/ic/ - Artwork/Critique

>>6524510
perspective >>6524324
more than one character is still a big issue if you want them interacting. pic rel is for AGI/SGI only.

>>6448425
(short) nice time frame you have there Mr Smarty Pants, care to elaborate? If you think increasing parameters will get you there, you are sorely mistaken.

>>6445920
As a tradchad and digicuck, yeah, it's pretty shitty. Simple commissions may end in a couple of years. Wont be long until trad is done by 3d printers holding paintbrushes.(must be a thing already surely)
Still, it won't be until AGI that image gen can make pic rel, pinup and one note bland portrait artists are fucked I guess.

>>6442563
Until AGI, there is no way for image gen to make complex scenes, draw what you want, but maybe push the envelope a tad, perspective, multiple interacting characters, clean intricate lines?

>>6430540
>>6430541
>>6430542
>>6430543
>>6430545
>The AI fag rages against the artists who have a strong point, by posting AI art of low hanging fruit of image making. He will not engage in further discussion, this is is last resort.

>>6231251
AI might not be as usable as you think, even with advancements. I generated 10k images in the beta of Stable Diffusion. Maybe 3 could be prints, but they still looked a bit messed up. I was quite ambitious with my prompts however, just asking for portraits it might have made more usable stuff(but shallow). I thought the img2img might fare better, but looking at the results, it's still a mess and nobody has produced anything of value from it. This type of image generation can't reproduce accurate perspective/size relationships properly. Multiple characters in a scene are all different sizes. objects are all random scales in a scene. Nobody has made 2 characters interact without limbs melting together. There are countless things that can go into a good image(pic rel) that AI is missing. What if these problems are beyond solving? we don't know yet, and shouldn't assume.

tatsuyuki tanaka