I look forward to a day when capabilities like this are trivial and boring to the average person. When my phone (locally) will be able to generate a fully voice acted 24 episode series anime on a whim for a meme with my group chat. It's astounding what we can do now, but will be completely ignorable before we know it, which is equally wild.
Mainly just pulling from how my friends use it now, for the fun of it. We have sent each other silly generated videos of stuff we did or talked about. Kind of like how we used up compute, storage, and bandwidth to send memes and funny videos. If it were the 80s we would spent proportional cost to share a cat picture perhaps, but look at what we did with the bleeding edge even then. Turned the computers into game machines, shared things, wasted time. This is just an extension of that idea. Everything cutting edge ends up getting used for silly things.
I think they are talking about making their own series, not selling something, which will be valuable to exactly one person (or in this case a group chat). Which is awesome.
> AI doesn’t increase the value of content, it makes it meaningless by destroying scarcity.
Most art is in some way or form derivative of another work or a combination of somesorts. I enjoy derivative works. To me, AI already has value. I don't seek endless gratification through originality.
A whole series would involve writing thousands of prompts, not just a few seconds. But since you bring it up, one thing I think we can eventually expect is an avalanche of fan created continuations of cancelled shows, like an 8th season of Star Trek: The Next Generation, or a decent 8th season of Game of Thrones. Instead of writing a tie-in novel, aspiring genre authors will write screenplays and turn them into AI movies.
Like the hot take but it is needlessly negative because it doesn’t go far enough.
You could make the same argument about musical instruments, or being able to record and playback music, dj tools, etc.
I think what you get is the power law distribution for tons of content. Some of the stuff is still mega valuable, but distribution just gets more and more important and it’s harder and harder to break through. This is what the “democratization” of any previously difficulty-gated endeavor does.
More niches will be created, more fragmentation in tastes, stuff like that. Not just completely valueless content.
Incumbents and platform providers get to win through it all though, because humans still want to fill their time.
Ironically I think it might, by serving as a synthetic generator to learn to better understand artistic values for those who need it. We wouldn't know what is ever so slightly wrong about replicator Earl Grey, without trying it for a while.
A lot of people right now claim that they cannot tell apart AI output from human art, while many of them seem to grow rather agitated and stressed after repeated exposure. I bet they're going to be forced into exclusively making art manually, or viewing exclusively human art at some point, and through that ways, AIs could increase value of human made data.
I find it funny that wishes for AI arts always seem to be more anime and end to Marvel slop. They want human slop go away thanks to AI slop? I'm not meaning to call out someone for contradiction or inconsistency - as I do understand the sentiment. But it gives me chuckles.
More content will be made in a single month than all of human history up to this point.
No more Disney-fication, no more Marvel / Star Wars "mass media slop". We'll have media that caters to people's long-tail interests. If you have a passion for Egyptology and Atlantis, you'll be able to watch a steampunk adventure about the Egyptians waging war with the Atlanteans. But perhaps with the serious tone of "The Wire". That would never have been greenlit before, but it'll soon be possible.
Good creators will arise just like good indie music, indie manga/comic, and indie game creators. Discovery will be the problem to solve for creators. There will be an abundance of talent that is finally able to create their vision rather than nepotism their way into one of 500 limited annual roles of autonomy.
Small creators who grow large like VivziePop [1] and PsychicPebbles [2] will be the model for the future of content. They start small on YouTube, grow large, and eventually have their own large-scale distribution and franchises.
The creative world is about to get orders of magnitude better. Not 2x, not 10x, but easily 1,000x.
I hate most movies and tv shows, but love the medium. The problem is most content produced just isn't my vibe. I like super artsy stuff, but also have particular tastes. That's going to change dramatically. Stuff will start fitting the shape of my interest graph.
> We'll have media that caters to people's long-tail interests.
This isn't inherently a bad thing but I don't believe it's without its costs, one of which being that with everyone watching (potentially) completely unique media there'll be no shared cultural artifacts to communicate with others about.
VivziePop may have started as a small creator but the recent creations such as Hazbin Hotel have now become things where the enjoyment can be shared with other fans, which in some ways is the secondary purpose of all media. Media enjoyed alone can still be rewarding, media enjoyed with others can be much more than that (except for going to the cinema with strangers, that can disappear immediately).
It seems like a common idea that if we can just generate a practically-infinite stream of media then we've solved some kind of problem with there just not being enough "content" (I hate using the term in this way but it's concise), and while I do sympathise with the points of view from people that can't find things they like, I also don't really believe that "Content is a problem".
