NAB 2016 Rewind - John Filipkowski: Cinema 4D's Stereoscopic Production Workflow

Photo of Cineversity

Instructor Cineversity

Share this video
  • Duration: 38:38
  • Views: 4754
  • Made with Release: 17
  • Works with Release: 17 and greater

John Filipkowski discusses Sarofsky’s process in creating main-on-end titles for Marvel’s Captain America: The Winter Soldier and Ant-Man.

John Filipkowski discusses Sarofsky’s process in creating main-on-end titles for Marvel’s Captain America: The Winter Soldier and Ant-Man. John discusses the stereoscopic workflow Sarofsky built around Cinema 4D, and provides insight into the process of creating these two dramatic title sequences.

02:55Captain America: The Winter Soldier
06:44Stereo Workflow in Captain America: The Winter Soldier
22:54Stereo Rivalry
24:57Animatic and Development Process
28:27Spline-Wire Look using Hair
33:40Take System

Recorded Live from NAB 2016 in Las Vegas.



- Hi, I am John Filipkowski. I work at Sarofsky. We are based out of Chicago. We specialize in design-driven production. We work with a broad range of clients, advertising, broadcast, network, film, entertainment industries, you name it. Basically, we do cool stuff. So let me show you what we work with and who we work for. I'll show you our demo reel. So enjoy. ♪ [music] ♪ So as you can see, we do a lot of high-end designs, commercials, television, film, new media, and what I'd like to talk about today is our film work. In case you haven't seen anything, or you may have seen it in the reel, we've done done work for Marvel. So we've done Ant-Man may-not-end titles, Captain America main titles, and all of that was produced stereoscopically. So I'd like to explain a bit about our stereoscopic workflow for these things. Before I dive in, I'd like to get you guys a little insider secret. Marvel, when we're working with these companies, or when we're working with Marvel, they are really hardcore about their security, so all the computers are not connected to the internet. We can't share any files, and what they also have us do is not say the name of the movie we're working on. So they have secret code names for every single movie we've worked on with them. So for Captain America: The Winter Soldier it's "freezerburn" Ant-Man was "bigfoot." Guardians of The Galaxy was "full tilt." And I'm not for sure, but I heard Batman vs. Superman was "megaflop," but I'm not quite sure. So if you hear me say "freezerburn" or something like that, I said it the whole project, so it just kind of comes out when we're talking about The Winter Soldier. So in case you haven't seen it, I'll let you guys take a look. This is the may-not-end for Winter Soldier. May-not-ends is what Marvel's been doing for a while now, instead of having the credits at the beginning of the movie, they'll put it at the end, and they've turned it into an awesome design piece for companies like us to be able to create, and it also gives something for the viewer to watch while they're watching the credits, and then they always have some cool sneak-peeks afterwards. So in case you haven't seen it, let me show you Captain America: Winter Soldier. ♪ [music] ♪ So that is Captain America: Winter Soldier may-not-end titles. So this was all produced in stereoscopic 3D workflow. So when we got a hold of Marvel, or Marvel got a hold of us, we had never worked on any major film projects before. So we went through the whole pitch phase, they liked our boards, and they wanted to work with us, but they had a lot of questions of if we could do this. So could we produce this may-not-end in stereo? Of course, we're going to say yes. We're going to do whatever we can to make this work. "Have you produced any film work before?" they asked, "No, we haven't." They asked have we produced anything in stereo. "Definitely not." After those two questions, I thought we were done for, but they really liked our work, they wanted to pursue working with us. They wanted to know how we were going to wash this down. We don't have a project to see this stereo work in our office, so how are we going to see it? We ended up using Autodesk Smoke for all of our stereo playback, and then for our NN pipeline. At the beginning, it's Cinema 4D, and we did use Softimage to start with, and then we pumped all our frames through Nuke to composite, and viewed it all in Smoke. So there is a few things going into this that we really were wondering about. Obviously Marvel had concerns about working with us, but we had concerns because we had never produced stereo before. So a few of our concerns were, "How do we produce this? Do we produce this natively in CG like Cinema or Softimage, or do we fake it in something like After Effects, something 2D or 2.5D?" They both have their pluses and minuses. With native CG, we have real-world scale, so we're able to create actual figures the actual size that they are. We can use actual measurements, and have relationships between them be accurate, which is really