Hacker News Reader

Comment by saguntum | Hacker News Reader

I wonder if they're going to license this to brands for heavily personalized advertisement. Imagine being able to see videos of yourself wearing clothes you're buying online before you actually place the order, instead of viewing them on a model.

If they got the generation "live" enough, imagine walking past a mirror in a department store and seeing yourself in different clothes.

Wild times.

View on HN

At that point, why even buy the clothes? Influencers will just post the video of the mockup on social media, which is the only reason they were considering it in the first place. Save themselves the foot fungus.

https://xcancel.com/Naija_PR/status/1904809073356251634

Then take the next step. Why even spend money going out? Generate a video of yourself with fake friends at a party and post that, while eating ice cream alone at home.

Now you're thinking with portals

Because food still tastes good whether or not it looks good. There are other sources of happiness than online validation.

Just wait for Musk's implent, it'll make ozempic pills taste like pizzas and burgers

I was criticising and making a joke prediction about the practice, not suggesting you actually do it.

I agree with you regarding online validation. I would even go so far as saying that depending on online validation or fame in general for happiness is unhealthy and anyone who does should make it a priority to find alternative sources.

Few years down the line:

"Five things you won't believe: We took an actual vacation"

>Why even spend money going out? Generate a video of yourself with fake friends at a party and post that, while eating ice cream alone at home.

Hey don't be giving away my JOMO secrets.

I'm fairly certain there is a scene in Minority Report just like this! Or at least, the advertisement says Tom Cruise's character's name.

https://en.wikipedia.org/wiki/Minority_Report_(film)

Here a clip of that scene: https://www.youtube.com/watch?v=7bXJ_obaiYQ

In 2023, Carvana ran an ad campaign that showed you a video of "your car" thanking you and talking about your time together:

https://adage.com/article/digital-marketing-ad-tech-news/car...

A little creepy, but very much in this vein.

We probably haven't even scratched the surface of what will be done with this tech. When video becomes "easy", "quick", "affordable", and "automatable" (something never before possible on any of those dimensions) - it enables countless new things to be done.

Its still just video though. Its not a new type of media. My guess is it will play out same as self publishing on amazon. Ultra specific generas that monitize the infinite long tail.

I feel like the main problem with buying clothes online is there is no way to tell if they are actually good or fit right. The photos are all fake where it's just an image projected on a stock photo of someone in a shirt. Doesn't tell you what the material is like, doesn't tell you if it actually fits (an AI video model is just making up the fit).

I don't have to imagine it because it's probably the most COMMON fantasy that people who work in advertisement and marketing have every day.

Now... take it a STEP further. Remember the scene in Futurama where Fry tries on the Lightspeed Briefs and looks in the mirror to see a rather aspirational version of himself?

https://www.youtube.com/watch?v=by0KQRJVFuk

Yeah.

its called Virtual Try On (VTO) and there are plenty of models going there for static gfx, it is very reasonable to expect soon emerge those for video VTO.

Accurate virtual try on however is quite difficult, and users will quickly learn to distrust platforms that just generate something that"looks right".

You can prompt with a normal size 8 dress and "kim jungle un wearing a dress" and it will show you something that doesn't help you understand whether that dress would fit or not. You can ask for a tube dress and it will usually give him a big bust to hold it up. It's not useful for the purpose of visualing fit.

It will definitely be used for such just like image models already are for cheap tenu clothes, and our onions shopping experience will get worse.

Maybe this needs purpose built models like vibe-net or maybe you cab train a general purpose model to do it, but if they were spending the effort necessary to do so they'd be calling it out.

You don't need generative AI for that at all, snapchat filters have existed for a decade and are the same concept. A lot of brands have already adopted that.

Or, on the genAI side, Google marketed this use case heavily for Flash Image 2.5 (even if that's not the same type of generative model because it's geared for editing, it's still in the taxonomy)

I'm surprised I had to ctrl-f this far for the first snapchat mention. Same. All I see here is snapchat except on any platform. Far from a tiktok competitor and far from revolutionary.

Seems like a nice feature but the most important aspect is "fit" and I wouldn't trust these models to do that accurately. They'll most likely make everything fit perfectly. Should be fixable tho.

When the dust settles , that's probably going to be the most common application of these video models. Making automated social content kind of defeats the purpose; people empathize with other people, not with AI . (I guess that's why they didn't also make their interview video via AI)

But Sora /VEO will probably also revolutionize movies and tv content

People said the exact same thing about AR furniture, and I'm 99% sure no one uses that.

It seems like 99% of apartment listings in the city of New York are virtually staged with AR furniture

The latter would feel like actual scifi to me.

Am I misremembering or didn't Meta announce few months ago that people will see their own faces in ads?

Reply on news.ycombinator.com