More variety itself isn't a problem, but I'm not convinced that in general there's a _lack_ of content. Humans have already reached peak saturation in terms of
the sheer amount of text, audio and video that's created every day. No single item of media is itself a problem, but perhaps the total aggregate of everything isn't helpful.
> There will be an abundance of talent that is finally able to create their vision rather than nepotism their way into one of 500 limited annual roles of autonomy.
Is it possible that part of what creates quality is having to collaborate with others? Would Star Wars have been as good as it was without Marcia Lucas?
Yay, I can't wait for entertainment to stop being soulless corporate sequels and reboots and derivatives, and now with an LLM I can now get soulless machine generated sequels and reboots and derivatives!
Sounds like hell. Yeah let’s have everybody live in their own bubble, that’ll work out great (not) and let’s have that bubble be able to be influenced in the tiniest ways so they will consume more products, amazing!
Maybe we can finally get shows/movies that are faithful to the books instead of whatever slop we normally get (Looking at you WoT). Obviously a book != a script (like script for a show/movie, not a code script) but with LLMs I feel confident it could generate a script that could be reviewed directly or review the final video then go back and tweak the script as needed ("Ahh crap, it misunderstood the 'you' in that sentence and the wrong character just spoke in the generated video, let me fix that and re-gen that scene").
>Egyptians waging war with the Atlanteans. But perhaps with the serious tone of "The Wire". That would never have been greenlit before
This is quite literally what the Marvel / Star wars mass media slop is, a sort of syncretistic version of every pop culture phenomenon scientifically engineered down to A/B tested audience taste. It's excatly what you're complaining about, never-ending banal stuff. It'll be Disneyfication on steroids.
It's a sort of media masturbation, hooking your brain up to itself and feeding it its own output, what the large studios are already doing with their franchises, just on an individual level. You'll basically sit in your own never-ending Marvel hell, basically D.F Wallace's Infinite Jest.
Art is obviously the opposite, something external and disruptive, a creative act from someone who isn't you.
One real example of AI generated video that in my opinion marks the start of common people consuming AI video are the bigfoot vlog vidoes. The first time ever i see normal people watching this and genuinely like it.
> No more Disney-fication, no more Marvel / Star Wars "mass media slop". We'll have media that caters to people's long-tail interests.
Lol. Where do you think these models will get the source data for these amazing generations?
Also, an average person's "long tail" is incredibly boring (I know mine is). You will need someone to create a next Breaking Bad, or a next Discworld, or... Your slop-generating machine will not be able to generate it from a "long tail"
There is a moment where people might be so engaged it might not be profitable anymore though.
Brands pay for ads because it generate sales. If people become vegetables that doomscroll all day, barely pausing to eat and have a dump, they will also stop purchasing stuff.
Or maybe people will be so addicted that social medias as a drug becomes a commercial product in its own riggt and tiktok can paywall it.
It will happen for different people at different times, but at some point the realization that you are looking at people who don't exist living lives that aren't possible to sell you products that won't help you will click for you, and when it does it will cut you off from society in a way we don't currently have words for.
I agree with you, but the owners and customers of ad-supported communities like Facebook have an incentive to inject clickbait from fake users posing as humans. Facebook is already well along that road, with a captive if aging userbase. Maybe they'd flee if they had an alternative.
Most people on social media don't interact, just consume. You are talking to small self selected minority
If there was an HN clone generated on the fly for you or the guy you replied to then what's the difference? Especially if you imagine you didn't know it was generated. That's the problem with this tech, for you there's no difference. But probably a difference for society.
You must have missed a train or two over the past fifteen years, so called “social media” have very little to do with the social part they used to, they are mostly algorithmically driven dopamine shots to capture user's attention, and TikTok is the purest form of it.
I don't doubt that we will be able to make a content machine. Honestly, we had that on tiktok before generative AI. But I feel like this kind of misses the point of art. The idea of art for me is on some level experiencing someone else's intention and connecting with others based on that. Generative models have no intention or experience so I am uninterested in what they have to say even if it is technically well executed.
Some of the shots are impressive but… Even among these hand-picked examples there’s a plenty of unnatural movement. And it seems like it was trained on the most hyperactive subset of tiktok as it apparently can’t hold a scene for more than 5 seconds.
While it pulls some pretty difficult things, it seems to struggle with other *seemingly* simple ones.
