Will 3D scans ever become the norm?
polycounter lvl 8
I find them to be fantastic reference, but I wonder if in the future, most studios will just cut out the middle-man and scan actors? Then all that needs to be done is retopology and texturing.
What do you guys think? Also, merry christmas!
What do you guys think? Also, merry christmas!
Replies
Happy Holidays.
Also :
[ame]
The process is just not as instantaneous as the video makes it look, but that's a bit of a given
Just my two cents.
Also, entertainment is always going to have elements beyond real life - sf and fantasy require imaginary creatures and environments. Resurrecting dead people without scan data will take incredible efforts.
And there's always going to be a need for stylized, non-existent human characters as well - ND still hand sculpts all their characters for Uncharted 4.
And in the end, if the asset creation gets faster and more simple, then the demand will increase to make the most out of the same budgets and schedules. If you can build a character in half the time then the client will ask for twice as many characters
So far as artists worrying about jobs for them, this is a GOOD thing, it means you spend less time modelling and more time being creative. If your only doing this cos you're good at topology or uv mapping or something then I don't really know why you're doing it. Your creative juices will be put to use elsewhere.
Real life people are never as attractive as what we see in movies. Bad hair day, skin problems, plastic surgery and so on - you will have to replicate all the efforts of make-up artists, photographers, lighters and all to get similar results and that takes as much artistry as their real-life counterparts.
Then you move to facial expressions and it turns out that no living person can do a perfect FACS AU set, they can't help but move every part of their face. The jaws and brows move all around and you need to have an artist at the helm to find out what part of the face you really care about - and to hand sculpt what the scans don't give you.
Then there are the full body scans, where you realize why body doubles still exist. Then you start animating and the director wants to change the performance or you didn't get proper data because of some funny unexpected issue. And so on...
And as for the tech side. Good technical quality is still important, it should take at least another 5-8 years before we can start to just straight export from Zbrush and build our rigs from the raw data instead of reworking stuff for optimal performance and quality. Some day we'll be able to use a million quads for a face but that's still beyond the horizon and so good modelers and tech artists can make a world of difference. and it will still help if the same person has both artistic and technical skills so you can keep the job in one hand instead of many.
for realistic characters imagine a database with millions of scans where you can blend sections of the body and face until you have that one unique face you like.
same goes with surface details.
The above could basically also be done for stylized character upto a certain degree. You`d have to build up your initial database of shapes though
the benefit of above systems would be that all of the inputdata is known and could translated into automatic lowpoly/uvs/rigs and all of that tedious work
for clothing stuff like marvelous designer gives a glimpse of whats happening with simulation but i imagine more userfriendly stuff will come along thats more aimed towards vfx.
and yeah it isnt because scanning data atm is still rough and requires a lot of handwork doesnt mean that it will still be this way in the future
Never thought of it like that. Same as how motion capture won't completely overtake keyframe animation.
Great responses. I don't know enough about scan data, but I'd definitely like to. It does seem like a very interesting area to explore.
Personally I see it as a tool to help and not replace any skills. Just like mocap is a tool for animation there is always the need to add layers and clean up data. Its the same for 3d scanning.
http://www.polycount.com/forum/showthread.php?t=140949
Could both situations be compared?
I think that's the most efficient and realistic pipeline for future development.
wait...
The joke is on you.
You're the ones who will be left holding the bag when it gets passed to you for animation so you'd better hope the modeler on hand is competent at what he/she does when it comes to cleaning up and finessing scan data.
But what if you want a face that fits in with your artistic vision but does not exist in reality. maybe you want someone who look like a combination of nick nolte and clint eastwood( horrible thought I know), then there is still room for artistic interpretation.
so scanning is really useful I feel , but maybe its not always what you need as slight stylization always looks better IMHO.
