Hacker News Reader

Comment by freeone3000 | Hacker News Reader

That’s the point of generative AI, though, isn’t it? You put in text, you get out an image, no more need for discernment or skill or labor. No amount of explanation of vector math will change the fact that article headers can now be generated from the article, or you can say “give me a spaceman on a planet looking at twin moons” and a Steam game header is generated with some basic cropping. Incidental and corporate art, which is how a lot of professional artists make a living, is in real danger of being automated away.

View on HN

You should look into the more modern tooling available these days, like the Krita plugin and InvokeAI.

Sure, you still use a prompt in those, but it's not really the focus in many cases -- you just use it to give a very rough guideline of what you're trying to do, then start drawing.

Depending on approach you'd either start sketching your astronaut and see the AI turn the sketch into something polished in real time, or you'd generate an image then polish up the details. I think most artists are more likely to prefer the first approach.

>"You put in text, you get out an image, no more need for discernment or skill or labor."

This sentiment is reminiscent to the attitudes around photography when it first became practical. People wondered how photography could be considered a form of art when all one had to do was point and press a button. The amount of effort is minuscule compared to what it takes to depict the same subject using traditional media. That being said, there is a lot of skill necessary to capture a photograph properly; lighting, composition, shutter speed, exposure, and so on. I agree that generative AI images will disrupt the market segment for stock photos and "clip art" used for articles, presentations, et cetera.

There will be a need to study and acquire skills related to the use of generative AI image creation. While technology will make simple prompts "good enough" for most outputs, just as our smartphone cameras make taking a picture "good enough", people will still need to study and practice in order to make high quality output.

> That’s the point of generative AI, though, isn’t it? You put in text, you get out an image, no more need for discernment or skill or labor.

The best tools in this space are spatial editors and node editors. Text is weak sauce.

Also, you can't use text modalities for film and 3D. It has to be art tooling to get good results.

Artists will always have a job. They'll just be doing way more than they were before.

You still need a prompt, but the spatial tools and controlnet models are where the magic is.

I think style transfer + pose detection + edge detection gives artists a bunch of new tools to play with.

OP I would look into learning comfyui

We will always need to work in the fields. With agricultural machinery we'll be doing way more than we were before.

if the amount of new offer is not paired by demand, there will be pain.

The fact that this comment is gray says everything that you need to know about this entire comment thread. A tiny slice of people that know what they're talking about and a giant stack of people drunk on dunning-krueger-fueled overconfidence in their understanding of art, on philosophical, practical, and professional levels thinking they've got the authoritative word because it's just so dang loud in the echo chamber. The tendency for technical people to think they can outmaneuver subject matter experts on their own terf with a few a priori thought experiments has always annoyed the hell out of me, but I can confidently say there has never been a more frustrating time to be both a professional artist and software developer.

Reply on news.ycombinator.com