Hacker News Reader

Comment by quantumHazer | Hacker News Reader

Like most AI image or video generation tools, they produce results that look good at first glance, but the more you watch, the more flaws and sloppiness you notice, and they really lack storytelling

View on HN

They don't have to be as good as the best film production team - they just need to be better than the average/B-grade ones to gain adoption.

With the media & entertainment hungry world which is about to get worse with the unempoyed/underemployed tiktok generation needing "content", something like this has to have a play.

Or replace cutscenes in video games, or short videos on ads on mobile (already small and people are barely paying attention)

Nowadays when I randomly open a news website to read some article, at the bottom of the page all the generic "hack to lose your belly" or "doctors recommend weird japanese device" or "how seniors can fly business class", I've been noticing lately 1/3rd of the images seem to be AI generated...

God, what a dismal future we're building.

Warning: browsing the Web without an ad blocker is hazardous to your mental health. If you regularly see ads permeating most web pages, and don't know how to avoid that, you may need to see a specialist.

Perhaps some of us choose not to rip off payments that are due to the people who provide a service that we're using. If I visit a website that's so infested with ads that I don't like being there, it's on me to stop visiting that website.

I simply don't think it's fair to cheat service providers when we don't like their service. You have a choice, and that choice is to not use that service at all. They're providing it under the terms that it is ad-supported. If you don't want to support it, but you still want to use it, then you're cheating someone. That is dishonest and unethical.

If ad-supported sites would return an optional response header indicating such, and their opinion on adblocking visitors. E.g.:

  Advertisement-Permission: [required|requested]

And my adblockers had a config option to abort pageloads with an appropriate error message, if `required` or `requested`, then I would use it happily.

In the meantime, I'm browsing every site with all content blockers set at maximum, because any other choice is incomprehensible on the modern web.

If I consequently visit some sites that want me to consume advertising of which I am unaware, then that is entirely their issue, not mine.

Jon Stewart did a gag in this week's episode where there were happy meal toy versions of a bunch of congressmen on screen for a few seconds.

A lot of content is like this - you just need an approximation to sell an idea, not a perfect reproduction. Makes way more sense to have AI generate you a quick image for a sight gag than to have someone spend all day trying to comp it by hand. And as AI imagery gets more exposure in these sort of scenarios, more people will be accustomed to it, and they'll be more forgiving of its faults.

The bar for "good enough" is gonna get a lot lower as the cost of producing it comes way down with AI.

But you don't have to outsource 100% of your creative work to your tools. This is a toolbox, not a complete automatic masterpiece generator. If you want serious production, don't remove yourself from the loop.

Drive the storytelling, consult with AI on improving things and exploring variations.

Generate visuals, then adjust / edit / postprocess them to your liking. Feed the machine your drawings and specific graphic ideas, not just vague words.

Use generated voices where they work well, record real humans where you need specific performance. Blend these approaches by altering the voice in a recording.

All these tools just allow you to produce things faster, or produce things at all such that would be too costly to shoot in real life.

AI used to be quite bad at coding just - what - 2 years ago?

Now it's "good enough" for a lot of cases (and the pace of improvement is astounding).

AI is still not great at image gen and video gen, but the pace of improvement is impressive.

I'm skeptical image, video, and sound gen are "too difficult" for AI to get "good enough" at for many use cases within the next 5 years.

it's still quite bad though

I don’t get how someone can look at these videos and think “wow there’s lots of flaws and it’s sloppy and no storytelling” rather than “holy smokes this stuff is improving fast!”

In 2 years we have moved from AI video being mostly a pipe dream to some incredible clips! It’s not what this is like now, but what will it be like in 10 years!

In the era of more software advancements, people have perceived the progress of tech to be incremental more than the opposite. I think even tech bigwigs of the past have said extrapolation is a futile endeavor because technology is unpredictable.

Bill Gates saw computers and extrapolated that there would be one in every home, Gordon Moore produced Moores Law which literally extrapolated compute in the 60s and almost still holding, and you can extrapolate that GPT-5 will perform better than GPT-4, and that there will probably be a GPT-6 that will be even better than that.

Extrapolating that technology will get better in the future when it has got better in the past isn’t a sure bet, but it’s a reasonably reliable one.

No, past performances are not an indicator of future performance. Sometimes it happens to be, but many other times it does not.

I think this is a strange take - Technological advancements aren't mutual funds.

Of course it's an indicator of future performance - Not a guarantee, but certainly a indictator.

I think what I genuinely have an issue with is that extrapolation ignores the finer details of research work being taken to understand the current issues that may affect the future goals of a technology. It basically just ignores the science of it in favor of "well we've seen this level work before". The thing is, sometimes the more these experts learn about the current goal to beat in their product, the more things previously unknown and unplanned for present themselves as new issues. Extrapolating is a bit past-facing when you see it this way, because those new problems are entirely new challenges that haven't been considered in the equation and may require an entirely new approach. Every success is different, every failure, and every challenge is different. So extrapolating seems pointless.

Yeah, no. That's not my point. Saying something will get better in /any/ amount isn't unreasonable. That's being positive towards innovation and it's harmless. That's not exactly what the modern tech consumer is being made to expect nowadays. It's always the next iPhone moment. The next GPT. Always a huge leap. Like if we landed on the moon and we've been promised we'll have colonies outside the solar system by 2040

You don’t even have to look close for some of these. The owl suddenly flipping direction in the first video was jarring

Yeah the owl ""animation"" is terrible, I bet they could have found better examples? If it wasn't the case I don't know what to think

It looked to me like the owl was turning around to land.

When it's in silhouette you don't know what direction it is facing, technically. I think what's happening is when you see a shot of something flying in front of something prominent (in this case, the moon), your brain naturally perceives it is going away from the camera and toward the object.

I think that's just the silhouette illusion[1]. In this case likely abetted by the framing elements moving near the edges.

[1] - https://en.wikipedia.org/wiki/Spinning_dancer

It doesn't flip, it's an illusion. The owl is always facing the camera.

Yea, but we're early days and I think that will go away as the tools get better. Also - did you watch the sample short films they have?

You should see the terrible results it's possible to generate with AfterEffects, Blender, Houdini, etc..

Reply on news.ycombinator.com