Workflow of generating concept art without AI (several days):
Pitch meeting asking for concept art of a particular thing.
Artist makes several examples of concept art.
Discussion meeting to approve or request tweaks (repeat steps 2 & 3 as needed).
Concept art accepted as “visual canon” for game development.
Finished art manually created and applied for various game assets (models, buttons, loading screens, etc).
Workflow of generating concept art with AI (several hours at most):
(Steps 1 through 4 of the previous workflow can be done simultaneously in a single “meeting” within just a few hours at most).
Finished art manually created and applied for various game assets (models, buttons, loading screens, etc).
AI art generation is getting better, but sometimes it gets stuck generating the wrong things (like inserting extra limbs or fingers where they shouldn’t be) and it takes a while to try and figure out how to correct the AI prompts so that it does what you think you’re telling it to do instead of what you’re actually telling it to do. Even then, it’s still a huge time saver when you can generate 4 different concepts within 30 seconds rather than 1 week.
No matter how good the AI is, it still takes time to find “just what you’re looking for” rather than simply create it directly from your head through your hand.
Go check out some AI art generating tools to see what they’re capable of creating. Here are two examples:
Midjourney: https://www.midjourney.com/showcase/recent/(currently the industry leader)
Leonardo: https://leonardo.ai/(you can still make free/trial accounts here)
Meanwhile over at company X that used to employ people to voice audio books…“I’ma just fire everyone cause I can use chatgpt to voice all the books instead.”
The extra time is there to entire the Artist’s vision of what they want to see and fully realize it, as well polish it before release. That’s generally how good art works.
…This doesn’t seem off. /s
This is missing some steps, isn’t it? Feels rather misleading here to just have 2 steps. And all without a meeting seems rather off as well. Especially since Art Cohesion is a thing they need to consider.
Meanwhile in the time it takes getting better at making hands, humans meanwhile, can make a hand in a short amount of time since… they use hands.
If i were the creator, doing this for my own vision, i would ask them “where exactly did they come from” and “what am i looking at”, as well of the copyright on this.
I get the feeling you’ve neither modeled a 3D object, nor textured/mapped one for a specific production purpose. In this case a video game.
The art style for WoW has been pretty much set in stone for decades. Changing it could be the kiss of death.
So you going to need a custom deep learning code so the the results will be repeatable.
I wouldn’t get my hopes up of having an “AI” produce a AAA game in a few hours after a coffeeshop concept meeting.
Will it ever be?
I don’t know.
I won’t say never, because never is a very long time and one can simply “get lucky” and have an eureka moment millions of years from now.
Depends on which one, if the OP is referring to the one Blizzard is working on it is built off of their own work for them. It wont be “searching the internet” like some do.
Also the OP is correct, AI is a tool that can be used to enhance the process for the artist. Get base ideas on paper quicker to present, quickly produce multiple variations, test new ideas without the risk of losing time investment.
I really dont understand why people are so hardcore against utilizing AI and AI art. Well i mean aside from the quote I used up there, if you use AI art check how its produced and make sure its not one of these ones that steal art.
Edit: AI art is also a fantastic tool for the talentless like me. I can actually make some concept art for my DnD characters to give my players visuals.
I am okay with AI in art as long as the artist is still in control and manipulating the AI, like for example all of the AI tools in modern Photoshop.
I am, however, not okay with AI doing all of the work and generating the art itself without any extensive amount of input from the artist themselves.
It is true that many digital artists use AI to make their workflow quicker/easier but it’s not like that one website where you type in “Shrek having a tea party” and it generates the art for you, it’s moreso the artist is still in control and is manipulating the AI to make tedious tasks a lot quicker, especially when it comes to 3D modelling and animation.
Yeah I think a lot of people fail to realize this.
Even something as simple as the basic software paint and the “fill” tool could be considered AI. Like do people expect the artist to fill in an area, individually coloring each pixel, or have an AI that fills the entire area instantly with the color the artist chose.
See this is the thing, people argue ai “can’t come up with” x.
That’s not the intent of it.
Upscaling old armor sets, taking a 2d rough draft concept and applying it to a 3d model, changing textures rapidly, it’s going to be a tool to make mundane work much much faster.
A lot of people angry that they are employed to do mundane work and it will be automated away.
Yes, better tools for scriptable actions for modeling/texturing/mapping would be nice to have.
But I really don’t see how an “AI” is going to create the assets from scratch with little more than a description string.
It’s going to need to make the skeletons, then models of the characters , clothes. armor , weapons, etc,etc. Rig them to the skeleton it made(good luck on this one).
Then weightmap, add jcm(joint controlled morphs), mcm’s(morph controlled morphs), animation IK chains(walking, casting, etc,etc), joint movement limitations…(there’s a rather long list of thigs to cover here, but this is the gist of it.)
“AI” and I use the term very loosly here, just isn’t to this point yet. The best it going to give right now is a random iteration of pre-made assets it was given to work with that may or may not function as intended.
You’re overhyping this and fearmongering over something that still can’t do hands and need other artists to be fed into, not to mention, to contend with copyright laws preventing it’s professional use since it uses other people’s images.
And knowing google, they would pour money into something before shortly abandoning it.