The piano in the beginning or the photo camera used by the photographer has "AI text" written on it. The old man with the beret in the cafe goes through his beret with his hand. The girl on the seaside looking back turns her head too much almost like an owl. The boy-in-a-bike-through-an-ewuropean-city scene ends on a square with an amorphous being in a unicycle under the tree...
> I wouldn’t be surprised if in 5 years all content is generated on the fly. You say something, and a 5-second video plays in response.
live mode means content stops being fixed assets and starts becoming ephemeral responses
video turns into output stream, not uploads
voice prompt is the new swipe
what they're doing isn’t pushing a format shift, they’re testing runtime content systems
on backend they’re compressing model infra with comet and tilting up llms that run cheaper and faster
that combo means they can serve gen content at scale without needing to batch or cache
if that holds, feed stops being a scroll and becomes a render loop
nothing about this is about media anymore, it’s turning the app into a low-latency model host disguised as video platform
really coo, but wheres the sound? i'd expect that they'd have built in the sound model since its gonna look like SOTA for video, VEO3 is great for video but the audios what knocks it out of the park
I work on AI solutions for a major video streaming company, and the problem with VEO3 is that it doesn't have any consistency between prompts. E.g. I can not upload a reference image of what a character looks like, and if I say in one video "the old priest bends down" and in the next video "the old priests picks up the coin", the priest will look very different between shots.
Veo3 does support image to video, so what you can do is create an image that is the start of a scene, and then use that to generate the actual scene. Unfortunately, Veo3 is really bad at this. I expect this will improve over time.
Although I'm not super excited about this Seedance model personally, I do really like that it focuses on consistency between shots. I hope this puts pressure on increased performance from Veo3 in that regard.
I’m starting to wonder if it will. There seems to be this pattern that an awesome T2V model will come out and everyone starts clamoring for an I2V model and then when the I2V version drops a couple months later it’s not as good. I’m starting to get the feeling that I2V is just intrinsically challenging in a way that makes it hard to do well at all.
This is obviously going straight to TikTok. The big issue is it's going to open the flood gates on their own platform.
Anyway, if everyone wants to be a content creator, why not charge them for the privilege of that desire? A content creator will forever need AI-generated something. So now we move from "you get to post your content for free" over to "you get to now pay us through this AI-gateway to post your content".
There is something in motion heavy videos that is making me nauseas/sick in my stomach. Last time I felt this was with first Sora release. It's not as bad as Sora, but its there. Veo 3 didn't gave me these feelings or may be I haven't seen its motion heavy samples.
Does anyone else feels same looking at motion heavy samples of Seedance?
Has the realism of AI already caught on to that of animated CGI movies?
I assume that an expert in CGI can point out obvious flaws in these outputs. But I wonder if it is possible to fix those details by prompting it to change only specific segments.
There is also the question of how much compute/money they are spending per second of output, compared to a high-budget Hollywood CGI.
Like every AI launch demo I've ever seen, the results are unbelievably high quality, but if you take a second to read the prompts they never quite match. Here basically every single example is ignoring a portion of the prompt; sometimes the camera directions, sometimes the atmospheric description, sometimes making up very distinct elements that were not mentioned at all. People talk about "AI slop" because these models are really good when you just want "content" and you don't really care exactly what it looks like, but if you are trying to produce something specific, which you are in every real-world use case I can think of, it is very frustrating and often impossible to get there.
Decent 1080p quality. Not bluray level, but getting close. Definitely ahead of every other video generator.
Video production just got a lot cheaper and requires very few skills. This is basically destroying the creative video production industry (ads, product videography, youtube content of all kinds) and probably VFX industry as well.
They've been running tests for weeks under the covert name "Unicorn" and just renamed the model to Seedance a few days ago.
edit: I'm not sure why I'm being downvoted for this, except perhaps not liking the ByteDance angle.
China produces incredibly good video models and have been in the market lead for at least a year now. All of the top video models, save for Veo 3, are from China.
In fact, the only open source video models of note are all non-American (mostly Chinese, and one Israeli model).
People are already way too easy to get to believe conspiracy theories. Shit like Pizzagate or whatever is only going to get more common when bad people start making, "and look, here's the video proof!"
And we've already got Tiktok and Youtube Shorts just pumping the dopamine centers in the brain for short form content. Generating shit you like dynamically is going to be an addictive nightmare. The moment it gets monetized we're going to see the equivalent of slot machines pumped at us from every channel -- flashing lights and emotional tugs to get us to part with our valuable money or attention.
