- Hi, everyone. Thanks for coming out. I guess I'll start off by...so
I'm Brandon. I'm one of the CDs over at Ghost Town. My main focus over there is
kind of heading up a lot of our 3D, and a lot of our look dev, but why don't I start
us off by just showcasing some of the recent projects that we've been doing?
♪ [music] ♪
So that's some of the recent projects that
we've been working on over at Ghost Town. Today, I kind of want to, kind of focus in
on the idea of C4D as a bit of a platform for us. So, really, up until we kind of
bounced into Maxon and Cinema 4D, we hadn't really been much of a 3D house.
We'd dabbled in some of the different packages, but it'd just never really taken
hold; it never really spoke that well to the arts, kind of, driven community that
we have at our offices. And once we found Maxon and Cinema, things just seemed to
kind of click, and from there it really... Cinema and Maxon turned to kind of this
Grand Central Station, where it, kind of, turned into, kind of, the bedrock of a lot
of the projects that we do. It was easy enough to, kind of, do R&D and for us to
kind of feel comfortable that we can toy around and, kind of, play and have fun in
3D in a way that we really hadn't been able to do in other packages up until this
point in time. The first thing that I want to kind of
touch into, though, a little bit, is really for us... When you do a lot of
experimental work and you do some more straightforward work you have to have a
really firm understanding of, kind of, the technology that you're using, and the
ability to, kind of, fix it if there's a problem. Every single piece of software
has its own quirks, has its own kinks to it and rendering for 3D is a major, major
item that also has a lot of debugging and a lot of, kind of, work that has to go
into it. And that really, kind of, came to a very specific, kind of, point with a
recent project that we did for this documentary called, A Faster Horse, and
I'll show you guys that real quickly, some shots from it.
So essentially, we had three major sections that we had to worry about for
the film; some general clean-up and your general v-effects approach. And then we
had this major scene that you're seeing right here, where we had to, kind of,
create this illustrative, kind of, cell-shaded-style approach and render, and
then this more of a CAD-style approach to kind of showcase all the bits and all of
the engineering that really goes in when you're designing a car from the ground up,
like they recently did with the new Mustang.
We had a really tight timeline, a very small staff that we're working on, here,
we don't have a massive render farm or anything, really, that we're relying on.
We do a lot of oversea renders, and we kind of have to be really nimble about our
approaches. And as we were going into the project and doing some of our general
R&D...let's see, I'll pull this up quickly. While we were talking with the
director, David, he really wanted to go for this feel of, kind of, this CAD,
industrial, kind of authentic look to really showcase this, you know, the
feeling as if you're going through the CAD sequence itself and you're really
understanding, like, how many bits and pieces are there in the car. And over the
course of the look development, which we basically had the better part of about a
week to do, we were like, okay, that sounds really, really good. But we still
wanted to have this certain, kind of, patina to it to kind of showcase the
physicality of the car, show the weight, show the dimension, and give it a little
bit more pizzaz. You're seeing some of our early look development as we were going
through, and thankfully David liked it and we moved forward with it, so it's great.
We go through, we do all the camera animations and get everything all sorted
out. And it was a Friday, we send off our renders and we come back in on Monday and
we weren't quite as far along as we thought we were going to be.
