Hacker News Reader

Comment by jermaustin1 | Hacker News Reader

Having worked with a lot of artists, they are seeing stable diffusion and other image generators as a way to generate lots of ideas. A base they would then ,usually starting from scratch, build their end product off of.

So even though to a casual glance, these images look amazing, if you look closely, you see all the flaws that come with them being generated. Weird artifacts, a lack of symmetry that humans usually add to their creations.

These flaws would not exist (to such a degree) when a professional artist paints the same scene.

View on HN

Yes - this will make artists hyper productive IMO; and will make a lot more people willing to make art! If you've ever had a little ditty and wondered what it would sound like as a violin concerto in D minor, now you can know! If you ever wanted your wife painted as a Rembrandt, there you have it. If your five year old comes up with an idea for a fun video game called Laser Chicken, now his friends can play it.

Imagine pairing this with bespoke automated clothing output - take a photo of an outfit, verbally describe the changes you want to it ("a little more debonair, dark lapels, 1920's styling"), click a button, preview it as it would look on you in several recent pictures you took, and a week later your tailored suit arrives. Now for sneakers. Hats. Bags. Watches.

The 20th century was about mass production to ensure everyone could have things: food, clothing, transport, entertainment. The 21st century may turn to expression: allowing each person to express themselves however they want in their goods and services. Or just following along to buy whatever your favorite tastemakers recommend. However involved you want to be!

The world is about to get a lot weirder and more interesting.

Making artists hyper productive will cheapen even further their output. If one artist can do the work that you are currently paying 8 to do, you only need to pay one artist that can wrangle these tools.

I would agree that this is like 20th century mass production. To be clear, I don't necessarily think that mass production is a good thing either. In fact, it has been probably the most detrimental thing to our environment that humans have ever done.

They cheapen the entry fees for learning, but the expectation will keep getting higher. It's like games where, even for indies, they won't accept N64 quality these days. Text to speech is quite good now, but people prefer real narrators. Everyone can create music with a cheap laptop and some midi keyboards, but only those really talented will make it.

Some very popular games are much less polished than N64/SNES quality.

Mobile, Indy, and Retro are all very popular, just look at what people are playing on Twitch.

> If one artist can do the work that you are currently paying 8 to do, you only need to pay one artist that can wrangle these tools.

And maybe the number of places paying for art will grow by a factor of 8.

Maybe, but it hasn’t worked that way for software. I think people might see the opportunity to inject bespoke art in a lot of places it wasn’t previously. College students who could only afford movie posters previously will have art commissioned, every building will have a mural, etc etc.

One of the many things that I loved about Lisbon was that art was hilariously pervasive. It felt like every vertical space was filled with beautiful and original art. The whole city was a gallery. It would be wonderful to see that in more places.

I thought Teddy K was dead. Maybe you're a bot trained on his works.

This concept was used in westworld [0]

[0]: https://youtu.be/2xm4feEKDjw

> These flaws would not exist (to such a degree) when a professional artist paints the same scene.

The flaws in AI generated art were 100x as obvious in systems like this only a few years ago. In 5 years I doubt anyone will be able to tell the difference between AI and human art.

When the photograph displaced most portrait painters, we invented a new type of artist - the photographer. I hope we’ll see the same thing here - artists who specialise in using stable diffusion (and friends) to make new art in a new way. This blog post is like one of the world’s first photographers saying “hey look how the photo changes when I move the subject relative to a light source!”. I can’t wait to see what results we get with deep expertise (and better algorithms).

How long before we have filmmakers using AI to cast, direct and shoot their films?

Maybe, maybe not. It didn't happen with CGI, the uncanny valley hasn't been surmounted, car chases aside.

The mulchers, as Bruce Sterling calls them, have a fresh meat problem. They've consumed all the words, and all the pictures, and we already know that feeding them their own mulch gives worse results.

We're not at the scale limit for data but we know where it is. It's not clear that refinements to the mulching process will create mulch good enough to tell apart from creation. It might. But it might not.

> with CGI, the uncanny valley hasn't been surmounted

I'm not so sure about this. While some scenes are still obviously using CGI, I think a lot of CGI in movies now passes unnoticed, even entirely digital characters.

We certainly notice when those characters do things humans can't do, of course, and when budget or schedule or both result in things being pushed out too early, but how would we know when digital characters look natural on-screen? We wouldn't!

I wonder if anyone is gonna take advantage of this point in time where the average person isn't aware of these breakthrough AI models, and sets up themself an account as a professional-grade artist on Fiverr, offering to draw highly detailed landscapes or whatever where the client can provide reference material and ideas.

You generate the image in 30 minutes (maybe less if you get the process down to a science), then wait around for a few weeks to keep up the illusion that you're actually doing the drawing by hand, and send it off to your satisfied client. You could be charging hundreds of dollars for your "artistic services," and have dozens of clients going on simultaneously.

real clients are going to want sketches and ideas so they can tweak them before final art.

Wonder if you could convert an image to a sketch

That makes the process a little less simple, but still easier than doing the real work

> Having worked with a lot of artists, they are seeing stable diffusion and other image generators as a way to generate lots of ideas.

But it also dillutes their own ideas. I know of painters painting AI generated stuff and that is likely a new genre but a lot of genuine artwork will lose interest on the market. It is what it is and don’t I love or hate it…

You don't just paint a better version of the generated images, you look at a bunch of generated images, and get a better idea of what you want to paint from that.

Say you know you want to do a portrait of a woman in armor, you can generate a dozen of those in around a minute on a 3090, look at the generated armors, the faces (usually all sorts of screwed up), and the composition. Its just a way to kickstart the creation process.

Yes, I get that but if everyone’s doing it it becomes a race to the bottom sort of thing… I hope I am wrong and things turn out in a completely different direction that I can’t see now

You say 'look at the generated armors' etc. but this does not mention what you use to look with: the artist's eye.

People so quickly assume that access to these tools will make everyone an artist, but the raw output is so lacking in a voice and intentionality. If you supply the voice and intentionality through your iterative process and a hybrid visual/text language playing the generator like a violin… you're playing the generator like a violin.

Your artistic skills have been translated to a wholly new set of vocabularies, and it's your eye that is tested most. Can you see/imagine better than the next guy?

The issue is an ever smaller percentage of the population can be successful at ever more difficult opportunities that result from ever better automation.

Manual labor still exists, but it’s a vastly smaller percentage of overall jobs. Productivity and automation seem like the same thing on the surface. However, the argument for a long tail of creators in an ever more wealth society breaks down when AI can start writing niche romance novels not just barely coherent news articles etc.

In theory we might have ever more new types of jobs, but automation isn’t just getting better it’s also getting faster.

Meanwhile I've gotten into a lot of very heated arguments with artists who fear this will actively destroy their livelihood. I lean towards seeing these developments as a massive positive for a society as a whole, but I find it hard to ignore that. Most of the flaws you see today will be reduced over the next few months, and people will get better of finding ways of working around them. It will cause substantial upheaval for a lot of people, including job losses especially on the low end. Maybe many, or even all, of those job losses will be compensated with additional jobs elsewhere, but we don't know.

I would argue that the technology will soon be at a point where these flaws are much less visible and will then further cheapen the work done by a human.

Reply on news.ycombinator.com