And that's to say nothing of the impact these tools have on artists and creative people or the costs to train and deploy these tools.
We're already seeing it today. The amount of 'footage' about LA right now that's showing some sort of war zone that is clearly AI generated, but being consumed as if it was real is staggering.
Yes, perfect AI content has multiple issues, that need to be addressed differently
- treating certain content in the same way that drugs are treated. Lots of countries are already moving towards age restrictions for social media.
- some kind of hardware-provided signatures for images and video, anything else must be assumed to be generated
Will be interesting for kids growing up - the peer pressure is now already very high to have smartphones, to be on Whatsapp, Instagram, TikTok, this will only get worse.
Maybe if I have kids I will found some Amish-like community with only 90's tech (only half joking).
We can only hope that people become aware that the Internet is a bullshit-machine and will only pull their news information from journalists, but I know this is wishful thinking.
We've basically flooded the information space with r-strategists.
In evolution, rapid reproduction gives an advantage to spamming low-quality offspring [1], and rapid selection without agglomeration [2,3] incentivizes antisocial behavior.
Ideas spread, mutate, and evolve just like animals [4]. So when the Internet made it free for anyone to transmit information to millions of people instantly, trustworthy information sources [5] and prosocial cultural values started dying [6], as literally the worst and craziest people become dominant [7,8,9,10,11].
...Presumably "AI" is going to make this even worse, and immeasurably so.
I feel so bad for the next generation who will never have watched man made movies, they will not be able to tell whether something is junk or not because there will be no baseline.
Only light skinned people on the video examples.
Ethnic diversity and accuracy used to be a problem with the models of the past. I wonder how the model would excel at prompts grokking at that.
I look forward to a day when capabilities like this are trivial and boring to the average person. When my phone (locally) will be able to generate a fully voice acted 24 episode series anime on a whim for a meme with my group chat. It's astounding what we can do now, but will be completely ignorable before we know it, which is equally wild.
Literally nobody will give a fuck about a 24 episode series that exists because you spent a few seconds writing a prompt.
AI doesn’t increase the value of content, it makes it meaningless by destroying scarcity.
Tea. Earl Grey. Hot.
Mainly just pulling from how my friends use it now, for the fun of it. We have sent each other silly generated videos of stuff we did or talked about. Kind of like how we used up compute, storage, and bandwidth to send memes and funny videos. If it were the 80s we would spent proportional cost to share a cat picture perhaps, but look at what we did with the bleeding edge even then. Turned the computers into game machines, shared things, wasted time. This is just an extension of that idea. Everything cutting edge ends up getting used for silly things.
I think they are talking about making their own series, not selling something, which will be valuable to exactly one person (or in this case a group chat). Which is awesome.
> AI doesn’t increase the value of content, it makes it meaningless by destroying scarcity.
Most art is in some way or form derivative of another work or a combination of somesorts. I enjoy derivative works. To me, AI already has value. I don't seek endless gratification through originality.
A whole series would involve writing thousands of prompts, not just a few seconds. But since you bring it up, one thing I think we can eventually expect is an avalanche of fan created continuations of cancelled shows, like an 8th season of Star Trek: The Next Generation, or a decent 8th season of Game of Thrones. Instead of writing a tie-in novel, aspiring genre authors will write screenplays and turn them into AI movies.
2 replies →
Like the hot take but it is needlessly negative because it doesn’t go far enough.
You could make the same argument about musical instruments, or being able to record and playback music, dj tools, etc.
I think what you get is the power law distribution for tons of content. Some of the stuff is still mega valuable, but distribution just gets more and more important and it’s harder and harder to break through. This is what the “democratization” of any previously difficulty-gated endeavor does.
More niches will be created, more fragmentation in tastes, stuff like that. Not just completely valueless content.
Incumbents and platform providers get to win through it all though, because humans still want to fill their time.
2 replies →
Ironically I think it might, by serving as a synthetic generator to learn to better understand artistic values for those who need it. We wouldn't know what is ever so slightly wrong about replicator Earl Grey, without trying it for a while.
A lot of people right now claim that they cannot tell apart AI output from human art, while many of them seem to grow rather agitated and stressed after repeated exposure. I bet they're going to be forced into exclusively making art manually, or viewing exclusively human art at some point, and through that ways, AIs could increase value of human made data.
