Hacker News new | past | comments | ask | show | jobs | submit login

I like this example because I tried making a greeting card "by hand" and it of course sucked ass. (As I have zero idea how to use Photoshop despite religiously reading the tutorials 20 years ago on a community site :D And, of course, I used PhotoPea.)

So by spending the same 20 minutes fiddling with some kind of graphics if the outcome becomes 10x better then I see that as an advantage. (That said I haven't tried this.)

Basically this is "exactly" that kind of capability that we already seen in "artist paints kid's drawing" [0] just commoditized. And this is where I (as a user) would be looking for the style transfer. Because I saw something famous/trendy/fancy/aesthetically-pleasing and now want to imitate it as a gag for this hypothetical birthday card. (Sure, copyright already doesn't care if I download someone's famous photo, and cut my friends head out from a photo I made, and put the head part on the copyrighted photo. Yet it was a straightforward derivative work. Just it's not commercial, damages are none, or even negative, ie. the artist benefits from that famous exposure! :D)

> legit benefits of this tech

I wanted to try to find an example that could apply to a lot of users. Because as awesome as using GenAI to generate Lean proofs (and then add a feedback loop by actually running the generated code through Lean) to solve Math Olympiad problems [1], it's not really an everyday thing.

> we evolve to where we push a button and get a birthday present. the logical conclusion is a system that does not even require to push a button.

Well, maybe! In no time we'll board the Axios and just chill. [2] But at the same time crafts and DIY and experiences (tourism, festivals, concerts) are ridiculously popular. (And it has its own problems. [3])

[0] https://www.youtube.com/watch?v=dB-Q0eNsUaQ KID'S ART Redrawn by a PROFESSIONAL ARTIST! - Ep.6

[1] https://news.ycombinator.com/item?id=41069829

[2] https://www.thelist.com/img/gallery/things-only-adults-notic...

[3] https://edition.cnn.com/2024/07/08/travel/barcelona-tourism-...




> Well, maybe! In no time we'll board the Axios and just chill.

This was sarcasm. That system would never happen. If birthday card requires no effort it is worse than nothing, literal noise.

> So by spending the same 20 minutes fiddling with some kind of graphics if the outcome becomes 10x better then I see that as an advantage. (That said I haven't tried this.)

Feels like you misunderstood. It is not "10x better" or "advantage" because literally the easier it was for you the less it is valuable.


> It is not "10x better" or "advantage" because literally the easier it was for you the less it is valuable.

... why? The closer I can approximate my imagined output (in a fixed time) the more value I see in it. The tool provides the added value.

We have better cameras than 30 years ago. Better pictures. It's easier to make a better photo in the same amount of time. I would be happier if my old blurry photos would be better.


the point it how difficult it is for you. the easier the present the less it is valuable. now you backtrack and mention that you have imagination and vision, and you spend effort trying to bring it to life, sure. but the original sentence was to just "ask some GenAI tool to make them a nice picture for their friend's birthday". which of course can be automated with a button.


You interpreted that part to mean "no customization", I meant as a tool for customization.

Value is relative, of course, that's why exchange/trade works, right? And the same applies here too, with the added complications of explicitly "unknowable price", after all I don't know apriori how much value my friend will assign to the card!

In general there are many situations that map closer to your understanding than mine, but I think you are dismissing also a lot of perfectly valid ones where your model simply doesn't apply.

For example if said friend values effort, but "apparent concept vs. actual results fidelity" is not part of effort for them.

Yet still we don't know if they value effort as in time spent (because if they value time spent then it's the same value in both of my scenarios) or if they value some "visible handcraftedness" (effort to produce something that I am known to be very amateur at). And even then the question arises, isn't picking the best tool for the job itself effort?

Of course if this friend simply abbors GenAI shit then they will think it's "low effort" even if I put hours and hours into it (that I wouldn't have put into a traditional image manipulation software, because there I realized soon that I would need months of learning, etc.)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: