[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/3/ - 3DCG

View post   
File: 61 KB, 600x424, 2ceb61b2eb0683395052389971ad40d2.jpg [View same] [iqdb] [saucenao] [google]
577453 No.577453 [Reply] [Original]

How close are we to a "make art" button /3/? When will we become nothing more than button pressers with little to no input when it comes to building a scene or making a character?

pic unrelated

>> No.577509

>>577453
It will never happen. Please fuck off.
Unless you want to talk about AI that is smart enough to do this on its own, in which case >>>/g/

>> No.577510

Probably like 100-200 years minimum.

>> No.577537

>>577510
lol no

10-20 tops

most 3d objects, including characters etc will be created just using neural networks using photos and keywords as guides

you can probably already automate every /3/ process in theory, it's just gonna take a while for the infrastructure to get aligned

also applies to rigging, animating and what not

>> No.577647

We are closer to the "ship everything to India, and pay them peanuts" button.

>> No.577651

>>577647
I recently got an email from a 3D studio in India. They offered me my personal 3D artist for 1500€/month. Pretty tempting desu.

>> No.577662

>>577537
Thats about recreating things, but if we talk about A.I being able to create advanced objects from scratch that don't have a counterpart in the real world then we are not even close

>> No.578387

>>577662
>create advanced objects from scratch that don't have a counterpart in the real world
can you honestly name any examples of this though?

any creative or artistic work humans do is just a rehash of previously consumed information. you can't create new information
we've already learned how to stimulate neural nets to hallucinate some imagined input. I would bet money that machines will
be able to produce coherent models for use in the industry in the next 3-5 years. anyone else who keeps up with the machine
learning scene could tell you we're making leaps and strides, it's just a matter of fiddling and training with lots and lots of data at this point.

this scares me about the state of machine learning. it's a sign the singularity is rushing towards us. almost none of the public seems
to be aware of how much they're being influenced by decisions which have been made on advice from data analyzing machines.

>> No.578411

>>578387

>can you honestly name any examples of this though?

Every animated movie ever, also a machine will not animate for you because before you write to it what it has to do (turn the arm 45 degrees from frame 210 to frame 250 then close the eye in frame 340 to frame 345) you would have animated it faster yourself.

>> No.578430

>>577537
I would give it 50 years but this answers the thread enough.

>> No.578459

I want to create a 3D model of my waifu, print her, give her a soul and life, so we can be together, not only in our hearts, but also as physical beings.

>> No.578484

>>578459

I too wish I could fug my waifu desu ;-;

>> No.578798

>>578411
That's old machines.

New machines can learn, and the only difference between "rotate forearm 37 deg z from frame 35 to frame 80" and "make a scene of andrew punching joseph in the face" is how many layers you want to put on your neural network.

The name of the game is finding the proper network architecture now.

>> No.578816

>>577662
>A.I being able to create advanced objects from scratch that don't have a counterpart in the real world then we are not even close

Already happening. AIs are getting better at making new stuff faster than familiar stuff, because new stuff simply needs to work, where as familiar stuff has to work *and* deal with the added complexity of being recognizable.

>> No.578863

>>577453
>What is Daz Studio