And so at this point I want to kind of bring up--and it's going to be super,
super techy at this point, but it's really, really worth understanding,
because any creative notion that you want to start off with, you have to understand
the execution path that you're going to plan on doing. Workflow is the make or
break of a lot of your creative. You may have a great idea, but if you cannot
render it in time, if you don't have the budget to do that kind of render, you're
going to have to adjust your creative. And so, really being able to kind of
troubleshoot on the fly; how can I make the look work based upon the heavy metal
that you have in the office, the budget, and the timeframe, and your staffing that
you have for the project overall. And, so, what we ran into with this one,
really specifically, and what I ended up learning this really, kind of, valuable
lesson on, is the full understanding of a multi-threaded versus single-threaded
application approach for the software. So I came back in on Monday, and we have
iMacs, we have big workstations. And I come back in and I'm looking at the render
timeframes between the two of them, and I realize that, okay, I'm looking at 21
minutes-frame for this machine. I walk over to another machine and I'm seeing 21
minutes of frame? Now, mind you, I'm looking at these two systems, and I'm
like, this doesn’t make any sense. System B, here, is a Duel Xeon 8-core,
hyper-threaded to 32 cores, has 64 gigs of RAM, solid state everything, it is a
workhorse. This thing melts frames for fun. And then I walk over to my iMac, and
the iMac rendered it in about the same amount of time. This doesn't make any
sense to me. This is going to be a huge problem if, all of a sudden my heavy
hitters are no longer being heavy hitters. So then, of course, I get to dive in and
say, okay, well why is your render slowing down? What's going on? What's your
optimization failure right now that you're not getting the results that you're
expecting to get? So then you go through, and I basically
started kind of ripping apart the render and really trying to understand what we're
looking at, here. So I did just a render using the physical render engine to get us
that nice reflection and sheen that was making me feel really good about my life.
And as you can see here, okay, now my big workstation is doing what it's supposed to
do. It's rendering at, you know, over 30-seconds faster a frame, which adds up
really really quickly. I'm like, okay, well that makes enough sense, I can live
with that. Then I went to the sketch and toon shader. So, this is where the big
uh-oh kind of flew up on screen, where I'm looking at this and I'm saying, okay, my
iMac is crushing my workstation, what is going on right now? And all of a sudden it
kind of re-dawns on me, oh, that's right, sketch and toon is a single-threaded
render operation, which means that you can have 64-cores in your system, doesn't
matter. You're bottlenecked to the fastest core speed of any one item that you have
inside of there. So, you could have a ton of cores, but if they're slow it's going
to be slow. So I'm like okay, well I have to re-think how we're handling our render,
here. We can't just batch everything out and hope that it's going to come in in
time. I need time to comp all this. So, we did a little bit of quick math,
and, you know, broke out the cell phone, because of course arithmetic is always way
beyond me, and I was doing some kind of simple calculations. And, so, you can see
here with the mixed renderer basically doing both the sketch and toon and the
physical shader at the exact same time, we were going to be clocking in at about, you
know, 210-or-so hours, you know, soaking it over to the iMacs versus the amount of
time it would take on these systems. And mind you, these hours are going to get
split across a bunch of machines because we have it all networked together. So it
wouldn't have been that long, but it just didn't feel all that efficient; I'm not
getting the most out of C4D, I'm not getting the most out of my systems, I'm
not asking my systems to do the right things.
So then I was like, okay, well then what if we compare the timeframes? And so, if
you look over here at the toon shade system, asking the iMacs and basically the
faster i7s that have a higher internal clock speed, you guys just worry about the
toon shade system. You big workstations? You worry about the physical renders. What
happens then? Now we're seeing we have some serious optimization that we're being
able to get out of our render timeframe, where our iMacs are being able to handle
all the toon shaders in about 150 hours, and our physical shader blasting through
all of our main workstations in 10.5 hours, which overall ends up being a
difference of about a little over 50 hours.