great in the grand scheme of things, because when you're talking about a movie this big, if something looks off, it's going to look really off, and the best way to be able to adjust for anything off is to know the actual size of things. So native CG is a plus on that side, but with that, you have really slow renders. If we wanted to speed up renders, we could go with something 2.5D, but then we get the arbitrary. So we're just doing 0 to 100%. What's the relationship between that? That's going to be really tough. So we ended up going natively through CG, which is great. And it worked very well for us. So what I'm about to show you is just kind of set up a quick scene in Cinema, show you how their stereoscopic workflow worked for us, and this is how we've done it. It's not the only way to do it, but it's the best way to do it because we did it. There's many ways to do this. So let me dive in and kind of show you a scene here. So I just want to quickly recreate something like this Captain America: Winter Soldier end tag. So let me just set up a scene here. Just going to grab a preset here, HD, and pull in a tube. What I like to do is also set up a camera. So camera is not set up, fun little tidbit I like is to customize the pallet, and in case you haven't used this before, it's called reset PSR, which is position, scale, rotation. You can select any option for any of your objects here in Object Menu, and you'll see all these coordinates here. If you just hit PSR zero, it zeroes out everything, so now my camera is in the zero, zero mark, which is great if you want to do something just flat, straight forward. So let me just take this camera, pull back a little bit, grab the tube, let's rotate it 90 degrees, adjust some of these radiuses, pull this out. I'm going to shrink the height, we don't need it that thick, and let me just raise the rotational segments so that we have some nice, smooth edges there. We also need a name for this, so I'll go and use the MoText tool. Super helpful. Super easy to use. For all of the titles, we actually went through Illustrator and type-set them all that way. For us, we are very specific about our type. It's very important to us to get it looking right, so Illustrator was the way to go for us, and it's all paths, so no one has to have a specific font anymore. You don't have to worry about font issues if we're sending it to the farm. So I'm using here MoText, which is great, but there's, like I said, many other ways to do things. So let me shrink it down here. And, of course, if we're working on something Captain America, we need a star, so pop in a star. Don't need eight points, we need five. Rotate this bad boy, and want to extrude that, so just going into extrude. Holding Option or Alt makes it a child of the extrude, and we automatically have some depth on that. So since we are doing stereo work, this is not enough depth if we want to fly through this and stuff coming at you. We're in 3D, so let's pull this stuff and give it some depth in between each object here. So what we have now is a really quick setup. It's very simple, easy to do. And what you see is I just picked the regular old camera from Cinema's camera options. A great thing in what Cinema has done is put in a stereo camera as a default option in there. So you can literally just go in here and select stereo. One thing that's great about it is that if you forget, like I did at first here, and you forget to choose the stereo camera, you can easily go into the stereoscopic tab here and switch that over. So if you just watch this camera icon and switch it to something like symmetrical, it will switch to a 3D camera. It does not mess up any of your settings at all. It's really super simple. Like I said, we've worked with both Cinema 4D and Softimage on this project. The Softimage guys did not have the tools like this in their program, so they had to actually go in, set up two cameras, do all the math. It was kind of a big deal. So they came over to us and they're like, "All right, we set up our camera, here's our math, what do you guys got?" And I said, "I don't like math. Let me show you what I have." And I clicked over to symmetrical, and they hated me after that. So super simple set up, super easy to do. Let me pull up a little bit better-looking project. So what we've got here is a stereo project. We're flying through the O here. It's got some nice depth. It's laid out the way we want it to. And to me, it looks like it would be cool in stereo, but how do I know if this is actually going to look good or not? Well, Cinema's made that quite easy to check out as well. So making sure your camera is stereo. You can go up here to the options, and they actually have a little tab here that says Stereoscopic, and it's got a picture of the old 3D glasses. So in case you didn't see what happened there, what Cinema does is actually separate out the blue and the red channels for you, which is awesome. So instead of having to go through and output