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

Sora is here

openai.com

Post reply on HN
Print view View original

481–490 of 1001 posts

Re: Sora is here

#481

Post by fraXis »

pier25 wrote: post #469
pen2l wrote: post #80

Every day that passes I grow fonder of Google's decision to delay or otherwise keep a lot of this under the wraps. The other day I was scrolling down on YouTube shorts and a couple videos invoked an uncanny valley response from me (I think it was a clip of an unrealistically large snake covering some hut) which was somehow fascinating and strange and captivating, and then scrolling down a few more, again I saw someth…

I wish Google would allow me to remove the AI stuff from search results. 99% of the times it's either useless or wrong.

Add a -ai to the end of your Google search query. There are also browser extensions that stop the AI content from displaying. I use the one for Chrome called "Remove Google Search Generative AI".

Re: Sora is here

#482

Post by mdp2021 »

toasteros wrote: post #435
minimaxir wrote: post #137

Earlier quoted context omitted.

Deliberately wasting electricity isn't exactly a moral win.

Generative AI is a waste of electricity by definition.

> by definition

"Definition" does not mean "...plus your own assumptions".

The results are there. Optimal, no; somehow valuable, yes.

Re: Sora is here

#483

Post by kurthr »

LASR wrote: post #273
jerf wrote: post #124

Earlier quoted context omitted.

It just plain isn't possible if you mean a prompt the size of what most people have been using lately, in the couple hundred character range. By sheer information theory, the number of possible interpretations of "a zoom in on a happy dog catching a frisbee" means that you can not match a particular clip out of the set with just that much text. You will need vastly more content; information about the breed, informati…

What you are saying is totally correct. And this applies to language / code outputs as well. The number of times I’ve had engineers at my company type out 5 sentences and then expect a complete react webapp. But what I’ve found in practice is using LLMs to generate the prompt with low-effort human input (eg: thumbs up/down, multiple-choice etc) is quite useful. It generates walls of text, but with metaprompting, that…

I'm not sure, but I think you're saying what I'm thinking.

Stick the video you want to replicate into -o1 and ask for a descriptive prompt to generate a video with the same style and content. Take that prompt and put it into Sora. Iterate with human and o1 generated critical responses.

I suspect you can get close pretty quickly, but I don't know the cost. I'm also suspicious that they might have put in "safeguards" to prevent some high profile/embarrassing rip-offs.

Re: Sora is here

#484

Post by nlehuen »

observationist wrote: post #461
miohtama wrote: post #437

Earlier quoted context omitted.

Users, not tools, should be judged. It is unlikely anyone is going to perform act of terrorism with this, or any kind of deep fakes that buy Easter European elections. The worst outcome is likely teens having a laugh.

Funny how all the negative uses to which something like this might be put are regulated or criminalized already - if you try to scam someone, commit libel or defamation, attempt widespread fraud, or any of a million nefarious uses, you'll get fined, sued, or go to jail. Would you want Microsoft to claim they're responsible for the "safety" of what you write with Word? For the legality of the numbers you're punching i…

You are posting this under a pseudonym. If you did publish something horrific or illegal, it would have been the responsibility of this web site to either censor your content, and/or identify you when asked by authorities. Which do you prefer?

Re: Sora is here

#486

Post by AntiEgo »

miohtama wrote: post #437
jedberg wrote: post #414

> We’re introducing our video generation technology now to give society time to explore its possibilities and co-develop norms and safeguards that ensure it’s used responsibly as the field advances. That's an interesting way of saying "we're probably gonna miss some stuff in our safety tools, so hopefully society picks up the slack for us". :)

Users, not tools, should be judged. It is unlikely anyone is going to perform act of terrorism with this, or any kind of deep fakes that buy Easter European elections. The worst outcome is likely teens having a laugh.

"Teens having a laugh" can escalate quickly to, "... at someone else's expense," and this distinction is EXACTLY the sort of subtlety an algorithm can't filter.

This does not need to become a thread about bullying and self harm, but it should be recognized that this example is not benign or victimless.

This genie is out of the bottle, let us hope that laws about users are enough when the tools evolve faster than legislative response.

[edit:spelling]

Re: Sora is here

#487

Post by minimaxir »

artemisart wrote: post #464
amelius wrote: post #410

Earlier quoted context omitted.

Can't you just give it a photo of a dog, and then say "use this dog in this or that scene"?

Yes, the idea works and was explored with dreambooth/textual inversion for image diffusion models. https://dreambooth.github.io/ https://textual-inversion.github.io/

Both of those are of course out of date and require significant training instead of just feeding it a single image.

InstantID (https://replicate.com/zsxkib/instant-id) fixes that issue.

Re: Sora is here

#488

Post by amzn-throw »

tom1337 wrote: post #190
MyFirstSass wrote: post #61

Wow this is bad. And by bad i mean worse than leading open source and existing alternatives. Is it me or does it seem like OpenAI revolutionized with both chatGPT and Sora, but they've completely hit the ceiling? Honestly a bit surprised it happened so fast!

Same goes with DALLE. It was cool to try it the first week or so but now the output is so much worse than Midjourney and stable diffusion. For me it can’t even generate straight lines and everything looks comic-ish.

To me this is just a simple artifact of size & attention.

Another example of this is stuff like Bluesky. There's a lot of reasons to hate Twitter/X, but people going "Wow, Bluesky is so amazing, there's no ads and it's so much less toxic!" aren't complimenting Bluesky, they're just noting that it's smaller, has less attention, and so they don't have ads or the toxic masses YET.

GenAI image generation is an obvious vector for all sorts of problems, from copyrighted material, to real life people, to porn, and so on. OpenAI and Google have to be extraordinarily strict about this due to all the attention on them, and so end up locking down artistic expression dramatically.

Midjourney and Stable Diffision may have equal stature amongst tech people, but in the public sphere they're unknowns. So they can get away with more risk.

Re: Sora is here

#489

Post by gus_massa »

8note wrote: post #478
observationist wrote: post #461

Earlier quoted context omitted.

Funny how all the negative uses to which something like this might be put are regulated or criminalized already - if you try to scam someone, commit libel or defamation, attempt widespread fraud, or any of a million nefarious uses, you'll get fined, sued, or go to jail. Would you want Microsoft to claim they're responsible for the "safety" of what you write with Word? For the legality of the numbers you're punching i…

that works for locally hosted models, but if its as a service, openai is publishing those verboten works to you, the person who requested it. even if it is a local model, if you trained a model to spew nazi propaganda, youre still publishing nazi propaganda to the people who then go use it to make propaganda. its just very summarized propaganda

Does this apply to the spell checker in Office 365 or Google Docs?

Re: Sora is here

#490

Post by makestuff »

pen2l wrote: post #80

Every day that passes I grow fonder of Google's decision to delay or otherwise keep a lot of this under the wraps. The other day I was scrolling down on YouTube shorts and a couple videos invoked an uncanny valley response from me (I think it was a clip of an unrealistically large snake covering some hut) which was somehow fascinating and strange and captivating, and then scrolling down a few more, again I saw someth…

I don't even know if this will be possible, or how it would work, but it seems like the next iteration of social media will be based on some verification that the user is not using AI or is a bot. Currently they are all incentivized to not stop bot activity because it increases user counts, ad revenue, etc.

Maybe the model is you have to pay per account to use it, or maybe the model will be something else.

I doubt this will make everyone just go back to primarily communicating in person/via voice servers but that is a possibility.

Post reply on HN