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

Sora: Creating video from text

openai.com

Post reply on HN
Print view View original

241–250 of 1001 posts

Re: Sora: Creating video from text

#241

Post by macrolime »

thepasswordis wrote: post #100

OpenAI demonstrating the size of their moat. How many multi-million-dollar funded startups did this just absolutely obsolete? This is so, so, so much better than every other generative video AI we've seen. Most of those were basically a still image with a very slowly moving background. This is not that. Sam is probably going to get his $7T if he keeps this up, and when he does everybody else will be locked out foreve…

>100% they would pay a lot of money to be able to hang out with Joe Rogan, or some only fans person, and those pornstars or podcasts hosts will never disagree with them, never get mad at them, never get bored of them, never thing they're a loser, etc.

All of these things are against the terms of service and attempting them may result in a ban.

Re: Sora: Creating video from text

#242

Post by matsemann »

break_the_bank wrote: post #160

In less than a few hours Gemini 1.5 is old news. Sam is doing live demos on Twitter while Google just released a blog. Didn't think Google would be the first of the Facebook, Apple, Google and Microsoft to get disrupted.

Did they have this ready to go to upstage whatever Google would release? Or just coincidental both things announced today?

Re: Sora: Creating video from text

#243

Post by partiallypro »

These are insanely good, but there are still some things that just give them away (which is good, imo.) Like the Tokyo video is amazing, the reflections, etc are all great, but the gaits of people in the background and how fast they are moving is clearly off. It sticks out once you notice it. These things will obviously improve as time marches on.

The fear I have has less to do with these taking jobs, but in that eventually this is just going to be used by a foreign actor and no one is going to know what is real anymore. This already exists in new stories, now imagine that with actual AI videos that are near indistinguishable from reality. It could get really bad. Have an insane conspiracy theory? Well, now you can have your belief validated by a completely fictional AI generated video that even the most trained eyes have trouble debunking.

The jobs thing is also a concern, because if you have a bunch of idle hands that suddenly aren't sure what to believe or just believe lies, it can quickly turn into mass political violence. Don't be naive to think this isn't already being thought of by various national security services and militaries. We're already on the precipice of it, this could eventually be a good shove down the hill.

Re: Sora: Creating video from text

#244

Post by Eisenstein »

dheera wrote: post #162
thepasswordis wrote: post #100

OpenAI demonstrating the size of their moat. How many multi-million-dollar funded startups did this just absolutely obsolete? This is so, so, so much better than every other generative video AI we've seen. Most of those were basically a still image with a very slowly moving background. This is not that. Sam is probably going to get his $7T if he keeps this up, and when he does everybody else will be locked out foreve…

OpenAI's moat is (a) talent (b) access to compute (c) no fear of using whatever data they can get. On the other hand, I think these moats will be destroyed as soon as anyone finds a drastically more efficient (compute- and data-wise) way to train LLMs. Biology would suggest that it doesn't take $100 million worth of GPUs and exaflops of compute to achieve the intelligence of a human. (Of course it is possible that at…

> Biology would suggest that it doesn't take $100 million worth of GPUs and exaflops of compute to achieve the intelligence of a human.

Biology suggests that a self-replicating machine can exist by ingesting other machines, turning them into energy and then using that energy to power themselves. Biology suggests that these machines can be so small that we cannot even see them.

How close are we to making one of those?

Re: Sora: Creating video from text

#246

Post by visarga »

jijijijij wrote: post #237
thepasswordis wrote: post #100

OpenAI demonstrating the size of their moat. How many multi-million-dollar funded startups did this just absolutely obsolete? This is so, so, so much better than every other generative video AI we've seen. Most of those were basically a still image with a very slowly moving background. This is not that. Sam is probably going to get his $7T if he keeps this up, and when he does everybody else will be locked out foreve…

I predict, this "AI" content generation will eat itself at last. It will outcompete the low-effort "content" industry as is. Then inevitably completely devalue this sort of "product". Because it will never get to 100% of the real thing, the "AI" content craze will ultimately implode. I bet we won't get AGI as a progression of this very technology. The impression of "usefulness" will end when "AI" is starting to drink…

Funny you chose the day of a huge leap in generative video to proclaim generative model limitations.

Re: Sora: Creating video from text

#247

Post by treprinum »

thepasswordis wrote: post #100

OpenAI demonstrating the size of their moat. How many multi-million-dollar funded startups did this just absolutely obsolete? This is so, so, so much better than every other generative video AI we've seen. Most of those were basically a still image with a very slowly moving background. This is not that. Sam is probably going to get his $7T if he keeps this up, and when he does everybody else will be locked out foreve…

It's going to take a while to make this realtime as you suggest. The lower the latency, the more $$$ it costs (exponentially).

Re: Sora: Creating video from text

#248

Post by golol »

idiliv wrote: post #141

People here seem mostly impressed by the high resolution of these examples. Based on my experience doing research on Stable Diffusion, scaling up the resolution is the conceptually easy part that only requires larger models and more high-resolution training data. The hard part is semantic alignment with the prompt. Attempts to scale Stable Diffusion, like SDXL, have resulted only in marginally better prompt understan…

The real advancement is the consistency of character, scene, and movement!

Re: Sora: Creating video from text

#249

Post by blibble »

drdaeman wrote: post #86
treesciencebot wrote: post #14

This is leaps and bounds beyond anything out there, including both public models like SVD 1.1 and Pika Labs' / Runway's models. Incredible.

Let's hold our breath. Those are specifically crafted hand-picked good videos, where there wasn't any requirement but "write a generic prompt and pick something that looks good", with no particular requirements. Which is very different from the actual process where you have a very specific idea and want the machine to make it happen. DALL-E presentation also looked cool and everyone was stoked about it. Now that we k…

they're not fantastic either if you pay close attention

there are mini-people in the 2060s market and in the cat one an extra paw comes out of nowhere

Re: Sora: Creating video from text

#250

Post by tehsauce »

It's fascinating that it can model so much of the subtle dynamics, structure, and appearance of the world in photorealistic detail, and still have a relatively poor model of things like object permanence:

https://cdn.openai.com/sora/videos/puppy-cloning.mp4

Perhaps there are particular aspects of our world that the human mind has evolved to hyperfocus on.

Will we figure out an easy way make these models match humans in those areas? Let's hope it takes some time.

Post reply on HN