Kind of like everybody does… Even if they think a little too highly of themselves…
This is pretty much the best one sentence summary of the human condition I have seen.
3 Likes
whoa, that is some harsh statement here, did you go through all of it? I mean this really deeply depends on the way you guide the AI by prompt skills…
This does not apply to anything in VFX, 3D art, Generative Art and so many other genres in art. I mean: Doing a Houdini Render is closer to studying higher mathematics, physics and biology, the aesthetic choices are based on reference images etc.
And I think that we should really be patient with judging these things now. The difference between V1 and V2 algorhithms of midjourney alone were, like, an exponential jump up in diversity, quality and possibilities of personalizing, and it was just released to public beta this week. I cannot even imagine how big the next steps will be in a few years, and at some point it will be indistinguishable from any other image based art…
What is more important to me as an artist is to think about how to personalize the putput, implement it to my works, use it as reference, or just as is and giving it enough personal touch, expand it and whatnot
1 Like
I’d still argue that bodies and the meaty bits of our brains are a lot smarter than we give them credit for. But those are basically impossible to account for in any meaningful way in an ML model, and I don’t expect that to be a popular opinion in this thread.
Not to put too fine a point on it, but this is quite pretentious. “Doing a render” does not equal understanding these fields or decisions.
Correct, what I was trying to say is “prior to even be able to render”, at least when you use Houdini for what is best at, which is clearly simulations and procedural workflow
Ah, so the problem isn’t that it’s not art, but that it’s not good art? How do you define art? And what makes it good? Maybe it doesn’t even matter if it isn’t art - most creative work isn’t.
This all just sounds like gate-keeping to me - right from the assertion that literally all of the works created ‘look like shit’ - which I’d go as far as to say is objectively wrong.
Intent in art can be important, but that still exists with the human providing the prompt, that journey you speak of is still there, this is just a short-cut to reach the destination.
Besides it’s all a matter of perspective. An event in your life might determine that you favour line-work and dark tones - and an AI can have that bias programmed into it. If the end result is the same who cares what sits behind it? At that point it’s a discussion of ego, right?
And this hits it, I suspect many of us will need to get over the ego-mound that our species has erected for itself. We’re fairly deterministic, and we’re inspired by that which we’re exposed to - there’s no reason that a sufficiently advanced AI can’t express its motiviations through prompts, that its programming and not neurons doesn’t itself invalidate the end result, at least for me.
Maybe we are ourselves are simply sufficiently advanced programming. If you discovered tomorrow that a higher intelligence created us would that suddenly disqualify all art that we’ve made?
1 Like
Credit for this one goes to JovenB/Midjourney, I just spotted it in a general channel and thought it was pretty beautiful!
3 Likes
I’m just expressing my opinion, and it almost sounds like you’re gate-keeping my opinion?
Do you suppose these ML models have opinions on any of this stuff? Are the opinions the annoying bit?
This is pretty dishonest, because it ignores my point (albeit subjective) that the destination has not been reached.
It is certainly a matter of perspective, but I think it’s dangerous and cynical to boil that down to “ego” like some kind of pesky psychological insect. I think it’s even dangerous to boil the work down into “line-work and dark tones” - you might be throwing a lot of subtle visual meaning away simply due to being unable to categorize it. The extreme of this line of thinking might be a dystopia where you can only buy “pork product” which equates all types of bacon, ham, pork belly, SPAM, etc.
That’s fair, you’re not presenting your opinions as facts - you just came in pretty hot
As a creative manager I could do with a lot fewer opinions
Depends on the objective - I already see Midjourney art used in practical, creative applications and the scope is quite wide. The biggest barriers with a lot of AI use right now are with licensing. Anyone that’s worked with stock imagery will see the immediate value in these tools.
However that’s different to art - I haven’t seen anyone suggest we hang any of this stuff in a gallery. However I’ve already seen one creative referred to as an ‘AI Artist’ that has some quite impressive collections.
It will be interesting to see how the medium evolves over the next couple of years and where its value shines brightest - no doubt it’s here to stay.
