Hacker News Reader

I think the OP wasn’t disappointed about the end result, which is now higher quality by their own admission. They can certainly use their skills and experience to engineer prompts that will return the highest quality work, no doubt.

What made them sad is they no longer have the joy and satisfaction of developing those models in a 3D rendering software. This is what they trained for.

In many ways, this is like the initial trauma of becoming a manager. Learning to be content with the end result and the success of the team, than your direct individual contribution. The difference here is they didn’t ask for it.

I fully empathize with them. It makes me sad to think what they’re going through. My advise to them will be to look for other ways to get that satisfaction - either through a hobby or considering a different role/profession. For better or for worse, the world is changing rapidly and I can’t see it going back to what it was.

Some people really love drawing 2d sprites pixel by pixel. Now people make 3d models to render down to 2d sprites. Some people really love drawing textures. Now people use Substance to make them procedurally. Some people really love doing polygon modeling for making characters for media. Now people do it with sculpting. Some people really love and took pride in creating high quality meshes that animated well with clean topology. Now remesh and retopo tools can fix all of that after the fact. Some people really love hand animating things like explosions and smoke. Now Houdini simulates all of it for you.

Every major efficiency advancement in any field is going to eliminate work that some people loved and trained for. Someone trying to do CG work for a game or movie based on the skillsets they loved and practiced 20 years ago is going to be doing something completely different if they want to be competitive today, and be buried if they don't. Does the OP lament the fact that the workflow he is a part of eliminated work that other artists would have done before them, and very likely spent as much time and love mastering as they did the 3d modeling?

I empathize with them - I've had former careers peter out like web design - but they should also recognize that they are beneficiaries of the same sort of change that is currently upsetting them. It sucks, but it is how progress has happened for a long time now.

Yes, this is exactly how I am thinking about this. We are all going to become "managers". The drudgery of work is about to increase a whole lot.

I'm getting whiplash from the pace of change: I think of pixel art rapidly rendered from 3D models as the disruptive new technology that's devaluing traditional drawing skills!

> What made them sad is they no longer have the joy and satisfaction of developing those models in a 3D rendering software. This is what they trained for.

I think the point is that this artist can still experience that joy, but they just can’t get paid for it anymore. That is indeed sad, but that’s life.

On the other hand, maybe in a post scarcity society, this artist won’t have to work to survive, and they can do all the 3D modeling they want without having to worry about rent and food.

> On the other hand, maybe in a post scarcity society, this artist won’t have to work to survive, and they can do all the 3D modeling they want without having to worry about rent and food.

The scary part to me is that there’s scant evidence that a post-scarcity society will be allowed to develop. It seems more likely that those at the top will reap the entirety of the fruits of productivity gains enabled by employing AI, much as they’ve been doing for the more incremental (but still massive) productivity gains of the past several decades. Very little of the pie will be shared with the working class.

I think the point is missed here. the post was made by an artist who really likes his craft, and now with the generated models coming to the picture he feels like his work diminished to simply polishing after them.

I am a coder who really likes coding, getting results quicker will not make me happier.

it is different in a work environment, but still, a lot of people simply enjoy the path and not only the outcome.

It's the same with every person who shows up with AI music composition. "Finally, you can be free from the tedium of composing music and get to the fun parts!"

But that is one of the fun parts...

as a hobby music producer, this thing make me sick :)

actually I think this problem started earlier, with many _musicians_ using pre made beats and auto tune. right now it just takes it to a whole other level.

music should be a tool of expression, not one to gain fame & social acceptance

> music should be a tool of expression, not one to gain fame & social acceptance

I don’t get it then why you care, let alone care enough to make you feel sick. Nobody is stopping you from using music as a tool of expression. You do you, yam in your basement or wherever you want. I bet that you will find like minded people who appreciate what you do, but even if not you yourself are saying “music should not be a tool to gain fame and social acceptance”.

So why do you care what other people do?

I guess I over exaggerated due to fast typing. it does not _make me sick_ I just don't understand the reason.

a bit of context: I've been producing music for ~15 years now, purely as a hobby (due to incompetence on my part to up a level).

I have friends which I helped and collaborated with for years on and off, and some of them seem to me like they are stuck in a loop of wanting to succeed without actually expressing themselves. they seem to be stuck in the mode of _this is how it's done, and this is what I want and anything else is incorrect_.

even when there are many other options to make it sound better and make it richer.

so correction: what makes me _sick_ is the fact that some people try to copy other successful musicians without even considering doing something original, purely because they want the same level of _fame and success_.

There are still masses of people making music without “pre made beats and auto tune” they just don’t get played in places like the radio because the radio is mostly for sanitized stuff that appeals to mainstream trends. I don’t expect most of the artists I listen to because an AI can produce a generic trap beat or something. Of course, there may be more garbage to sift through.