all of these frames, bring them into Nuke, get them comped, pull those frames into something like Autodesk Smoke, and view it on a stereo TV like this, all of our artists could now view it right on their monitors as they're working. So it's something super simple, super easy to do, and for me, it was really great being able to look at all of our artists with these goofy-looking 3D goggles on. We're working on something, you know, high-end, new, fresh, and they've got some old technology right on their noggin. So it's pretty funny. I love it. It's great. So now we're able to see this in actual stereo and see that we are getting some depth, and how much of that depth, and kind of mess with our options. So let me go through the options that we have here. When we first started, I thought we were going to have a ton of options to work with, it's really pretty simple. Stereo works with the left and right eye, so one image in left eye, another image in the right eye. It's the same image, but it's offset. So the offset there, the separation of those images and the angle that you see from either camera is how your brain determines how much depth there will be. So the right amount of each other, there's going to be no depth. The more you separate that, the more depth you have, the more you'll see in the background, the more in the foreground. A cool thing I want to show you over here is...let me pull up this other one, and you'll be able to see the cameras here. I've got to turn the camera back on. If you turn this back to mono, you see one camera, if you turn it to symmetrical, it actually shows two physical cameras there. Let me delete this one. So you can actually see that there is two cameras, the left and the right eye being recognized. So the main option that we were working with was eye separation here. It's defaulted at 6.5 centimeters, and in case you did not know, that is the default... "the default." Humans are default. It is the average distance between most people's eyes. So unless you're a sloth, or someone from the Goonies and you have some massive displacement there, most likely you will have a 6.5 centimeter distance. So it is kind of a good starting point, but it is not by any means the default "everything works" kind of way to go. What we've noticed is when we're working and looking at this on our screens, most of us have small, 17-inch, whatever, 20-inch monitors, we see it and there's a certain stereo depth that you get when you're viewing it that way, but when you project this, like they are going to be on the final project onto a 40-foot screen, those, what you think is a minuscule amount of depth, can be huge. So working with Marvel, we've learned a lot, and this has been a constant process with them just because we don't have the ability to have a 40-foot projection screen in our office. So what we like to do is...the artist will be able to, on their screen, through these glasses, be able to take a good look at the stereo, find a good option that they like, and then we give them what we call a wedge test. So we'll give them something a bit more depth and a bit less depth, and it kind of gives them the option to either pull us one way or another. Usually, we give them way too much depth, just because it looks awesome. I mean, you want stuff flying in your face. So like I said, the main option, or the main thing we changed here was this eye separation. So as you can see here, we've got a 6.5 centimeter distance, and you've got some good separation here. But when you get closer, this separation becomes wider and wider, and when watching that, it's going to hurt the viewer, and it's going to tire their eyes, because their eyes are trying to focus, and they start separating out. It's either separating, or becoming more closed-in. So MAXON made it super simple, we can key frame that, and that was a really helpful option over a lot of other programs that we were using other than Cinema. So we're able to just go in here, tweak it down, key frame that, pull back out, and now we've kind of lost that separation. So we can just straighten it back out. And this is where I'm talking about, that the 6.5 is default, but really when you're working with images inside of CG, there's no default. It's really what works at the time. So it is a real world application, we're using real measurements, but it's kind of all over the place. So it's really simple to set up, really easy to use, and for us, it was really great, because we were able to focus on the design side, and then when we had to create this in the program, it was easily done. So we were thinking to ourselves, "This was no problem. We can do this. Let's do another one." And that's why we started Ant-Man. So Marvel liked working with us, we liked working with them, and so, we decided to take on a bit more of the...a bit bigger of a project. So instead of this stark nature, very small amount of color pallet, we ended up going with something a bit more complex. So in case