1 Like
I accept this use case - “art as product.” This is what the market demands which is why ComicCon sprung to mind. Art-for-hire, Disney-licensed art, art on board games, etc. I’m sure this is a billion dollar market and ML could wipe out millions of young illustrators’ jobs and their annoying opinions to the glee of loquacious dilettantes and managerial types everwhere. “Optimize organizational structures” if you will. I am not interested this use case of art personally, but I definitely see the utility from a capitalist perspective. My hope is that people will tire of its samey-ness the same way they are beginning to tire of the superhero factory.
If only! More practically it will allow them to do more meaningful, valuable work. Junior Designer isn’t a vocation, it’s the first step in a career path, and that career will still exist.
This is less like factory automation which leads to redundancy and more like how software/computing in general changed the way the creative industry works. Designers just moved from drafting tables to Macs.
Whether the medium holds value in the more traditional art space only time will tell - no doubt there will be steadfast detractors either way.
People said the same thing about drum machines after all.
1 Like
It is very engaging to watch and listen to Jeff Mills work, for example.
1 Like
Something that was afforded by the way the machine could be interacted with.
Maybe in the not so distant future AI image generation tools can work more like Maya and less like a search engine - which could have a similar impact on both how much you can impart on the work and how seriously critics will take it.
3 Likes
That’s what DALL•E-2 already does (in a very rudimentary way) by impainting specific areas, to change them, and I already see many more of these surgical tools following in the near future. At some point it will become a totally “natural” tool like Photoshop Brushes, Subject masking based on ML, imagetracing like in Procreate / Houdini, etc.
All these tools are either made by someone else and massively used without always giving credit to them, or even fully based on AI under the hood and not really questioned.
I would say that ALL digital art is, to some degree, based on specific algorithms. Look at Pen Plotter art, which is massively based on numpy or p5js datasets.
Look at Refik Anadol whose “impressive AI art” is nothing but vector field / perlin noise based particle / fluid simulations based on realtime data (and he even does not program the algorithms himself…there is a massive team behind him).
So should everyone now give credit to all the programmers that are mentioned on the splash screen during Photoshop Booting? Or give money to the remaining family members of Perlin / Mandelbrodt etc.?
If we go down that route, artwork booklets would be several pages long, only giving credit to every bug that had to die to make a painter paint specific colors 
1 Like
I’m looking forward to this, because I think this is when it’s going to start getting good.
I should not have lumped this type of stuff in with the “type text into HTML form -> magic AI bot shits out PNG” paradigm. I think this stuff is a lot more interesting, personally. I don’t disagree with what you said but it requires more nuanced understanding and engagement with the data sets, and I think this is a more effective use of ML. It’s just textural data the way LFOs or sequences or envelopes are just modulation sources - the fact that it came from an ML set is just an educated aesthetic choice. Nothing wrong with that.
Not sure if you’re introducing the idea of crediting all the giants whose shoulders we’re standing on, or if I spaced out and missed that conversation. This is kind of a weak strawman, though - who reads or even distributes (printed?) art booklets anymore? Names or URLs of the libraries used probably suffice, and then you can hop through and traverse the layers of citations of each depending on your level of interest. But yeah, I don’t disagree on either side of that argument - there are person-millenia of code beneath each of these works, and it would be tedious as hell to credit them all.
1 Like
I signed up for DallE2 and Midjourney for work (I’m a graphic designer but get calls for illustrative bits which I am poor at). So far Midjourney has been the most effective, given a bit of training it by getting variations on the best of each prompt.
Here’s an interesting one - I asked for a padlock made of shattered glass:
5 Likes
That’s a nice one too👍🏿
Dall-E is starting to look nicer than Midjourney….to me.
1 Like
just cancelled my midjourney membership. it’s been fun but novelty has worn off, i can only see the libraries and algorithms now and it’s got a bit samey. not much surprises me and i don’t know if it’s the unimaginative prompts or limits of the system.
their upgrades and new features seem exciting, i’ll revisit “AI” art one day when it becomes more interactive and checks in on certain elements of a prompt it wants clarity on.
i do prefer the idea of an informed collaboration rather than a human tagged library and coded process masquerading as intelligence.
3 Likes