[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Home]
Board
Settings Home
/3/ - 3DCG


Thread archived.
You cannot reply anymore.



File: mitos-1080x675.jpg (281 KB, 1080x675)
281 KB
281 KB JPG
Automatization is going to replace a big chunk of the jobs in the industry and we are too many. What can we do to renew ourselves? Learning procedural modelling and tech art? Starting with Houdini maybe?

https://youtu.be/fQlQQSsC47g?t=437
>>
>>680916
>What can we do to renew ourselves?
Go to war among ourselves, fueled by national-populism and xenophobia, until the excess population disappears and those lucky enough to survive can find a job in the rebuilding of civilization.
>>
>>680918
I started to learn hunting and farming but that sounds good too
>>
>>680916
Unlike most sheeple I am well prepped. I have a complete 'bug-out kit' underneath my bed, when shit hits the fan and society comes crashing down I will tie this steel wire around the rail of my balcony, the other end around my neck and then I'll see you in the next world mother suckers.
>>
nah man andrew price it's all good in the blender conference

so it's all good.
>>
>>680916
Automation suffers when you want stylization. It's why I'm an artist and not a modeler. Procedural modeling only has its uses in landscape, props, and cool androids to render.

Call me when a robot makes something with soul that has prefect topology for working not just a random mess of vertices.
>>
>>680945
There are stylized character and prop generators in houdini. Unless is a very unique style, most stylized games are overwatch-fornite like and procedural can do it. So, yeah.
>>
>>680916
Being a c.s student that does 3D stuff on the spare time I'm confident to say that you're over reacting this by 10000x.
A.I is still really dumb, we are now being amazed by all this cool things it can do but in reality we are miles behind from having A.I to replace 3D artists.
You may ask why:

1. A.I can't produce AAA assets today, not even close to it. And when it does it will fall into number 2
2. A.I returns really unoptmized data. And to optimize it falls into number 3.
3. A.I to be efficient requires insane ammounts of processing power. You may say "oh but we have super computers that the studios can use"
yeah, we may have the power required but it isn't financially viable compared to what we have today with humans doing the job.

A.I can be awesome, but it's not magic. "but in the future maann" well, in the future we will have flying cars, don't we? oh fuck.
>>
>>680951
I'm not finding anything when I look up houdini stylized character. Also key word "soul" we can tell something artistic made by a human apart from something made by machine not to mention the amount of human input needed to work houdini and get a favorable result.

It's as much automation as when we moved on from sculpting on a computer instead of manually inputting dots from clay. We just adapt and there's still a usefulness to 3d scanning hand made props
>>
>>680991
Cont.

We will always need the human element. Like the other anon said it's not magic and in the past few years what have we even achieved with Ai? Almost nothing but variations of cleverbot just taking in 1s and 0s and trying to put out similar strings of 1s and 0s it's seen before.
Other Ai only exists to serve a short lived purpose like taxes, chess, or doing random things until it finds a random string that gets it to walk longer than other versions of itself.
>>
>>680916
All you can do is start learning how to use A.I. and A.I. assisted programs just as you've been learning how to use other new programs and tools to stay relevant in the market. Civilization is advancing and if you don't advance with it you get left behind, welcome to the rules of nature.
>>
>>680916
How about you machine learning brainlets start by building something to do automatic retopology. Computers don't seem to be able to replace even the slightest bit of creative work yet. Right now I can't take this kind of luddite concern trolling seriously at all.
>>
File: Kim-Black-Bear.jpg (164 KB, 800x550)
164 KB
164 KB JPG
>>680990
C.S. minor reporting in. Can confirm, this man speaks the tru tru. I'm honestly amazed how hard people's dicks get over buzz words like "machine learning" and "AI" these days. The flying car analogy is spot on.
>>
Remember when we had this thread like three years ago?
>>
>>680916
smashing that x button rn.
"machine learning" is a fancy way of saying "brute-forcing an algorithm"

it's just gonna be another tool in your toolbox to help you come up with ideas.
>>
>>680916
why its cool
Instead of fumbling for weeks with modeling. rigging, texturing you just type "hey model me a hot chick"
>>
I dream of making incredible designs and having AI finish the textures and weight painting to my vision. Fuck you.
>>
>>680927
Baha
best comment ever
>>
remember 6 years ago people excited that 3d scaning and printing is going kill modeling by hand?
>>
>>684041
this time it's different because pretty much everyone is a fucking monkey compared with modern AI
>>
>>684047
Me: Yep, that's a car.
AI: lol what is a car?
>>
>>684047
>everyone is a fucking monkey compared with modern AI
LOL 'modern AI' is mainly a scripted scam and sucks at pattern recogniton
>>
>>683785
Not brute force. Machine learning is multi-dimensional interpolation. It needs lots of well structured examples to be useful.
>>
>>680990
>AI to be efficient requires insane amounts of processing power
>efficient
>requires insane processing power