Now, 50 hours in a 4-week timeframe for delivery is night-and-day. That is you
getting files right now in your hand that you can begin to get comping on, you can
begin to do look dev on, you can really, kind of, figure out any quirks or any
issues, versus just sitting around twiddling your thumbs waiting to get
everything at the last second before you realize, “Oh, I probably should've tweaked
the camera on shot three. ” You don't want to be in that position. You don't want to
be put up against that wall where all of sudden now you have to make compromises to
the overall project just because you were sitting there hoping that it was all going
to come in at the right way. So, this was such a wild optimization that I really
kind of wanted to take the time today to really, kind of, bring these kinds of
items up because it's not just about the good ideas, it's about what you bring to
the actual execution of the idea. A good idea and a good set of mood boards is one
thing, but being able to actually figure out how that workflow is going to work,
how you can get the most out of all your systems, and how you can be efficient with
your process means that you have more time to be creative. And that's really, for me,
the paramount of why I really enjoy Cinema 4D so much is because I really enjoy being
in it. So you want to kind of reduce these uh-oh's as much as you possibly can during
the process. So, let me show you really quickly,
because while this is all pretty dense, kind of, talk right now. So we have the
same file up here: super dense, super heavy, and all this math, all this talk,
all this tech that went behind there, it was as simple as a little three-second
blip where essentially... So, as some of you may be familiar, you have your render
settings. And so we had our baseline render settings for the scene, you had
everything all set up, all your multi-passes that you felt super good
about, right? In order to basically split this up, toss it over to the server and be
able to give myself these two delta projects to be able to run from, it was as
simple as literally just taking my base render file, dupe it. We're going to call
this one "physical." Toon. Now, all I have to do is, I'm going to take my toon, turn
off ambient inclusion, because that's not really important, go into my sketch and
toon, make sure that I'm set to color, not off or texture so, that way this guy's
only worrying about toon shade, not having to think about anything else, so color's
good. Go into our options, let's get rid of all this other stuff that's just not
really worth anything when you're doing toon shading. Drop this down to five ray
depth. Now all of a sudden, I have an optimized toon shade render ready to go
out, ready to be deployed to all the iMacs and let those guys start tearing. I go
over to my physical shader and literally, I can do that or I could just swing myself
over to the actual physical render engine and it would automatically get rid of the
toon shade, because that only works in the standard and advanced render engine. And
that alone was enough. That's all it took for me to get my project set up, sent out,
and I just saved myself 50 hours of render time. Or if you're sending off to do
remote rendering, that could be a lot of money that you just saved on the project
or didn't go over. So super simple, really straightforward,
and actually, this is really easy to handle now also in the take system, where
you could be assigning all of this and, basically as you're doing your different
builds and different looks you just assign to the two different takes and you know
that your renders are ready to roll and kind of take it out from there.
So I wanted to spend a quick moment on that, and just kind of walk everyone
through this general notion. Now of course by doing this, I now have these extra
layers that I need to kind of now deal with, essentially, over in comp. Being
able to think kind of open about what your... We'll ignore that. Being able to
think, kind of, openly about... There we go. It's fine...about your multi-passes
and how to get the most out of them; thinking about what the data that's coming
in, so that way you can kind of continue to refine and build the look dev, not
necessarily asking the 3D to do everything for you, what can we get done in comp?
So we have actually--This is an actual project build from the project here, and
you can see it's not many layers that are really, honestly, going on. So if I kind
of walk through here, we have our background. I used drop codes horizon to
basically get this nice, happy little grid system. I got my camera brought in from
Cinema 4D, so it wasn't really that big of a deal. Then we have our baseline physical
footage, it's all pretty and reflective which is nice. Prep the field. Our lines,
simple little bit of color burning that's happening inside of there, and then this
was actually from our AO pass that was giving us this nice kind of illumination
or this kind of energy inside of all the pieces and all that literally was, if I go
into here, is... So we have our general AO. Invert, tweak the levels, my old
friend colorama, a little bit of curve to give it a pop, little bit of glow, and
then finally, our depth of field. Swing back over to our overall builds, last
little touches here and there, we have our UI system, and then finally, the color
correction and you have your shot. So even though I had to split it all up, I had
more layers than I was planning on originally from my comp build, it really
wasn't that big of a deal for me just to bring everything back together again and
get everyone to dance nicely together. Now that's one iteration of using the
sketch and toon system to kind of give you these nice, kind of, highlight and, kind
of, more graphic edges to everything. Then on the other side of it, we had to
actually do a build where we were doing actual, real, sketchy, kind of tooney
scell shade. And don't worry, it'll start drawing in a second. And you can see that
we're getting this nice kind of cell shade-like, analog feel as we were trying
to kind of create this draftsman table, and showing the iterations and kind of the
advancement of the car through each one of the designs. And the really kind of
interesting thing that, for me, stood out in this build was the idea that I was
going to be using the exact same tool over and over and over and over again, because
we didn't have the timeframe to actually, literally, bring in an artist to be going
through and drawing all of this. So I had to, actually, really try to kind of create
a variation in style and approach to the actual car's concept art using the exact
same ingredients each time. And I cannot stress enough the idea that
what they've built with the toon shade system in here is so wildly flexible that
you can honestly get totally, totally lost just, in a good way, in being able to kind
of find all these really, like, fine-tuned, nuance builds for the sketch
and toon shader. Being able to kind of create this line and sync system and, you
know, being able to do the overshoots, and I could spend the rest of the hour today
just going through and showing you every single in and out of the toon shade
system, but it is such a powerful, powerful feature when you're trying to
kind of create that nice, patina-ed, nuanced kind of look. It doesn't give you
that classic, cartoony, cell shade approach.
