/3/ - 3DCG

[Anonymous No.918887]

Hearing a lot of big talk from industry vets on other boards now - they say that sculpting realistic humans like grasetti is dead and buried and at best looks uncanny. They say that everything is scanned and cleaned up when going for realism in production. Does this resonate with you, /3/?

[Anonymous No.918888]

Scanning and clean up doesn't excuse lack of developing the skills necessary for likeness and realistic characters. The process of handling scans still requires a large effort which can be detrimental if used as a crutch, ignoring the time it takes for artists to land it, ultimately being executed disappointingly poor despite having (incorrect) scan data. See the comparison between WWE 2k22 and WWE 2k20 for an example.

[Anonymous No.918906]

Photorealism always bored me to no end. Drawn or painted it might wow you for a moment but that's all there is to it. Photorealistic digital sculpts are even 1000x more boring.
Creative intent behind artwork is what makes shit interesting.

[Anonymous No.918918]

>>918887
Photorealism is here to stay because
a) it's cheaper
b) it tricks the audience into thinking "woah, great graphics!"
c) it may be more resource greedy but with these new cards this is becoming a non-issue

Why is it cheaper?
Every artist on the planet knows how to do realism because you just imitate what's real. This means you don't have to waste resources on finding a stylized aesthetic that appeals to the masses. Think of "Photorealism" as the default artstyle that everyone accepts without complaint.

Also all these new game engines and creative technologies (Unreal Engine, Marvelous Designer, Substance painter, etc) are tuned for realism meaning the formula for "realism" is built into the tool itself. So 3D asset creation is now more about doing the work rather than being creative.

Realistic assets, because they have the same artstyle, are interchangeable between games. A Panzer tank is going to look the same no matter what Battlefield remake we're on at this point.


The downside of photorealism is that there's an upper limit to how far you can go. The moment movies and games reach that point the audience will get bored and will actively seek out stylized art. Look at all the success Overwatch has had. Or Fortnite. I 100% attribute it to the artstyle. I'll fight you on this point. Fall Guys would have been a complete failure if it had a realistic artstyle.

Going forward I believe studios will pay the big bux to the artists whose WAIFU style appeals to the masses. Not joking. There's this one guy that all he does is these close super rendered shots of anime+western waifus. Japan paid him (a lot of?) money for the rights to his artstyle. Pic related.

[Anonymous No.918919]

>>918887
>Hearing a lot of big talk from industry vets on other boards now

Oh no you heard big talk. Well my god tells me to kill all infidels. It doesn't get more big talk than that.

[Anonymous No.918969]

>>918888
>See the comparison between WWE 2k22 and WWE 2k20
come on now. We are not talking about games designed for the PS4 here, but film

[Anonymous No.918975]

>>918888
>incorrect scan data
oops i scanned the dog . . .

[Anonymous No.918976]

>>918918
>Fall Guys would have been a complete failure if it had a realistic artstyle
Yes probably, because it wouldn't suit the genre and purpose of the game

[Anonymous No.919017]

>>918918
lolno, realistic models are tougher, expensive, and time consuming

if you think otherwise you have dunning kruger

[Anonymous No.919018]

>>918969
film, game, doesn't matter. What you had proposed for the future of the industry where leaning on scans and for artists to not bothered learning to approach human faces despite necessary for directive changes after the base scan, because anons on other boards said it was dead and buried, and at best looking uncanny.

>>918975
You think the data has been handled correctly from the steps of initial scan to asset?

[Anonymous No.919020]

digidoubles have been scanned for a long time; no one's sculpting <marvel actor> from scratch. but translating scans to production models still requires a tonne of human expertise.
will it get mostly automated by the 2030s? sure.
right now? no.

[Anonymous No.919022]

>>919018
>What you had proposed for the future of the industry where leaning on scans and for artists to not bothered learning to approach human faces
we're not talking just about the human face, but the entire body.

[Anonymous No.919032]

>>919022
>OP picture is the face of Alita, a hot topic when it comes to photo realism and stylistic adjustments
>uncanny predominately refers to the face for examples
> no I'm talking about movies
> no I'm talking about entire body scans
Again, It doesn't change shit about what has been mentioned. Only because you don't like the specified examples. The face is part of the scanning process an entire body double. If you want to talk about scans for something like retarget mocap to doubles where the actors face is not seen in the shot at all then specify, otherwise I will point out the issues of adopting this approach and applying to creating humans. This includes the face.

[Anonymous No.919033]

>>919032
anon, industry veterans on other boards are saying that scanning has replaced the need for sculpting realism. Is this encroaching on your area?

[Anonymous No.919034]

>>919033
I'm tired of boogeymen. Newer workflows don't immediately nullify existing processes and knowledge about how the world works. They are used when beneficial. As the other anon mentioned scanning has been used for a long time but doesn't mean it is good to go without any interaction after the scan. You will have to do manual work at some point, you will push the vert, you will listen to the demands of the pixel fucker in the dailies, you will shot sculpt on frame 3, 34, 49 because it is quicker than to push it back up the pipeline for fixing.