This seems congruent with a thought I've had, which is that generative image models might enable people commissioning art to be more coherent about what they want. I only have FOAF-tier knowledge of commissioning artwork, but my understanding is that a major source of frustration on all sides is that a lot of work is often wasted because someone commissioning an artist has only the faintest spark of a concept or doesn't know how to communicate it effectively. I imagine a fair number of professional artists would be receptive to something along the lines of "here are my prompts/models, here's the set of the outputs I have opinions about and here's what I do and don't like about them; please use your expertise to compose this into something that actually works".

The market of people who care enough about those aspects to pay an expert is, I'm guessing, much smaller than the market for people who will be fine with whatever they can come up with themselves.

How many of those people would actually have paid an expert vs. downloading some stock photo or artwork and going close enough?

If we can have an infinite variation of personalized stock photos, I'm betting that the "good enough" side will win.

I'm still not scared about my job since I provide specific solutions (C++ architectures) that an AI cannot produce (yet), but it's worrying for everyone and every kind of job at the same time.

The market will decide if those "good enough" people are right.

Perhaps the market is wrong. Just because something is profitable doesn't make it good.

If you don’t like the taste of the people who want to pay for art I suggest doing what is traditional and either (a) paying for it yourself, or (b) taking over or setting up the government funding for artists so that people who share your taste are in control.

My concern is that there would be no one to pay for it because they have been driven out of the market.

That doesn’t make sense. Buyers don’t get driven out of the market.

They do if they lose their jobs. Art isn't something you must buy. I like to buy art, but if I lost my job, it's one of the first things I'll be cutting back on. And the less people are buying, the more the remaining artists have to charge to keep being able to afford to do art, which just spirals more people out of the buyers market.

What is a recession or a depression?

Economic collapses are a thing.

Sorry, when I said "no one to pay for it" I meant "no one who could be paid to do it". In retrospect, my wording was ambiguous.

I’ve been fiddling with midjourney recently. There’s definitely a learning curve to it, but ironically gpt has been helpful for generating prompts. I expect the edge that artists have in this regard to erode further with time

> People with a good understanding of different styles of art and framing/angles/etc should be able to create the "right" prompts for a customer much much faster and get better results than people who don't know art.

When the Bing image creator released a couple of days ago, I used ChatGPT to craft a prompt for it.

> In one sentence describe the art style of impressionism

< The art style of impressionism is characterized by visible brush strokes, emphasis on the changing effects of light, and an emphasis on capturing the fleeting moment.

Then used that in my prompt for the creator:

> okinawa street on a rainy night, visible brush strokes, emphasis on the changing effects of light, and an emphasis on capturing the fleeting moment

The result: https://imgur.com/a/QdT1Gtk

This was the first shot. No prompt engineering, no parameters, nothing.

The point is you knew to use "impressionist" as a concept. ChatGPT probably wouldn't have helped you with that kernel. It's not clear whether GPT added any value at all in your use case as you just replaced "impressionist" with "{synonym of impressionist}". How would you repeat this for:

> okinawa street on a rainy night, in the style of fauvism

> okinawa street on a rainy night, in the style of constructivism

> okinawa street on a rainy night, in the style of De Stijl (Neoplasticism)

> okinawa street on a rainy night, in the style of later Expressionism, e.g. George Grosz and Otto Dix

If you weren't aware of these stylistic options at the time?

> The point is you knew to use "impressionist" as a concept.

I could have surfaced, or learned about it 20 seconds earlier from a different ChatGPT conversation (throwing vaguely related questions at GPT[1]), or something as old fashioned as a search engine. Maybe turning up the "novel" kernels of information that you won't get from LLMs without knowing to ask will be its own thing in the future. A few Pinterests for prompts are already out there if I'm not mistaken.

Great choice of art styles, by the way!

[1] https://sharegpt.com/c/jnpoQRv

Couldn’t you just have prompted:

> okinawa street on a rainy night. Impressionism style

If I just wanted results, I'm sure I could have. I just thought it was interesting to see if this more abstract description would map to results that would be considered impressionism.

>I have no idea who Charles Sheeler is. But if someone knows 1,500 different artists/photographers/genres/styles and the nuances of them, they can immediately select the right look for the client

Don't overthink it. The SD GUI I've seen people use has a button that appends a random artists name at the end of the prompt. You mash that button until you are satisfied with the style.

>People with a good understanding of different styles of art and framing/angles/etc should be able to create the "right" prompts for a customer much much faster and get better results than people who don't know art.

Strong disagree. It's a completely different discipline. One doesn't "know art", they know a specific flavor of it, and 3d modelling is not... sentence constructing.

As someone with no artistic knowledge or skill who has been playing around with this stuff, I totally agree. I have found a large part of the learning curve to be getting a sense of how to describe the things in the first place.

It does lower the barrier to entry significantly for anyone wanting to get into it, but it's still going to be the case where some people are demonstrably better at it than others.

Of course developers will just solve it by making to be UX such that it uses some genetic algorithm to help you iterate to the exact thing you want even if you don't know how to describe it.

The exponential part of this ride really feels like it's coming in to full view.

Yet, giving a prompt has a much lower barrier than actually being an artist.