I'm gonna let you think about that one

>>680992
>All AI is basically cleverbot

Opinion humbly discarded

>>683785
>Machine learning is brute force

If you were right, something as simple as mnist would be intractible with its 256^784 ~ 10^1888 possible inputs. Gradient descent methods are greedy algorithms, which has the possible downside of getting stuck in local optima, but they are usually searching through such high dimensional spaces that that becomes unlikely

>>684047
3D modeling is currently outside my field of research, but

>Put psychedelic sticker in FOV of modern computer vision system
>Suddenly it believes everything is a toaster
>"everyone is a fucking monkey compared to modern AI"

>>684112
>Confuses digital image processing with machine learning

>>680916
On one hand, GANs (from your video) are notoriously hard to train and still require real data as a starting point. Where is that real data going to come from? On the other hand, GANs are gaining traction because generating synthetic data to train models is cheaper and less time-consuming than gathering real data. Is there a tipping point to where we will no longer require real data anymore? Only time will tell OP
>>
Linux Kernel dev there.

>>680990
A.I have made tremendous progress in processing powers, in large part because of OpenCL, CUDA, and associated GPGPU. Nowadays you can train a neural network (or convolutional NN) somewhat fast. That's what NVidia does everyday of the week. Kudos to them to think about a convolutional NN for denoising, though that's more or less what they're for, so not many kudos anyway. CNN are best for image processing, and NVidia thought of them to process image. That's not particularly Fields Medal worthy.

Still a lot of what the current state-of-art does provoke a stupid amount of fear amongst the low T artist community. Yes, they can fill landscapes with their tech, color images, and even produce decent pictures. Yes, they can denoise image. No, they will never be able to produce assets beyond the most unimaginative of rock textures. You need intellect and artistry for that, and convolutional NN aren't at that level. And they won't before our AIs overlords completely murder us or decide a singularity is best for everyone.

My point is that you're all panicking like a bunch of sheep, while you're doing one of the most human, soulful, work ever. You're an artist. If NN aren't able to control all the facets of your life, you still have a job. If NN are able to control all the facets of your life, you're either dead or in a Singularity, so who cares.

Think for yourself.
>>
>>684212
Ok smart boy how much of ai, in this context, wouldn't just be something spitting out a design based on some information and loads of previous data sets relevant to the task?

You also missed the other part of the post where I point out the simplicity of other things we use ai for unrelated to generating original content.

Man quit pretending to be smart no ai in existence can make something original and not random without an extensive data set to base it off of not even a human can. In the end it's still put data in get data out when simplified.

You didn't even come in this thread to make a point you just came to jack yourself off about ai knowledge. Its not impressive I mean this is /3/ and you said you don't even do modeling. Why are you here?
>>
>>684259
lmao, being this assblasted.
>>
>>684259
>why are you here?

/3/ is the default board for the clover app, literally the only reason. Also my response to OP is pretty much what you're tl;dr'ing me
>>
>>684263
I just don't understand how someone can be smart and incredibly stupid at the same time. You had to scroll to get to this thread and beyond that you haven't set up clover so it doesn't default to /3/. It just baffles me. Don't say you stayed because you thought maybe 3d would be a fun hobby you can't possibly be that stupid.

>>684262
The guy replied to half the people in the thread to share his ai knowledge. He's a bit of a jackass. I wasn't even talking to the guy and he points to my post to say "opinion humbly discarded". I just don't understand how someone could have their head that far up their own ass. It wasn't shitposting obviously.
>>
>>684264
Your first post I responded to is a mess. To the best of my interpretation you reference NLP with cleverbot, taxes (?), search algorithms with chess, and genetic algorithms, maybe even reinforcement learning with your doing random things bit. This argument is completely scattered and frankly unhelpful to OP, possibly damaging to others who read your post. You were better off here >>680945. Maybe I am a jackass, but when someone audaciously claims all recent advancements in AI boil down to better cleverbots all I can think is
>opinion discarded

