Live data from Hacker News

Meta Movie Gen

ai.meta.com

451–460 of 1001 posts

Re: Meta Movie Gen

#451

All the vids have that instantly recognizable GenAI "sheen", for the lack of a better word. Also, I think the most obvious giveaway are all the micro-variations that happen along the edges, which give a fuzzy artifact.

That sheen looks (to me) like some of the filters that are used by people who copy videos from TV and movie and post them on (for example) facebook reels.

There's an entire pattern of reels that are basically just ripped-off-content with enough added noise to (I presume) avoid content detection filters. Then the comments have links to scam sites (but are labelled as "the IMDB page for this content").

Re: Meta Movie Gen

#452

Why do these video generation ones never become usable to the public. Is it just they had to create millions of videos and cherry pick only a handful of decent generations? Or is it just so expensive there's no business model for it? My mind instantly assumes it a money thing and they're just wanting to charge millions for it, therefore out of reach for the general public. But then with Meta's whole stance on open ai…

Consistency and continuity is the main problem. Take a look at the “Super Panavision” AI videos on YouTube.

Those videos are a good measure for monitoring AI video improvement.

Re: Meta Movie Gen

#453

Earlier quoted context omitted.

Creativity isn’t magic, it’s a skill. There is no creativity without the application of it. By definition creativity produces something. Without skills it’s not possible to produce anything. The act of creating teaches you to be better at creating, in that way and in that context. This is why people with practice and expertise (e.g., professional artists, like screenwriters and musicians) can reliably create new thin…

To an extent. Take cooking for example though- I don't doubt that writing recipes and trying them builds ones creative muscle, on the other hand, I don't think being we'd be at a loss for great chefs if we were to automate the cutting of onions, the poaching of eggs, and the stirring of risotto.

Of course we would. That’s my entire point.

Take poaching eggs for example. Let’s say you automate that 100% so as a human you never need to do it again. Well, how good are your omelettes then? It’s a similar activity — keeping eggs at the right temperature and agitation for the right amount of time. Every new thing you learn to do with eggs — poaching, scrambling, omelettes, soft-cooking for ramen — will teach you more about eggs and how to work with them.

So the more you automate your cooking with eggs the worse you get at all egg-related things. The KitchenBot-9000 poaches and scrambles perfect eggs, so why bother? And you lose the knowledge of how to do it, how to tell the 30-second difference between “not enough” and “too much.”

Re: Meta Movie Gen

#454
post #111

You know it’s going to happen: “I want a funny road trip movie staring Jim Carey and Chris Farley, based in Europe, in the fall, where they have to rescue their mom played by Lucille ball from making the mistake of marrying a character played by an older Steve Martin.” 10 minutes later your movie is generated. If you like it, you save it, share it, etc. You have a queue of movies shared by your friends that they like…

> Generation failed: I'm sorry, but I cannot generate videos of real people. Please try another prompt.

Also no violence, or alluding to conspiracies, or historical events.

Re: Meta Movie Gen

#455

Earlier quoted context omitted.

I don't see it as that much of a problem. It's like washing machines taking away people's job of washing clothes, what are they gonna do with their time now? Maybe something more productive.

… something more productive than art? that’s quite a productive thing. art has tremendous value to society. why don’t we automate the washing machine more instead of automating the artist?

Washing machines and roombas were the low hanging fruits in the real world.

Automating more in the real world is much (much) harder than grabbing the low-hanging fruits in the digital world.

Re: Meta Movie Gen

#456

I've been saying for years that generated content is an impending tsunami that's going to drown out all real human voices online. The internet may become effectively unusable as a result for anything other than entertainment.

This is interesting and i see some of this now. Even here on HN and other forums i thought were mostly "human". Even one of my group chats i can tell one of my friends is using ai responses, but one of the other members cant tell and replies earnestly. I am grossed out by this. my instinct is to avoid ai slop. The interesting part to me is: What next? Where do we go? Will it be that "human" forums are pushed further…

Cryptography-secured/signed generated content / interactions?

Re: Meta Movie Gen

#457
post #157

Why do these video generation ones never become usable to the public. Is it just they had to create millions of videos and cherry pick only a handful of decent generations? Or is it just so expensive there's no business model for it? My mind instantly assumes it a money thing and they're just wanting to charge millions for it, therefore out of reach for the general public. But then with Meta's whole stance on open ai…

I'd guess 1 in 10 model demos turn out to be useful product, at best. This and Sora are particularly annoying, though, for how they put together these huge flashy showcases like they're announcing some kind of product launch and then... nothing. Apparently there's value in just flexing your AI-making muscle now and then.

to be fair, Sora was one of the most mind blowing technology showcases of my life, and openai is successful at raising tons of money

Re: Meta Movie Gen

#458

Earlier quoted context omitted.

> It is totally possible right now to at least generate images with no discernible tell. I have yet to find examples of this

In the linked webpage, the following videos would be good enough to trick me: - The monkey in hotspring video, if not for its weird beard... - The koala video I would have mistaken for hollywood-quality studio CGI (although I would know it's not real because koalas don't surf... do they?) - The pumpkin video if played at 1/4 resolution and 2x speed - The dog-at-Versailles video edit If the videos are that good, I'm s…

> limited to a certain style of close-up portrait photography

Not to take away from your point but it's more limited than one might think from this phrase. As an exercise, open that page and scroll so the full image is on your screen, then hover your mouse cursor within the iris of one of the eyes, refresh and scroll again. (Edit: I just noticed there's a delayed refresh button on the page, so one can click that and then move their mouse over the eye to skip a full page refresh.) I've yet to see a case where my mouse cursor is not still in line with the iris of the next not-person.

Re: Meta Movie Gen

#459

My kids both have creative hearts, and they are terrified that A.I. will prevent them from earning a living through creativity. Very recently, I've had an alternate thought. We've spent decades improving the technology of entertainment, spending billions (trillions?) of dollars in the process. When A.I. can generate any entertainment you can imagine, we might start finding this kind of entertainment boring. Maybe, at…

Creativity is about having original ideas. So far, AI isn’t that good at that, and neither at maintaining a consistent idea throughout a production. Will AI be able to come up with a compelling novel series, music album, video game, movie or TV series in ten years? Possibly, but there’s also a good chance that it won’t.

Re: Meta Movie Gen

#460

Things are about to get weird. We can't control this at any level: At the level of image/video synthesis: Some leading companies have suggested they put watermarks in the content they create. Nice thought, but open source will always be an option, and people will always be able to build un-watermarked tools. At the level of law: You could attempt to pass a law banning image/video generation entirely, or those without…

I agree the cat is out of the bag, but GANs do not work like that. One of the common failure modes in training a GAN is that the discriminator gets too powerful too quickly and the generator then can no longer learn.

Hard to say anything is impossible off of one point - but discrimination afaik is generally seen as the easier problem of the two, given you only need to give a binary output as opposed to a continuous one.

Post reply on HN