you haven't seen it, here are the Ant-Man may-not-end titles. ♪ [music] ♪ So as you could see, that was way more complex than what we had done before with Captain America. So what we had with this was more complex camera moves, more scene elements, more multipasses, and center eyeballs. Let me explain center eyeballs quickly for you. So for Captain America, we did create it in stereo, and it uses the left and right eye. Being our first job, we wanted to know...there's two sets of deliveries for the may-not-end titles. There's a 2D delivery, obviously non-stereo, and then there's a stereo delivery. So what we had done, and what we were recommended to do is for the 2D delivery, just render out the left eye of the left and right eye, and deliver that as the 2D delivery. Now, what happens is because you do have two cameras that are separate from each other is everything that you might think is centered, especially that end logo, is a bit offset. So we had to go back in and readjust certain scenes specifically just for that offset. So for Ant-Man, we were like, "We are not doing that." We made use of the center camera, rendered that out separately, and then had our own separate left and right eye. And you are able to do that so simply, like I showed you, with just the click of a button here. So we can render this out mono, just the center camera, and then symmetrical, and we don't have to adjust anything. Wish we would have done that a bit earlier. So where are we at? So there's a lot more options, a lot more issues that we can come up with, and one of those issues is stereo rivalry. What stereo rivalry to us is when the image in your left eye is not there, or is not the same as the image in your right eye. And let me show you an example. There we go. So here I made a pretty easy, simple kind of star scene. Flying through it, looks like it's going to be great. Let's ship it. Client loves it. This is it. We're done. It's never that easy. But as we go through, you'll see some of these objects come pretty close to camera, we'll want to check this out in stereo. So what we think looks good right now, let's say right here, what we think looks good right now in...if we check it out in stereo, is not so good. So what we want to see when we're looking through this is a left and right eye, so that would be we need a blue and a red channel. We're only seeing the blue, there's the red, but because of the eye separation, when it's going past us so quickly, or so close to the camera, we start to lose one of the eyes. So when you're viewing that, this is another example of what's going to drain and tire out the eyes of the user is that you start to see images in one eye and not the other. So it just kind of freaks you out and it startles you, and this comes by so quickly, and you're trying to figure out what that was, and it was a real pain for us to try and figure out how to get rid of these options. But what we ended up doing is rendering out a bunch of different passes for everything so we we took...let me show you. So we rendered out separately the main objects here, we rendered out the dust separately, all of these things separate. It's kind of a large process, but it was definitely worth it to have that much control. So before I dive deep into kind of how we spline this, I'm want to show you a bit of our process...a bit of our process on how we go through from start to finish. So when we talk to Marvel, we go through a pitch phase, we send them a bunch of ideas with the story board, they end up liking one of our ideas. So the first thing we do is say, "Hey, these were in still frame, can we create this and actually make this look cool in motion?" What happens is that our artists use a bunch of different programs for making storyboards, so After Effects, C4D, Photoshop, you name it, it's used. These happen to be made in After Effects. So it's really easy to try and create an animated mock-up. So this was all created in After Effects, and as you can see, it's pretty close to the actual final piece, but it was done in After Effects. So if you remember what we lose when we're doing it in 2.5D is we don't have that scale, we don't have those great measurements that we're able to use to our advantage. So as much has we could send this to the client and say, "Hey, here you go. Yeah, we can do whatever you want, " they don't know we were faking it, and...well, now they do. But we need to find a way around that. So during this whole process, there's two separate teams going on. There's the R&D team trying to figure out how we're going to do this in Cinema, and then there's the actual animatic team trying to figure out the design and framing of each frame. So here's a really early-on animatic phase. We just want to show them here's what each scene is going to look like. Take a look at that. The pacing is okay, it's all right. It's not totally there. We've got the titles up, but we really want them to focus on what the scene looks like. So you'll