I find it funny that wishes for AI arts always seem to be more anime and end to Marvel slop. They want human slop go away thanks to AI slop? I'm not meaning to call out someone for contradiction or inconsistency - as I do understand the sentiment. But it gives me chuckles.
1 reply →
Who is going to spend time watching those episodes if content is so easy to make? Everyone will be busy watching their own generated content.
AI is going to watching those. AI watches much faster than humans.
Because few are great storytellers, people like celebrities, and having a shared cultural reference point.
Even TikTok's main fuel are likes and shares. Isolated material will just cease to be interesting, because it's from your own imagination.
1 reply →
Me too!
Can't wait to create shadowrun movies for example :)
More content will be made in a single month than all of human history up to this point.
No more Disney-fication, no more Marvel / Star Wars "mass media slop". We'll have media that caters to people's long-tail interests. If you have a passion for Egyptology and Atlantis, you'll be able to watch a steampunk adventure about the Egyptians waging war with the Atlanteans. But perhaps with the serious tone of "The Wire". That would never have been greenlit before, but it'll soon be possible.
Good creators will arise just like good indie music, indie manga/comic, and indie game creators. Discovery will be the problem to solve for creators. There will be an abundance of talent that is finally able to create their vision rather than nepotism their way into one of 500 limited annual roles of autonomy.
Small creators who grow large like VivziePop [1] and PsychicPebbles [2] will be the model for the future of content. They start small on YouTube, grow large, and eventually have their own large-scale distribution and franchises.
The creative world is about to get orders of magnitude better. Not 2x, not 10x, but easily 1,000x.
I hate most movies and tv shows, but love the medium. The problem is most content produced just isn't my vibe. I like super artsy stuff, but also have particular tastes. That's going to change dramatically. Stuff will start fitting the shape of my interest graph.
I'm so excited.
[1] https://en.wikipedia.org/wiki/Vivienne_Medrano
[2] https://en.wikipedia.org/wiki/Zach_Hadel
> We'll have media that caters to people's long-tail interests.
This isn't inherently a bad thing but I don't believe it's without its costs, one of which being that with everyone watching (potentially) completely unique media there'll be no shared cultural artifacts to communicate with others about.
VivziePop may have started as a small creator but the recent creations such as Hazbin Hotel have now become things where the enjoyment can be shared with other fans, which in some ways is the secondary purpose of all media. Media enjoyed alone can still be rewarding, media enjoyed with others can be much more than that (except for going to the cinema with strangers, that can disappear immediately).
It seems like a common idea that if we can just generate a practically-infinite stream of media then we've solved some kind of problem with there just not being enough "content" (I hate using the term in this way but it's concise), and while I do sympathise with the points of view from people that can't find things they like, I also don't really believe that "Content is a problem".
More variety itself isn't a problem, but I'm not convinced that in general there's a _lack_ of content. Humans have already reached peak saturation in terms of the sheer amount of text, audio and video that's created every day. No single item of media is itself a problem, but perhaps the total aggregate of everything isn't helpful.
> There will be an abundance of talent that is finally able to create their vision rather than nepotism their way into one of 500 limited annual roles of autonomy.
Is it possible that part of what creates quality is having to collaborate with others? Would Star Wars have been as good as it was without Marcia Lucas?
Yay, I can't wait for entertainment to stop being soulless corporate sequels and reboots and derivatives, and now with an LLM I can now get soulless machine generated sequels and reboots and derivatives!
Sounds like hell. Yeah let’s have everybody live in their own bubble, that’ll work out great (not) and let’s have that bubble be able to be influenced in the tiniest ways so they will consume more products, amazing!
Maybe we can finally get shows/movies that are faithful to the books instead of whatever slop we normally get (Looking at you WoT). Obviously a book != a script (like script for a show/movie, not a code script) but with LLMs I feel confident it could generate a script that could be reviewed directly or review the final video then go back and tweak the script as needed ("Ahh crap, it misunderstood the 'you' in that sentence and the wrong character just spoke in the generated video, let me fix that and re-gen that scene").
1 reply →
>Egyptians waging war with the Atlanteans. But perhaps with the serious tone of "The Wire". That would never have been greenlit before
This is quite literally what the Marvel / Star wars mass media slop is, a sort of syncretistic version of every pop culture phenomenon scientifically engineered down to A/B tested audience taste. It's excatly what you're complaining about, never-ending banal stuff. It'll be Disneyfication on steroids.