Here is one for the doom and gloom crowd, Until the day of the subjective human director becomes extinct and replaced with the AI and who seeks out judgement to whatever is correct by algorithm and what yields the best outcome for the consumption by other AI, No one cares how you create the initial form be it sculpt or scan and no one will care after. In end making robot films for robots to enjoy is what you will get.

[Anonymous No.919035]

>>918887
>Does this resonate with you, /3/?
I personally don't give a fuck
colour me resonated when you can scan a stylized dragon

[Anonymous No.919036]

>>918887
Yes I'm terrified by this sudden imposition of revolutionary technology. It's not like it's been standard vfx practice for well over 20 years or anything like that that

[Anonymous No.919044]

>>919017
just more time consuming
I know this is /3/ and basic anatomical knowledge about the human form is seen as the Holy Grail... but there's a reason why the majority of the artists in the industry aim for a realistic style: lowest common denominator

there are more photorealistic games out there than stylized games even though the most popular games are stylized games

numbers do not lie

[Anonymous No.919045]

>>919036
It hasnt been standard to scan for realtime.

[Anonymous No.919048]

>>919044
>just more time consuming
Clearly you are unemployed if you think this doesn't automatically make it more expensive.

[Anonymous No.919060]

>>919048
I am unemployed.
Why would I ever want to work for a studio?

[Anonymous No.919090]

This board is completely dead until some fucking idiot makes some idiotic thread like this and all of a sudden everybody decides to reply. This board should just be merged.

[Anonymous No.919114]

>>918887
This is true in a production environment because its way more time and labor efficient to do a scan but the idea that all unscanned realistic character renders are uncanny is just wrong. Likeness is very tricky, time consuming, and takes a lot of skill/knowledge

examples:

https://www.artstation.com/artwork/nQ2LXK
"Getting the likeness was tricky as we were not provided with any scan or references, so I had to google/pinterest around to get good ones in order to get it right."

https://www.artstation.com/artwork/nQ2LXK

https://www.artstation.com/artwork/aR4gbk

[Anonymous No.919128]

>>918887
this movie sucked and it bombed because she looked fucking weird. this type of 3d will never sit well with western audiences.

[Anonymous No.919137]

>>918887
Yeah industry friends are saying similar stuff. Basically the future is Daz + scanning then using ai to help clean up.

[Anonymous No.919172]

>>919114
think about all the time it will take to do the many poses for blendshapes

>>919128
it was a good flick. I'm thinking of buying the 4k desu

[Anonymous No.919186]

>>919044
Those popular stylized games you’re talking about are free. I doubt their art style was a major factor here. RDR2 was also verd popular and it isn’t stylized.

I kinda agree with the first paragraph though.

[Anonymous No.919209]

>>919045
Yes it has. People been using texturexyz since forever

[Anonymous No.919217]

>>919209
forever must mean something else in your burger language, zoomer.


I for one remember like it was just yesterday that I was playing the original gameboy because that was the lastest thing

[Anonymous No.919224]

>>919172
800 blendshapes
I can do that in 30 days
Starting from scratch (don't have prior blendshapes) you can do that in 3-4 months

If you're making a face based on an actual actor it will double the time. No scans needed. Just 2400 pictures of said 800 blendshapes.

[Anonymous No.919226]

>>919186
> RDR2 was also verd popular and it isn’t stylized.
> was

> Those popular stylized games you’re talking about are free
> free
> the real value of Fortnite in 2022 is at least $50 billion

I don't understand how people can make such statements and the fact that they invalidate their own argument completely flies over their heads.

For a game like Fortnite, RDR2 is a rounding error.

[Anonymous No.919228]

>>919224
>If you're making a face based on an actual actor it will double the time. No scans needed. Just 2400 pictures of said 800 blendshapes.
uterly ridiculous. If you need that many pictures you might as well scan. Imagine if you need to change something entirely

It will look a lot better and be faster.

[Anonymous No.919233]

>>919228
2400 isn't much, 3 pictures per blendshape, and they're to be used as references. Scanning would mean 800 scans

Any asshole can take a picture to be used as reference, but photogrammetry at that level only a few places offer it. Also you need to fly the actor there and pay for the time and accommodation. Or you can pay the actor directly to take a bunch of selfies. Assuming it takes 5 min per blendshape reference it's like 2 weeks of EASY work. You think you can scan a face in 5 minutes and make the data instantly usable for the artist?

[Anonymous No.919234]

>>919233
>Or you can pay the actor directly to take a bunch of selfies.

Anon...we are done here

[Anonymous No.919267]

I've only ever worked on stylized productions, but what your saying sounds about right in terms of the 'we need to do this fast and turn a profit' attitude that most vfx houses have. And also what's the point of hand sculpting a likeness when you can just scan the actor? Photorealism is boring anyway and if you can sculpt a 1:1 likeness that's a neat party trick but not much more than that. It doesn't have much practical use since most things in CG are stylized to some degree.

[Anonymous No.919288]

>>919226
Hmm ok, but for example gta v and pubg have been popular for a long time as well, people are still playing them. But do you really think the stylized aspect of them is what keeps them alive?

Anyway, idk how we even got into game popularity and why it should matter.

[Anonymous No.919398]

>>918887
>scanning a face that doesn't exist

[nutsack yo momma No.919968]

>>918919
of course it does. i decided to kill your god. the end.