[Aya] [Burichan] [Cirno Blue] [Futaba] [Gurochan] [Marisa Black] [Murderer White] [Nazrin Grey] [Photon] [Reimu Red] [Spooky Suwako] - [Home] [FAQ] [Thread List] [Manage]

[Return]
Posting mode: Reply
Leave these fields empty (spam trap):
Name
Email
Subject
Comment
File [
CAPTCHA
Password (for post and file deletion)
  • Supported file types are: GIF, JPG, PNG
  • Maximum file size allowed is 8192 KB.
  • Images greater than 256x256 pixels will be thumbnailed.

File: 1675976376217.png -(1147606 B, 760x1152) Thumbnail displayed, click image for full size.
1147606 No.747  

now that the dust has settled, HONEST THOUGHTS on ai generated art?
new AIya just dropped

>> No.749  

warmly waiting to get more consistent results with it

>> No.750  

Same as always
One crop of midwits replace another
No impact is had on the actual field itself
Adjacent/reliant fields benefit from the advancement of crude RAD tools

It's still just neverwinter nights portraits tier but it's better neverwinter nights portraits
constructoids and nftfags do still be seething though lol

>> No.755  
>No impact is had on the actual field itself

im sure more than a few people are no longer willing to pay for lewd commissions when they can prompt it themselves myself included

>> No.756  

>>755
literally not relevant at all wtf is this nonsequitur
you weren't commissioning anyone relevant and you weren't commissioning anything relevant
literally

>One crop of midwits replace another
>constructoids do still be seething though lol
>> No.758  

>>755
I have seen an increase in demand for human drawn pornography due to the overwhelming flood of AI art looking terrible

>> No.759  
File: 1675994038393.png -(2182278 B, 1024x1280) Thumbnail displayed, click image for full size.
2182278

AI art is to illustration what photography was to painting. Illustrations for practical purposes (marketing, instructions, etc) will in the long term be largely replaced by AI-generated art just like how photography for a large part replaced portraits and other semi-practical depictions of reality. A lot of people making illustrations as a craft will need to find different work, and people skilled at prompting will take their places ('you just enter some text' is the new 'you just press the shutter'; it will gradually be acknowledged as an artistic skill in itself), but the artform itself will survive, still being made by enthusiasts and and commissioned on a smaller scale.

To me personally it doesn't matter whether something was 'handcrafted' (to the extent that can still be said with most art being digitally made), prompted, or a combination of both (which I think has a lot of overlooked potential). Some people claim there is no artistry behind prompted art, but it's still the result of someone trying to convert an idea in their head into something visible, and the prompter makes many decisions to get the end result as close to their vision as possible. The AI art generator is a tool, just the same as the drawing and image editing programs current illustrators use. This tool just eliminates the dexterity requirement (which many illustrators trained for years), just like the camera did for depictions of reality. It's understandable that people are upset that the skills they trained through years of hard work are now in less demand, but as it's a way to transform your imagination into something perceptible AI prompting is definitely art.

Of course, as with photography, the lower barrier entry does mean most of what gets produced is garbage. But the fact that people are taking poor selfies to post on Facebook does not disqualify proper photography as an artform, and the same should apply to art prompting.

>>747

>now that the dust has settled

The dust hasn't even begun to settle. We're still at the very beginning of discovering what is possible, not just technically but also artistically. Meanwhile, it's only now slowly starting to get used in society, and legislation around it is completely in flux right now. The artform is currently in a similar stage as photography was in the early 1800s, and it will become very different as it evolves.

>> No.766  
File: 1676033682386.jpg -(1020304 B, 2979x3000) Thumbnail displayed, click image for full size.
1020304

>>756
uh yes i was and yes i was (to me, which is all that matters)
also please stop making up words i can barely keep track of them anymore

>> No.777  

>>759
At last I truly see...

>> No.779  
File: 1676089616507.jpg -(86220 B, 600x600) Thumbnail displayed, click image for full size.
86220

>>759

>Some people claim there is no artistry behind prompted art

It's understandable, because the renders themselves are not works of art. With pic related, the actual work is this:

>"kono_subarashii_sekai_ni_shukufuku_wo! megumin 1girl ^_^ backlighting blush brown_hair cape closed_eyes collar facing_viewer fingerless_gloves flat_chest gloves hat lens_flare short_hair short_hair_with_long_locks smile solo staff sunset witch_hat dacchi"

But why is the render not art in itself? Well ultimately art is an act of expression, and the AI's role is deterministic once the prompt and seed have been chosen. At best the person who curated the training data is the one who made the decisions there, but that would make the model itself the work, which I can agree with.