It's a sort of media masturbation, hooking your brain up to itself and feeding it its own output, what the large studios are already doing with their franchises, just on an individual level. You'll basically sit in your own never-ending Marvel hell, basically D.F Wallace's Infinite Jest.
Art is obviously the opposite, something external and disruptive, a creative act from someone who isn't you.
1 reply →
One real example of AI generated video that in my opinion marks the start of common people consuming AI video are the bigfoot vlog vidoes. The first time ever i see normal people watching this and genuinely like it.
> No more Disney-fication, no more Marvel / Star Wars "mass media slop". We'll have media that caters to people's long-tail interests.
Lol. Where do you think these models will get the source data for these amazing generations?
Also, an average person's "long tail" is incredibly boring (I know mine is). You will need someone to create a next Breaking Bad, or a next Discworld, or... Your slop-generating machine will not be able to generate it from a "long tail"
10 replies →
The future is something like the TikTok algorithm, but generated on the fly.
As you scroll, it learns what you like and generates more videos.
With enough context fed into the model of what you react to, the content will be so mesmerizing that you won't be able to look away
This is chilling and also seems inevitable long term
I imagine there are some boundary constraints - mobile video can only be so compelling. I would guess we're already pretty close to that threshold.
2 replies →
The only winning move is not to play. Just don't use stupid apps. Problem solved.
It's more depressing if it DIDN'T happen soon. The technology should be here, years ago.
I think it will also try to influence what you like to maximize engagement unfortunately...
There is a moment where people might be so engaged it might not be profitable anymore though.
Brands pay for ads because it generate sales. If people become vegetables that doomscroll all day, barely pausing to eat and have a dump, they will also stop purchasing stuff.
Or maybe people will be so addicted that social medias as a drug becomes a commercial product in its own riggt and tiktok can paywall it.
There's a cliff though.
It will happen for different people at different times, but at some point the realization that you are looking at people who don't exist living lives that aren't possible to sell you products that won't help you will click for you, and when it does it will cut you off from society in a way we don't currently have words for.
3 replies →
I think this is an unfortunate misunderstanding of why people like and use social media.
ChatGPT can already generate endless "comments". And yet you're here.
I agree with you, but the owners and customers of ad-supported communities like Facebook have an incentive to inject clickbait from fake users posing as humans. Facebook is already well along that road, with a captive if aging userbase. Maybe they'd flee if they had an alternative.
No one knows if you're a bot or a human.
HN users that actively engage with comments are probably 0.0001% of overall social media users
> And yet you're here.
Some people are talking to ChatGPT though. We are here, for now.
1 reply →
Most people on social media don't interact, just consume. You are talking to small self selected minority
If there was an HN clone generated on the fly for you or the guy you replied to then what's the difference? Especially if you imagine you didn't know it was generated. That's the problem with this tech, for you there's no difference. But probably a difference for society.
You must have missed a train or two over the past fifteen years, so called “social media” have very little to do with the social part they used to, they are mostly algorithmically driven dopamine shots to capture user's attention, and TikTok is the purest form of it.
Later on a "live mode" which is realtime generated content, guided by your voice. Netflix could also have this as a feature.
The thing is that people don’t know or won’t say what they like.
The TikTok algorithm is good at figuring out what you do. Not what you say. So the content will be a lot more engaging.
3 replies →
Will it learn that I don’t like ads?
It will figure out how to make ads that you do watch. If a generated video is good enough, people will watch it even though it’s an ad.
3 replies →
Everything is ad nowadays. They have already been embedded in every single "content creation crap", movie, tv shows, music videos.
That's the secret - everything will be an ad
It might MAKE you like ads..
That's the near future. Look farther down the line and it's netflix. Keep scrolling and it generates entire movies and shows based on what you like.
Probably before that though we'll see AI movies pre-generated before we see them generated on the fly during scrolling.
I don't doubt that we will be able to make a content machine. Honestly, we had that on tiktok before generative AI. But I feel like this kind of misses the point of art. The idea of art for me is on some level experiencing someone else's intention and connecting with others based on that. Generative models have no intention or experience so I am uninterested in what they have to say even if it is technically well executed.
Some of the shots are impressive but… Even among these hand-picked examples there’s a plenty of unnatural movement. And it seems like it was trained on the most hyperactive subset of tiktok as it apparently can’t hold a scene for more than 5 seconds.
While it pulls some pretty difficult things, it seems to struggle with other *seemingly* simple ones.