Also I like the default layout with the safe boards in alphanumeric order followed by the unsafe boards. It pleases my autism
>>
>>681120
Because history has shown that industries are uncaring to kicking out jobs in order to save a buck. The industry is heading towards A.I. tools that optimize work flows and minimize time spent on tedious task, like modeling, texturing, and even animation. Maybe not today, or even next year, but eventually those unable to adapt to the rising demand of tech art based skills will be shoved out, i.e. art for artist types.
>>
Don't be alarmed. Any task requiring creativity, aesthetic sense, an understanding of the context of the work to be produced, etc, is very far from being doable for AI. Artistic jobs are among the safest from automation. *If* they ever become tractable by AI, finding a job may be the least of our worries.
>>
>>684212
>Confuses digital image processing with machine learning
they are obviously related you retard
>>
>>684291
They are related, and ML even borrows techniques from it. That does not mean they are the same, faggot
>>
>>684268
>smart guy doesn't know about the significance of ai for the future of tax collection and compliance

Quit being retarded
>>
>>680916
Spearhead the VR porn market, particularly with various fetishes in mind.
>>
>>684296
AI can be applied to more than just taxes, what is so different about it that deserved such specific mention?
>>
>>684362
It was a huge hilarious big deal when they started using the Watson ai supercomputer for taxes. Are you underage or something? Maybe I should have used jeopardy instead of taxes.

It's a huge thing though ai for use in taxes alone even the irs is developing its own ai solutions. I'm sure you already knew about all that though you're really smart. It's a major specialised direction for ai that is unrelated to ai that would be used to create 3d models it shouldn't be confusing why I brought it up alongside chess solving ai and so meant to emulate something walking when talking about ai and where we are with it today.

It's all data in data out. The point in time we are now is using unbelievably expensive technology to save people 30 bucks a year on taxes and beat them at jeopardy and all creative ai lacks a "why" to what it's doing just a what and how and the only thing that brings to the table right now is novelty. It can help out the creative process by fixing things, but that's really it for the foreseeable future.

I'm going to keep being an ass because I really don't like you anon. Doesn't matter why I'm probably projecting a bit with the reasoning but I just genuinely don't like you.
>>
>>684373
>It's all data in data out.
"gg ez" -- HAL 9001, December 2054
>>
>>684373
You seem to think I disagree with you about AI lacking creativity when I actually do not. OP's video is about GANs, and you mention vaguely a small collection of unrelated AI. Your first post should have been your last. You wouldn't have gotten your feelings hurt when I called out your shallow at best knowledge on the subject.

I'll also say again, there are many areas where AI is applicable. The Army and the Air force are researching offensive (ATR) and defensive (TrojAI) uses for deep learning. Google is tackling explainability. Nvidia is leading synthetic data generation. Malware for AI and Malware AI is an emerging problem. There are advancements all the time that make AI more efficient, more portable. Google is positioning themselves as an AI first company now. The IRS isn't the only organization creating AI solutions. 2013 was a breakthrough year in AI, and we've made much progress in these past six years. Now if you'll excuse me, I'm gonna chat with cleverbot about my tax returns
>>
>>684384
I think you might be autistic. I was pointing out some examples of other ai out there for drastically different tasks which haven't really taken people's jobs and are incredibly specialized. I figured that was obvious enough not to get but for some reason you assumed I was calling chess ai cleverbot. Why would someone bring up ai aside from gans and shit in the context of "will ai take our jobs"?
It is a good point and not harmful like you think. We still have accountants and chess players. I assumed that someone reading would get that.
Up until this point I was mostly arguing that you're dumber than you think you are. Now I understand it's autism.
>>
>>684405
You are way out of your league, son.
>>
>>684409
That was an incredibly autistic thing to say holy shit anon. It's like I can see the fedora and shit eating grin thinking you just one line schooled a guy on 4chan. I didn't think you could post anything more stupid or shit sniffing until now.
>>
>>684410
I'm just stating a fact. The way you deal with it is up to you, and, by the looks of it, you're not doing well. Get a grip and stop embarrassing yourself. I'm not the guy you were calling autistic or dumb before, by the way, so don't even think I have a horse in this race. I simply cannot stand bullshit -- and you're full of it.
>>
>>684435
I knew I would get a "I'm not that guy" post. I expected him to post afterwards saying it wasn't him who said that but this is even better.

by the way my dog typed out this message and pressed post so I'm whatever I post isn't representative of me.
>>
>>684446
Whatever helps you sleep at night, mate.



Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.