video clips before we got coherent ones. Seems like we should expect the same progression for hour or more long clips. Are people trying this with the current tools?

@JoshuaHedlund I would say 3-5 minutes 1 shot "not physically painful to watch" is within current model capabilities. probably need someone else to confirm/deny, though, because my tolerance for AI is unreasonably high.

@logan I’m not asking about the maximum length of not-physically-painful videos. I’m curious about 60-minute videos - is anyone generating that length at all yet, regardless of quality.

@JoshuaHedlund I'm unsure what you're asking.

The video I linked was 4 minutes long and passes my bar for "must only be at the level of the worst filmmakers". I'm not claiming you can't tell its AI or that it's something I would actively seek out. But I am claiming if you forced 100 people to watch it or Rock Me Tonight and then rate their pain on a scale of 1-10 the AI video would be rated less-painful.

edit: never mind. I misread 90min as 90sec. Yes. 90 minute videos generated by AI are still physically painful every single aspect you would expect to be a problem is a problem: screenwriting, voice acting, character consistency, etc.

I would say the reason we don't see more long-form 1-shot AI generated videos is because it's both very slow and expensive, and it's just more satisfying to spend a day and make something that's 4-minutes and borderline non-painful than a week to make something that's 30 minutes and excruciatingly so.

@logan thank you, yes that’s an example of what I was looking for. Do we know if that 80 minute video example was from a single prompt with no human edits?

@JoshuaHedlund Yes, all of the videos I linked were generated 1-shot from a short (1-2 sentence) prompt. The prompt for the 80 minute video was a high fantasy film in which a elf princess is rescued from her kidnappers by a loveable rogue and they must cross a treacherous wilderness to bring her home

@JoshSnider thanks, this is really great information and experimentation. Sounds like it was very much not a 1-shot process, though, even to get to what you got. Looking forward to your continued attempts.

Things are getting to a point where we will probably see some attempts at resolving this in the next couple weeks/months so would like to get a better sense of where everyones head is at.

The main limitation that I see for this (and the high quality market) is that cobbling something together wouldn't clear it, having a director at all wouldn't clear it, and having a script definitely wouldn't clear it, "to a prompt" does a lot of heavy lifting. It is, in large part, a market on context rot being solved or not by then. It does require an end-to-end solution that scales to a fairly long script to clear it. I don't think video quality was likely to ever be the main limiting factor.

I understand it does not have to be able to do it consistently, and doing it once to a prompt would probably clear the threshold, but producing something that is 1h20+ in a way that is consistently somewhat coherent still seems ways away. I haven't been tracking every new releases, but it does seem like every feature length project that would come close to the quality bar did not involve writing the script and had a lot of humans in the loop. I'm curious if someone has some examples of things that they would consider to be close to warranting a resolution that goes in the direction of "to a prompt".

@LuEE I think at current capabilities, someone could probably do this with Opus 5.5 using Seedance and like $10k

@LuEE I've already seen ~1 hr long vertical microdramas that probably clear the "quality" bar here or are very close. These films have some amount of human curation but like... I think Opus 5.5 could write a better script than the incredibly bad and pointless scripts that these films rely on, so I don't think that's the bottleneck.

@bens Thats the main disconnect, I don't think current models can take on a task as long as a feature length script even with a good harness. Unlike something more modular like coding a large project, writing a feature length script does require the model to keep the whole thing in mind, at least a lot more of the context in much more granular detail. Context cluing from heavily compressed representation leads to a lot of inconsistencies so I'm not sure compacting is the way forward. And it doesn't feel like this is an axis where improvements have been quite at the pace of other capabilities, the couple long reasoning benchmarks I'm aware off basically publish with bad results across the board and are ignored entirely thereafter, doesn't feel like anyone is looking to deal with context rot or knows how too. Altough bad movies tier script might very well be within reach here

What’s surprising is that “to a prompt” seems the clear bottleneck to me although I don’t have access to the latest models. I don’t know if the trivial minimal prompt examples provided will be sufficient for an AI to do it in the following narrow sense: the AIs have access to tools now so it could scrape a bunch of tool sets and other prompts that contain the structure/instructions to produce such a movie, but this would seem to be a gray area that goes against the spirit of the market, yet on the other hand we might not even likely know that it it did such searching unless it was a locally run AI or relied on prompting where we said “don’t search the internet and don’t use the internet for any resources for this project.”

@LuEE I think this should only resolve to Yes if there’s a clear consensus that Yes, this is now possible. One offs with an unknown amount of human curation shouldn’t be sufficient.

sold Ṁ5,335 NO

Seems like we might be at this point already. What will you do if it costs ~$10k in tokens to do?

@TheOtherKC Now I haven't seen The Room or any of the other bad film examples to compare, but this one is pretty bad. The main character is obviously a different person in each scene, and he keeps doing things that make no sense like going in the opposite direction of what the plot suggests he should be doing.

oh hi mark

I like to periodically revisit this. Here's what can be created using only open source tools (gemma4, LTX23, Flux-Klein-8b). Took about 9 hours to generate. Obviously you could do much better by using commercial models (Gemini-pro-3.1, NanoBanana Pro, Veo3) but would cost a few hundred dollars.

bought Ṁ1,500 NO

I don't think the intelligence difference between making Who Killed Captain Alex and Saving Private Ryan is that big from an AI's perspective. I would think this should be much closer to the high-quality movie market.

@ItsMe there is no intelligence difference between Who Killed Captain Alex and Saving Private Ryan