The piano in the beginning or the photo camera used by the photographer has "AI text" written on it. The old man with the beret in the cafe goes through his beret with his hand. The girl on the seaside looking back turns her head too much almost like an owl. The boy-in-a-bike-through-an-ewuropean-city scene ends on a square with an amorphous being in a unicycle under the tree...
ByteDance has been testing their model on the Model Arena for weeks. They were covertly calling their model "Unicorn" until just a few days ago.
It's already ranking better than Google Veo 3:
https://artificialanalysis.ai/text-to-video/arena?tab=leader...
By a LOT too lol, not even close, wow, that said, i imagine if they enabled veo3's sound... it would win lol
> I wouldn’t be surprised if in 5 years all content is generated on the fly. You say something, and a 5-second video plays in response.
live mode means content stops being fixed assets and starts becoming ephemeral responses
video turns into output stream, not uploads voice prompt is the new swipe
what they're doing isn’t pushing a format shift, they’re testing runtime content systems on backend they’re compressing model infra with comet and tilting up llms that run cheaper and faster that combo means they can serve gen content at scale without needing to batch or cache if that holds, feed stops being a scroll and becomes a render loop nothing about this is about media anymore, it’s turning the app into a low-latency model host disguised as video platform
really coo, but wheres the sound? i'd expect that they'd have built in the sound model since its gonna look like SOTA for video, VEO3 is great for video but the audios what knocks it out of the park
I work on AI solutions for a major video streaming company, and the problem with VEO3 is that it doesn't have any consistency between prompts. E.g. I can not upload a reference image of what a character looks like, and if I say in one video "the old priest bends down" and in the next video "the old priests picks up the coin", the priest will look very different between shots.
Veo3 does support image to video, so what you can do is create an image that is the start of a scene, and then use that to generate the actual scene. Unfortunately, Veo3 is really bad at this. I expect this will improve over time.
Although I'm not super excited about this Seedance model personally, I do really like that it focuses on consistency between shots. I hope this puts pressure on increased performance from Veo3 in that regard.
> I expect this will improve over time.
I’m starting to wonder if it will. There seems to be this pattern that an awesome T2V model will come out and everyone starts clamoring for an I2V model and then when the I2V version drops a couple months later it’s not as good. I’m starting to get the feeling that I2V is just intrinsically challenging in a way that makes it hard to do well at all.
Why do all the examples have a large circle in them?
Is this available somewhere to use?
> Seedance 1.0 will be integrated into multiple platforms in June 2025, including Doubao and Jimeng
> https://www.doubao.com/chat/create-video
> https://jimeng.jianying.com/ai-tool/video/generate
https://arxiv.org/pdf/2506.09113
This is obviously going straight to TikTok. The big issue is it's going to open the flood gates on their own platform.
Anyway, if everyone wants to be a content creator, why not charge them for the privilege of that desire? A content creator will forever need AI-generated something. So now we move from "you get to post your content for free" over to "you get to now pay us through this AI-gateway to post your content".
There is something in motion heavy videos that is making me nauseas/sick in my stomach. Last time I felt this was with first Sora release. It's not as bad as Sora, but its there. Veo 3 didn't gave me these feelings or may be I haven't seen its motion heavy samples.
Does anyone else feels same looking at motion heavy samples of Seedance?
Has the realism of AI already caught on to that of animated CGI movies?
I assume that an expert in CGI can point out obvious flaws in these outputs. But I wonder if it is possible to fix those details by prompting it to change only specific segments.
There is also the question of how much compute/money they are spending per second of output, compared to a high-budget Hollywood CGI.
Given how bad regular (non-animated) CGI often is in Hollywood movies nowadays, I don’t think the bar is that high.
Change management will indeed be interesting.
"Old man" doesn't look that old to me (guess that means I'm old!)
As a Chinese,I am proud of ByteDance. It make the China's AI industry at the top of the world. Although we have been banned by USA
you guys keep making everyone else kind of look bad. all eyes on china soon/now for AI
Very nice being able to read an actual paper on a powerful text-to-video model.
Yeah. It is great. So apparently separating spatial / temporal attention works if you are careful and train with large enough dataset too!
Like every AI launch demo I've ever seen, the results are unbelievably high quality, but if you take a second to read the prompts they never quite match. Here basically every single example is ignoring a portion of the prompt; sometimes the camera directions, sometimes the atmospheric description, sometimes making up very distinct elements that were not mentioned at all. People talk about "AI slop" because these models are really good when you just want "content" and you don't really care exactly what it looks like, but if you are trying to produce something specific, which you are in every real-world use case I can think of, it is very frustrating and often impossible to get there.