see here we've got some scenes that are there. There's another where we go into a bridge here, and then we were trying to go's a highway that didn't make it in. Trying to go into the trees. It's a constant ongoing process. I can't tell you how many of these options we have. So after we get sign-off on the sketch phase, what we do is we actually lay it out in CG. So this is still super rough, but what this does for us is give us the ability to set up cameras, set up timing, set up placement for everything in this environment so that if there's any tweaks, we already have the setup, and we're much further along in the process. So after we get this kind of rough sketch layout, and hopefully they approve it, we get into the actual spline look. So all of this, much like Captain America, was done in C4D...let me take that back. So all of the effects and everything were done in Nuke, but what we had done this time different than Captain America was we had taken...this is about 95% C4D. So Captain America was was a bit more Softimage, a little less C4D. We saw the power in stereo camera workflow, and just the workflow in general, and we went full-on C4D, so there might be one or two scenes that were Softimage, but really, this is a C4D-heavy project. So like I said, here is where we ended up before we take it into Nuke and create this nice look with all the glows and everything. So working and trying to create this look of this spline look that has some you can actually tell what it is, but it's still organic, took quite a while to do. We tried a bunch of different options. We tried X-Particles, like we took an object and used tracers, and tried to have it wrap around the object. It looked cool, but we just didn't have the real...we couldn't control it as well as we wanted to. We tried projection-mapping splines on there. That kind of got us there, but it really wasn't there. So the way we found out how to do it was actually just doing it by hand. Let me show you how we did that. So if I take an object, we'll just grab a sphere. One of my favorite things is these protoplasm kind of things. We actually did a lot of research for the molecular level, and in case you didn't know, Ant-Man is taking off of a video from the '70s called Powers of Ten where it's kind of talking about the spacial distance between us and the inner body and the molecular level, so it keeps going in. It's a really cool video, especially being made in the '70s, but keeps going in 10 times more, 10 times deeper and deeper. So we kind of went with that because Ant-Man, he loves to shrink himself like an ant, so it makes sense. So like I said, did a lot of research for this area, its molecular area, but if there's any doctors or people in the house, surgeons maybe, that can tell me if it's looking correct. No? Didn't think so. So to create something similar to that, we're just going to use something similar, something easy, like a Displacer. Really simple to do. Can make a lot of cool stuff. If you go into this custom shader, you can pick a noise, and we can pick any noise. Pick a noise, any noise. And just go into the Displacer, we want a lot more displacement. So just grab this height under the object tab and just pull it out. Now we've got something cool. Sure, it's not super scientific or accurate, but it gets the point across. So what we used, instead of just wrapping a spline around it, we found out that we could use the hair tools. So if we just pull out this menu, there's an option here called hair guides, and what that does is allow you to pinpoint where you want the hair to be. So most of the time, what I was familiar with, is just throwing on some hair and the whole object is just boom, it's got hair all over it. This way, you can pinpoint exactly where you want your hair to come from. So we're going to shrink down these radiuses. Right now, it's going to pop...come on. Oh, one thing I forgot to do. This will not work unless this is it's own object. So right now, we're using a Displacer on a sphere. We need to create an object out of that. So if I just do a simple currency-to-object, it creates this object out of geometry. So now we have this here. Now we can paint some hair on there. As you'll see, like I said, we need to shrink this radius down. So I just click add guides, we get the option here, and we only want one at a time, one hair at a time. It's how we do it. It's almost like painting. Painting hair. If it was only this easy in the real world, right? So we can kind of wrap around, and what's great is that these hair guides wrap onto the geometry in every little crevasse. So we're able to wrap this very nicely. But we're wrapping it in hair, we don't want it in hair.we want this to be a spline. So we were looking all over for it, and there's a guy named Jaider who... I think it's Jaider. We found a script online that actually took these hair guides and turned it into splines. So it actually finds your first point, takes your first point, and