Again, the idea of being able to try to optimize your setups here, I was being
able to get my builds going by being able to split out my general paint approach,
which was using, actually, the art shader. And if you haven't played around with
that, your ability to like hop online, go download some kind of grungy texture, toss
that inside of the art shader--and I'll show you one right now...
I'm tossing that in the luminens, because I don't care about lighting, because it's
illustrated. We have a texture that's basically dropped inside of the art shader
and then I'm doing a color shift to it. But basically, this is just a very simple
jpeg of just kind of a Garrosh kind of look, and that just gets projected over
there and it give you this nice, kind of, illustrated feel for the whole piece.
Renders incredibly fast, because we're not asking it to do a whole heck of a lot. So
now I have kind of my general kind of, illustrated look, I can bring that in. I
can start getting my comp builds going, I can bring in different paint effects,
textures and different layers to kind of have the paint-on effect working for
itself, and then I send off my line pass, which... I would click render right here
but it takes a minute to go through and do. So I can show you guys this one
instead. We don't need to save that. Don't worry about that.
So this is actually the active build that we actually used for that whole scene. And
as I dive in here, one of the really cool, cool things with this is, for me
specifically, I'm not a huge animator. I'm not a guy who's going to sit there and key
frame the whole thing six ways to Sunday. I really like procedural builds. If I can
tell the system to go do something, and then I can walk off to another computer
all together and let it just kind of cook on it, money. Money, money, money, money.
I'm so happy to be doing that, because then I can be worrying about other builds,
other setups. One of the great things with the
toon-shade system is, and let me swing back over to that real quickly, is when I
go to right here, animate, I can give it all of these animation commands of saying
how I want it to draw, how I want it to actually come up on screen, do I want it
to be doing all the strokes at once, do I want it to be a sequential build? And then
the stroke order, I can say long or short. There's all these different permutations
that you can really sit there and fine tune the way in which you want this thing
to kind of appear in front of you. So for our usage here, I'll hop into the
draw-on and you can see I just asked it to draw it for me. I didn't want to worry
about it, and I gave it a couple simple commands and it gave me back that.
Brilliant. This is a great base for me to be able to work from that I can then go
through and really kind of take all of my different water color shaders and
everything else, and start getting it all really built out and feeling more analog.
Toss on, you know, a little bit of a posture-ized time and now we're getting
that nice kind of sketch and tooney, shady, kind of feel.
So yeah. So that basically kind of brings me to the close of all this. I want to
thank you guys so much for taking the time and hanging out with me for a few minutes.
If you want to get any more information about any of these projects or more, you
can always find us at GTMVFX, that's the main company website. Facebook at GTMVFX,
and Twitter @GTMVFX. If you want to yell at me and give me all kinds of notes about
what I could do better in the future, you can always find me at Instagram is
b_parvini and Twitter is @bparvini, and you can tell me exactly why you're better
than me at this exact kind of style of work.
Thank you guys, I really appreciate it.