Man I hate when sites hijack the page up/down keys.
Decent 1080p quality. Not bluray level, but getting close. Definitely ahead of every other video generator.
Video production just got a lot cheaper and requires very few skills. This is basically destroying the creative video production industry (ads, product videography, youtube content of all kinds) and probably VFX industry as well.
It's beating Google Veo 3 in the model arena:
https://artificialanalysis.ai/text-to-video/arena?tab=leader...
They've been running tests for weeks under the covert name "Unicorn" and just renamed the model to Seedance a few days ago.
edit: I'm not sure why I'm being downvoted for this, except perhaps not liking the ByteDance angle.
China produces incredibly good video models and have been in the market lead for at least a year now. All of the top video models, save for Veo 3, are from China.
In fact, the only open source video models of note are all non-American (mostly Chinese, and one Israeli model).
Can't wait for infinite ai generated video slop
Sadly another closed model that will never go open weights i'd guess
We're so fucked, as a species.
People are already way too easy to get to believe conspiracy theories. Shit like Pizzagate or whatever is only going to get more common when bad people start making, "and look, here's the video proof!"
And we've already got Tiktok and Youtube Shorts just pumping the dopamine centers in the brain for short form content. Generating shit you like dynamically is going to be an addictive nightmare. The moment it gets monetized we're going to see the equivalent of slot machines pumped at us from every channel -- flashing lights and emotional tugs to get us to part with our valuable money or attention.
And that's to say nothing of the impact these tools have on artists and creative people or the costs to train and deploy these tools.
We're already seeing it today. The amount of 'footage' about LA right now that's showing some sort of war zone that is clearly AI generated, but being consumed as if it was real is staggering.
Yes, perfect AI content has multiple issues, that need to be addressed differently
- treating certain content in the same way that drugs are treated. Lots of countries are already moving towards age restrictions for social media.
- some kind of hardware-provided signatures for images and video, anything else must be assumed to be generated
Will be interesting for kids growing up - the peer pressure is now already very high to have smartphones, to be on Whatsapp, Instagram, TikTok, this will only get worse.
Maybe if I have kids I will found some Amish-like community with only 90's tech (only half joking).
That's been happening since Photoshop. Describing this as some apocalyptic event is absurd.
Photoshop requires skills. The comparison with Photoshop is absurd.
3 replies →
We can only hope that people become aware that the Internet is a bullshit-machine and will only pull their news information from journalists, but I know this is wishful thinking.
We've basically flooded the information space with r-strategists.
In evolution, rapid reproduction gives an advantage to spamming low-quality offspring [1], and rapid selection without agglomeration [2,3] incentivizes antisocial behavior.
Ideas spread, mutate, and evolve just like animals [4]. So when the Internet made it free for anyone to transmit information to millions of people instantly, trustworthy information sources [5] and prosocial cultural values started dying [6], as literally the worst and craziest people become dominant [7,8,9,10,11].
...Presumably "AI" is going to make this even worse, and immeasurably so.
---
1. https://en.wikipedia.org/wiki/R/K_selection_theory
2. https://journals.plos.org/plosone/article?id=10.1371/journal...
3. https://en.wikipedia.org/wiki/Context_collapse
4. https://en.wikipedia.org/wiki/Memetics
5. https://en.wikipedia.org/wiki/Decline_of_newspapers
6. https://theweek.com/culture-life/third-places-disappearing
7. https://globalnews.ca/news/1157137/internet-trolls-are-sadis...
8. https://www.engadget.com/2018-03-19-study-shows-distribution...
9. https://old.reddit.com/r/slatestarcodex/comments/9rvroo/most...
10. https://www.ipr.northwestern.edu/news/2023/why-are-online-po...
11. https://en.wikipedia.org/wiki/Kakistocracy
[dead]
Hollywood is cooked isn't it?
Rshtach ag
I feel so bad for the next generation who will never have watched man made movies, they will not be able to tell whether something is junk or not because there will be no baseline.
I dont think they care as long as the content is good. Even memes popular at that demographic are AI generated today.
Only light skinned people on the video examples. Ethnic diversity and accuracy used to be a problem with the models of the past. I wonder how the model would excel at prompts grokking at that.
0:10 into the video, that's light skinned to you?!
This model is from China. It isn't even being made available internationally (yet).
[flagged]