follows in order that you've created, and makes a spline out of it. So I will pull up the script manager, open the script up. It's super simple, you can find it online. Where are we at? Here we are. Ant-Man hair. Pull up the script. And super simple thing to do, super easy, is they have this icon here, we can also pull that up in our menu so that we don't have to keep opening that up. So all I have to do is select the hair, select all my points, hit this, and what it does is create an actual spline in the placement of all those guides. So now what we're able to do is go in with these crazy-looking objects, or with the more realistic-looking objects, and create an organic feel that we have control over without having to use something like X-Particles or something where it's just kind of more random. So we can give it that kind of random feel, organic feel, and have the ultimate control over it, which is great, because that's what Cinema has been giving us, is ultimate control over everything. So what we also have been using R17, there's a system...has anyone used takes before? Is everyone on R17? Anyone? Okay. For these other projects, for these last two, we are using R15. R17 came out, and the take system came out, and this was a saving grace for us. For this next project that we were using it for, we had actually used it to separate the passes, so we actually...if it pulls up. You go into the take system here and you can see...what we used was for hero shadow, text shadow, text beauty, you can really separate the passes, see what each pass looks like in the system without having to render out a frame as a whole, and how that's beneficial for us is through the stereo process, like I said, for Ant-Man, we had to render out the dust separate from the plane, from the clouds, from each individual piece because of that stereo rivalry we get, we needed control over that. So this way, we're able to take a look at that using the take system, and control the stereo eye separation of each object in one project. So we don't have to save out a project for the dust, and save out another project for the plane, save out...just saves us time. Keep it all in there using takes. So I will show you a little bit of how the take system works, and then another option. This is a simple scene, just a bowling scene. We've got shiny balls here. This is how I see it using it in a real world application for...we do a lot of advertising. It's just kind of creating simple options. So if I were to create...this is just simple dynamics. First of all, we get into our take menu, you got a main take. A main take is where everything starts, so the default. So if you want anything to be changed across the board, you need to change it in your main take. All we have to do now, if we want to start this take process, is go up here, there's a New Take button. So you get into the new take, I like to keep on Auto-take, but you'll see when we get into...there we go. So we click on Auto-take, all of these options here turn blue, and what that means is it's recording all of your options. So we can go in, change any of these options, and it's going to record them in this take. So what I'll do here is take this ball, and I like to also keep auto key frame on, super simple. So we'll start a bit back, move forward in our timeline, and just kind of push it through. As you'll see here, push it through. It's kind of a weak toss, maybe a little baby toss, little baby bowling. Let's say we want to go a little crazy. Let's go deeper. Balls-deep into this bowling option. What we can do, instead of saving out a new project, we can actually create a new take, it resets us, in case you didn't see. Let me delete that take. Here we were in this take. Create a new one, the ball's shifted, we're back into the default main take, and now we can just do it all over again a different way. So we've got...make sure we're in this take, grab a bowling ball, key frame is on, start back here, get up here, and we're going, going all the way. So we play this down. Much more dynamics, another option. And the simple thing about this is you can just flop through the take system so easily. So we've got the short one, we've got the deep one, we can keep going through back and forth. So not only can you switch out just animation, you can switch out colors, you can switch out anything. So let's just make a really quick change here to the colors. Let's make a green one. There we go, and turn our balls green. Now we've got green balls. So in this take one, we've got green, switch it here, we've got blue balls. So we've got total control over everything. So this has been even more exciting for us. So we are total control freaks at Sarofsky, and for us to be able to have a total master control over this has just been so great. So I would suggest if you haven't gotten R17, it's definitely worth to check out just for that take system alone. So I am pretty good right now. So thank you for watching.
Resume Auto-Scroll?