Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

nvlabs.github.io

Post reply on HN
Print view View original

111–120 of 162 posts

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

#111

Post by whilenot-dev »

stephen_cagle wrote: post #99
mccoyb wrote: 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 suspect these models will be like old Gutenberg's printing press. A rapid rise in the amount of content; most of it not that great. However the sheer volume will result in even more high quality content actually being created in aggregate. Put another way, the average game quality will go down, but the actual rate of "Great" games will go up.

Have to disagree here as I don't subscribe to your analogy. GenAI can be considered a tool, yes, but it's less a "circular saw for workshops"-tool, and more a "microwave for kitchens"-tool... and I doubt microwaves led to higher quality content in aggregate.

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

#112

Post by marc_g »

resist_futility wrote: post #73

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

My 70mbps connection couldn't even buffer the videos so I gave up trying to watch them. It didn't even look that high quality...

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

#113

Post by vessenes »

bobkb wrote: post #85
vessenes wrote: post #8

Earlier quoted context omitted.

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.

It’s hard not to believe that Google doesn’t have an amazing model in-house with all that Youtube content available. But, agreed the Chinese models seem best in the last year or so, and agreed an open policy on training data def makes for better quality

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

#115

Post by Keyframe »

mccoyb wrote: 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…

You touched upon the substance of what we're witnessing et large with AI. Stuff feels hollow and worthless even though it looks amazing. Graphics, images, video, music, text, code..

So, what's the deal?

As with ANY work in life, the quality of the result is a direct reflection of care and intention behind it. Simplified, it's a reflection of how much _you_ put effort in it. It always shows. Even in AI day and age. It's just that path to a result (without effort) is now way shorter so volume is showing up and diluting the overall impression. The latter kind of cheapens every field it touches, so even more effort will have to be put in to show up on the radar.

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

#116

Post by fredoliveira »

debugnik wrote: post #80
robot_jesus wrote: post #45

Earlier quoted context omitted.

No Man's Sky, Terraria, Dead Cells, to name a few.

Dead Cells just arranges a few pre-designed rooms together for each stage, doesn't it?

If it does do that, it doesn't feel that way. I never found it particularly repetitive.

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

#117

Post by Azantys »

ctoth wrote: post #59
unfitted2545 wrote: post #39

Earlier quoted context omitted.

If we use world models to train AI systems, are we not essentially forcing something to live so it can gather data for us? Yes, we haven't gone that far with creating consciousness yet, but there is gonna be a lot of money around neural computing devices for consumers in the coming decades, so that will speed up knowing what sense data you need for consciousness.

> are we not essentially forcing something to live so it can gather data for us? Wait until you learn about what we do to chickens.

The Matrix is imminent

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

#118

Post by ssl-3 »

harshreality wrote: post #79
resist_futility wrote: post #73

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'…

It's really something. It was using ~400Mbps of my mostly-idle 500Mbps connection (slow for some folks, but pretty speedy in my particular ghetto).

It appears that there are 62 videos on the page. They're generally 16fps and 60s long. All are h.264, 1280x704. The median bitrate is 4.962 Mbps.

I don't know enough about JS to try to understand WTF it is doing, but there's only 1.3 GB of video on that page. At a transfer speed of 400Mbps, the whole mess of them should be downloadable in around 30 seconds.

But it wasn't behaving that way at all. It instead behaved as an excellent bandwidth-waster.

(Woe to those who click this link on metered connection, I guess.)

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

#119

Post by nl »

jubilanti wrote: post #53

Model weights coming "soon" == currently vaporware. So the weights aren't even open, how can this be "open-source"? Everyone is right to be skeptical of this coming from a 2.8B model. Weights or it didn't happen.

The model is out here: https://huggingface.co/Efficient-Large-Model/SANA-Video_2B_7...

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

#120

Post by stephen_cagle »

whilenot-dev wrote: post #111
stephen_cagle wrote: post #99

Earlier quoted context omitted.

I suspect these models will be like old Gutenberg's printing press. A rapid rise in the amount of content; most of it not that great. However the sheer volume will result in even more high quality content actually being created in aggregate. Put another way, the average game quality will go down, but the actual rate of "Great" games will go up.

Have to disagree here as I don't subscribe to your analogy. GenAI can be considered a tool, yes, but it's less a "circular saw for workshops"-tool, and more a "microwave for kitchens"-tool... and I doubt microwaves led to higher quality content in aggregate.

I feel like a circular saw is a creator tool, while a microwave is mostly a finishing tool.

I take raw material and make something out of it with a circular saw, largely unrestrained by anything other than cost, skill, and material.

With a microwave, I make things hot so I can eat them.

Aside: Also, I wonder why that is? Why do we regard the microwave as "degenerate" compared to the oven? Why is baking seen as a calling while microwaving is, well, not? Is it the ease of the microwave makes the effort less impressive? Maybe it is that you can't achieve certain effects like browning? Is it because of it's 1970's association with "radiation" and tv dinners? Is it just cultural inertia?

Post reply on HN