100 photo reality is really boring, maybe a bit souless
I find the hardest thing in 3d is getting the base volume right and I think scanned reference would be awesome for me to get that stage done faster
Out of interest how much would it cost me for a decent hand held scanner and which are the best makes ?
you need Xtimes a good DSLR you can shoot from your PC or at least all at once, some softboxes and flashes for good even light and a rig for all of this, and agisoft or the autodesk solution
i've seen/worked with scan data of a few recognizable actors and this is very true. a lot of work goes into making them appear just as you'd expect them to on the big screen, not like that monday morning face staring back at you with a blank expression.
it also gets ... interesting when the scan session wasn't all that well done (e.g. done quickly on-set somewhere in-between things) and the person is not easily available for a re-shoot.
Eh? What's the point? The riggers and technical artists will be the one pesting against erronous topology, not the animator, which job is just to animate, as in to bring it to life.
Here's what you didn't get. Let's look at both positions the same: Animators and Modelers.
Modelers: Based on initial illustrations and concept design, a 3D modeler has to form, sculpt and bring shape a character, a scenery or an object within a 3d Software ( let's put it that way).
Animators: The animator's job is to take that object, character, element (after it has been rigged) and bring it to life by giving motions with a selection of set controls.
Both artists find pride and satisfaction in their job because, even though, they are often constrained by directors , animations and modeling lets them CREATE something. To create, to produce something that hasn't been done under cartoony, realist, abstract and surreal visions...
(Un?)Fortunately, years ago, people managed to bring a new technology to take Real-life motions and apply them into computer imagery (that is Mo-Cap).
At that point, 3D animator's work was brought down as they would only be called to work on creatures,some complex actions scenes and such. They weren't needed anymore to animate humans when games or movies settings were set in a realistic world(animaiton movies aside), their job in doing so was only to CORRECT human motion mo-cap data.
Now, another technology is being brought forward: 3D scans..
And it's exactly the same thing as Mo-Cap but for Modelers. Soon, the technolgy will be so advanced that whenever they will need to model seomthing from real cities, let's say New-York, they will simply be able to scan the street to get some base shape right and you guys job will be to CORRECT the topology and shape to make it work.
All that because it will be cheaper, more accurate and closer to reality..just like Mo-Cap...
EDIT:
I don't hate Mo-Cap. Without it, we would have never seen epic characters like Gollum set to life so vividly on screen but when it's a matter of video game, I don't like when devs favors cheap Mo-cap in place of fantastic original key-hand animations, even if it's cheaper..
They ended up NOT using most if not all of the Mo-Cap from Serkis for Gollum. Most of what you saw in the films was hand animated.
Id wager a lot of Mo-Cap is re-animated instead of just cleaned up. Just like many scans have a lot of resculpting to get them to where they need to be.
Personally, Id appreciate it if the future of scanning was just to get the base forms and structure and then let the artists iterate on top. Kinda like what Ninja Theory is doing with HellBlade
[vv]106186817[/vv]
Asset production is a bottom up process.
Complex 3D scans will have to be cleaned by the modeler EVERY time. In some cases, whole sections will need to be remodeled as retopologizing simply takes too long. What this means is that compromises to the fidelity of the model will have to be made and not all will be conducive to animation compared to modeling from scratch.
The other factor here is time. No one wants to be the one to have to explain to the lead designer why their asset has yet to be passed on through the production pipeline. The modeler is likely to hand over a 3d scan asset that is not quite 100% optimized under such pressure. Riggers could end up with a model that makes their job of finding the ideal areas of geometry for smooth deformation that much harder which means you, as an animator, will have to try and make that work.
Your initial statement was one of mockery at the modelers expense. Yet, all the hardships that modelers deal with are not experienced in a vacuum - everyone in the production pipeline suffers. So yes, the joke is still on you.
In many studios the animator is also the person in charge of rigging. Just something to keep in mind !
While it is true that mocap doesn't always save as much time as it first may seem, it can be an awesome tool when it comes to realistic animation. Nothing is more creepy than a realistic human character with overly animated movements and cartoony poses
I was looking at scans from Ten 24 when the thought popped into my head. Judging by a lot of the scans out there, Ten 24 are one of the only companies to produce good scans via touch-ups. Example below.
I agree with MrHobo. The whole aspect of combining both scans and 3d modeling is a probably future for this sort of stuff.