To put it another way, prompts are a degenerate case of prose. Speaking, cabinet making, and simple skills are all "arts" too, in a traditional sense. Prompts as a medium are a much narrower space than poetry or narrative writing: Word order is not relevant, for example. Maybe not the case for language model prompts, but then again, do any of those produce any good results? I don't really care. The danbooru-query prompts are very sensible and produce impressive enough results. The future of prompt based art is probably in producing, rigging, and rendering 3D models. Frankly that would account for most of the shortcomings in consistency and form, and make the whole thing a hell of a lot better for working with. Get a hand model and position it. You don't need AI to get that right. It just needs to render it with a style.

Anyway, compared to other mediums, prompts are fairly low on the spectrum of high and low art. Unfortunately, many supposed "real" artists operate at the same level. See: all the professional artists crying about "muh style." (https://archive.is/W8CTK) "Professional" here is just a coded way of saying that one's livelihood is dependent on selling "rights" to a group that extorts people effectively. None of these artists maintain any kind of virile tradition. They will not pass on any substantial skills or artforms as part of a coherent culture. Their work is destined to be sold back to themselves and their successors. They are sold the lie that they create "art," when in reality their purpose is to supress it by drowning society in an ocean of low effort copy work. They say that limitations breed creativity, or at least, that's what corporate marketing drones say when they are told to draw mystery meat blob human #7043. If AI art puts even one of these firms out of business I will say it has been a success.

>> No.780  

I believe you held something back for too long
it grew strong
and energy got it's own will
and people think they make music still
but music is there without you or me
we just manipulate for better or worse
so let it situate

>> No.781  

>>766
frogsnake energy

>> No.782  
File: 1676098999418.png -(2738561 B, 1408x1024) Thumbnail displayed, click image for full size.
2738561

>>779

>I don't really care. The danbooru-query prompts are very sensible and produce impressive enough results.

Isn't this just "you just press the shutter"? Word order does matter, and there are various other ways to weight certain words. There's also negative prompts, image-to-image, inpainting, and a bunch more stuff. You don't need proper lighting and a good angle to make a photograph, but there are plenty of variables to tweak for artists who don't just settle for 'good enough'.

(Image not mine, neither is the previous one.)

>> No.786  
File: 1676136367574.png -(5339584 B, 1223x2445) Thumbnail displayed, click image for full size.
5339584

>>782
Does word order matter with the kind of prompt I posted? I know about the other kinds, but all the best images I've seen are these keyword prompts. I know you can weight words, which could be used to impose an order, although the current bracket syntax makes this unreasonable beyond simple prompts. As I said, I know the language model stuff exists, but my impression was that it sucks. The amount of control from just using tags seems superior. Image-to-image is cool, and works pretty good, and in that case I can see that it is a really useful asset generator, which is why I suggested that 3D rendering is where this stuff will really shine.

>press the shutter

What I said before applies equally from photography to drawing as it does with ai to drawing. Photography introduces some new variables, but it really is just a degenerate case of drawing. The photography gives up a huge degree of control and in favor of convenience. 90% of photographers just take nicely lined up pictures, and painters filled a similar role in the past. (Double exposure and other effects notwithstanding. Development is arguably an artform in itself, but most people completely ignore this today.) It's good though, because many realist artists never would have realized that their life work could have been replaced by a mechanical box with a button, forcing them to set their sights higher. Photography is an art, but it inevitably devalued lower works of art through increased supply. This is probably a big reason why so many styles developed in the past few centuries, to differentiate themselves from realism. On one hand, thats partially responsible for low effort trash, and the situation I described in the latter half of my post, but obviously there are many excellent artists who go completely unnoticed too. If AI does something similar to arts in general, including music, it could go even further by destroying the existing IP system and lead to a golden era for artists who actually want to pursue high art.

>> No.787  
>it could go even further by destroying the existing IP system and lead to a golden era for artists who actually want to pursue high art.

I don't really see how it has any impact. The demand for high art won't change in a response to a change in the supply of low art. The increase in general wealth will be good for it but the people who aren't pursuing high art in the first place aren't not pursuing high art because they don't have time, it's literally because they're subhumans and/or philistines.

>> No.850  
File: 1676687483795.png -(1321929 B, 1024x1536) Thumbnail displayed, click image for full size.
1321929

new big stinky gem

>> No.851  

>>850
eyes are a bit fucked but its still an emerald

>> No.882  

moar

>> No.994  
File: 1679084365540.png -(47958 B, 1316x350) Thumbnail displayed, click image for full size.
47958

Behold the future.

>> No.995  

in the futures doorbells sing



Delete Post []
Password