Live data from Hacker News

SANA-WM, a 2.6B open-source world model for 1-minute 720p video

nvlabs.github.io

81–90 of 162 posts

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#81
post #16

I struggle with these world models from the perspective of video games (so this post is a particular perspective). I'm not a game developer myself, but some of my favorite games carry a deep sense of intentionality. For instance, there is typically not a single item misplaced in a FromSoftware game (or, for instance, Lies of P -- more recently). Almost every object is placed intentionally. Games which lack this inten…

> In general, I have the feeling that we are hurtling towards a world with less intentionality behind all the things we experience. Everything becomes impersonal, more noisy, etc.

You’re right - but that world is not the end of the story. The intentionality matters. Human creations matter because they connect us. I don’t know how long it will take, but people will build judgement as to what makes for good use of these tools to make meaningful things and expand our creative horizons in deeply human ways. Mind you, there will always be shallow slop. It’ll just take time for creators to learn how to use these tools to make something that isn’t slop.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#82

warning: viewing the videos that auto play on that page shot up my downloads to 350Mbps on that page

I only noticed after more than an hour with the page left open in a tab. Is it really streaming and re-streaming the same videos? There's too much to cache so it keeps re-transferring them indefinitely? I hope nobody leaves that page open on a metered or capped network connection. I'm surprised github hasn't suspended the page. Are AI researchers so used to burning through compute and network resources that they don'…

Nearly every website for papers about AI applied to graphics hangs my phone browser, so I'm assuming the answer's yes.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#83
post #16

I struggle with these world models from the perspective of video games (so this post is a particular perspective). I'm not a game developer myself, but some of my favorite games carry a deep sense of intentionality. For instance, there is typically not a single item misplaced in a FromSoftware game (or, for instance, Lies of P -- more recently). Almost every object is placed intentionally. Games which lack this inten…

I think your comment can be split into two questions: 1. Games derive some appeal from their intentionality and hand-crafted nature. Will these less-intentional experiences be as appealing? 2. Can these less-intentional tools still be used to create intentional designs?

On that first point I think it's important to remember that the lineage of video games comes from board & card games and sports. There's always been an ability to inject more complexity and less-intentionality into those things. Sports in some ways are like a simplified and altered role-play of war battles, and more realistic war roleplaying does exist but it has less appeal.

As humans we like solving things and noticing patterns and the intentionality of games taps into that appeal.

On the latter point I do think these world models will eventually be used to meaningfully contribute to building games. I think people will have to find new ways to design that balances intentionality against the freeform nature of these simulations, but it may take a while to have the capability to do so.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#84
silly question: what's "world" about what's being generated here? is the an actual abstract representation of physical space (like, eg, a game-engine style scene graph?) or does it just mean "this video generator is more coherent physically than other video generators"

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#85
post #8
post #6

The trouble is the lack of training available to these models compared to the ones like Seedance and Kling who seems to be tapping into their unlimited video inventory. Many models like LTX is technically good but when it comes to slightly different camera movements or the subject interacting with objects they struggle. For a recent example we had to use sample videos generated by closed source models and then use th…

I tend to think of these NV Labs models as architectural demos and ‘free razor blades’ — they’re more intended to inform internal R&D, get customers something that lets them do what they want quickly, and enhance the state of the art. In this case, what looks interesting is the one minute coherence and the massive speedup - they claim 36x over open models with similar capabilities. You can tell they aren’t aiming for…

Agreed the marketing angle. But beyond the marketing angle what seems to matter is the access to data - look at Seedance , various Kling models etc which are far ahead of others.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#86
post #16

I struggle with these world models from the perspective of video games (so this post is a particular perspective). I'm not a game developer myself, but some of my favorite games carry a deep sense of intentionality. For instance, there is typically not a single item misplaced in a FromSoftware game (or, for instance, Lies of P -- more recently). Almost every object is placed intentionally. Games which lack this inten…

> Games which lack this intentionality often feel dead in contrast

Like for instance... Dwarf Fortress? Minecraft?

Generative AI is just another method to go procedural generation. Not necessarily a better way. Or you could even argue that procedural generation is a form of generative AI... But either way, there are games where the lack of intentionality is central to the appeal.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#87
post #61

Nice, now instead of just reading slop you'll soon be able to experience slop Worlds, in 3D! /s It's honestly impressive, on the surface. The visuals are gorgeous... but it's still empty. What makes a "World" a world is precisely it's coherency. It's not about how it looks but rather how it "works". The plants in an ecosystems are a certain way because of the available resources, all the way to forces like gravity. I…

ai is exciting because it shows us what really matters...

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#88
post #16

I struggle with these world models from the perspective of video games (so this post is a particular perspective). I'm not a game developer myself, but some of my favorite games carry a deep sense of intentionality. For instance, there is typically not a single item misplaced in a FromSoftware game (or, for instance, Lies of P -- more recently). Almost every object is placed intentionally. Games which lack this inten…

> Games which lack this intentionality often feel dead in contrast Like for instance... Dwarf Fortress? Minecraft? Generative AI is just another method to go procedural generation. Not necessarily a better way. Or you could even argue that procedural generation is a form of generative AI... But either way, there are games where the lack of intentionality is central to the appeal.

> Like for instance... Dwarf Fortress? Minecraft?

DF for some doesn't fit into this category for me. Minecraft feels dead to me, while many other games that utilize procedural generation are not.

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#89

Earlier quoted context omitted.

Which game would that be apart from Minecraft?

Dwarf fortress, no man's sky, elite dangerous, ... The combination of "many", "most popular", and "nothing" is overstating it by a wide margin but for example the majority of the vegetation in games as far back as oblivion was procedurally placed.

Battlefield 2 had procedural trees and terrain the year before. I think it more or less came with open world maps?

Re: SANA-WM, a 2.6B open-source world model for 1-minute 720p video

#90

Earlier quoted context omitted.

>We are able to create less satisfying, less humane experiences faster? Yes, exactly. Inundate the world with superficially plausible yet hollow content, including any desired themes. People who aren't very discerning won't complain; the others will be outmatched and find that 99/100 pieces are all noise and they will need to spend increasing amounts of time trying to find the 1, if they can. I think there are some g…

I think you also can create satisfying, same human experiences faster, but not as fast. But the dopamine descent require strong discipline to stop there, and most don't .

[dead]
Post reply on HN