BACKGROUND

What initially drew you to the world of VFX and film?

Thrain: I've always loved visual effects and seeing films with visual effects in them, ever since seeing Star Wars when I was a kid. I didn't think I would actually end up working in visual effects, I originally trained to be a graphic designer then got into computer animation and spent many years doing commercials and computer animation for television. Eventually I started doing the same kind of thing for films, which took me into visual effects, which I thoroughly enjoy.

What are some of the features that you have previously worked on?

Thrain: Recently I've worked on Charlie and the Chocolate Factory, The Hitchhiker's Guide to the Galaxy, and before that King Arthur, which were all very different projects with very different challenges, and the same again with V For Vendetta. Some films are very surreal and the effects are very much in your face. Something like King Arthur is more about invisible effects. On this film things like the crowd scenes that we're doing are not really meant to call attention to themselves as computer graphic shots. Having said that, there are a number of shots in this film which are a bit more visible as effects, like the knife trails that we're putting in for when V is doing his knife fights, which are meant to call attention to his slightly super human powers.

MULTIPLE Vs

What was your first introduction to V For Vendetta?

Thrain: When I first heard about the project I was quite interested in it because I knew the name Alan Moore. I had read Watchmen many years ago, but I didn't read V For Vendetta until I heard about the project, and I thought it was very interesting. Then I started getting involved in the actual film, and there's definitely a very interesting story, and lots of different things going on. It's not really a big effects film, but there was a nice opportunity for doing some interesting things within it.

What were some of the challenges you had to tackle initially?

Thrain: The biggest thing we knew we would be doing was crowds of computer graphic Vs. I think we were lucky in that, normally, when you're doing crowd stuff you need to have everybody look as different as you can, but in these shots the idea was that they're all dressed up like our hero, so that made it a little bit easier. At the same time, I think one of the biggest challenges with that was the fact that V has such a distinctive look and a distinctive costume, so we had to match that very, very closely. The hardest bit was probably his cape, which is a very distinctive part of his character.

We had to come up with methods for creating simulations of that movement of the cloth, and we had to do it on quite a big scale as well. There's one shot where we've got about 10,000 characters running around, and they each have a cloak, and they all had to be moving realistically. Depending on the shot you can see more or less of that detail, but there are a few shots where you really can see the capes moving independently and realistically. That was probably the hardest technical challenge.

How did you ensure the fabric of the digital cape looked the same as the original?

Thrain: The props department kindly lent us the whole costume, so we'd take the material and look at it under different lighting conditions, and we did a fair bit of R&D in what we call shading models, to see how something like cloth realistically appears when light catches it. Things like his hat and his cloak are very velvety matte materials, which are traditionally difficult to create in the computer, but I think we successfully got our lighting and shading to react in a very similar way to the real thing, so our computer graphic Vs look very close to the real one. Most of the crowd shots are very distant, and the characters you see are very small, but in order to calibrate what we were doing to the real character we actually set them up at about half a full screen size and compared them to some live action reference of the stunt man dressed up in the costume. Basically we used that as our goal, and matched very closely to that so we could be certain that it would look like the actual character when you saw it on screen.

Did you design a number of different ways the capes would move within a crowd?

Thrain: Yes, that was a key to it. You have a number of different people in the crowd ‚ different heights, different weights, and that kind of thing ‚ so we'll do slightly different speeds of walk and slightly different actions. You don't really want their capes to be doing all the same thing, so that's why we had to do what we call simulation, where the computers work overnight and when we come back the following day we basically have a simulation for each cape on each character, reacting to their particular movements and the physical forces in the scene like gravity. That means we come out with a completely different animation on each cape.

Aren't the thousands of Vs only made from ten or so humans in motion capture?

Thrain: In some of the shots we've probably got at least 10,000 characters, which are then replicated from about 20 different models of different heights etc. We got both female and male models, although you can't really see that because they're under the costume, but I think it has some effect on the visual appearance. In some of those shots every cape will be doing an individual animation that has been simulated physically correctly.

We can make characters go around something in the middle of the road and have the cape react in a realistic way. Depending on the shot, you see less or more of that. There are only a small number of CG crowd shots in the film, but there's one that's quite close where they're walking across a bridge, and you really can see their capes flopping around and interacting. We've looked at it countless times to try and spot all of the little things, and the Visual Effects Supervisor [Dan Glass] will also go over it with a fine tooth comb. You do really have to pay attention to detail.

For the sequence where all the Vs converge on Trafalgar Square, how much work has that shot gone through before 3D begin work on it?

Thrain: Those shots were helicopter plates, which had a bit of a movement on them that needed to be changed. Also, the time of day was completely different, it was shot at dawn, whereas it had to be at midnight. So that plate was taken, locked off, then matte painted to give it the appearance of being at night, with light sources put in. Once we had that, we could really start our work. We started blocking out the animation of the crowds, and started working up our lighting to match. It's not quite a one, two, three process as it seems when I say it, for instance this plate kept evolving as our work evolved, and our lighting continues to evolve as the matte painters improve what they're doing.

Did you have to create a vector-based model of Trafalgar Square initially to work out the movements of the Vs?

Thrain: We had some rudimentary buildings and geometry (which we didn't render because we had a plate to work with) and we blocked out the crowd system, which is software that has been developed by Cinesite, to do these massive crowd animations. If you want the crowd to avoid a certain area you can put in an object that repels them. In Trafalgar Square there's Nelson's column, and you don't want them to be trying to climb up that, so you tell them to avoid that area of space, and they'll walk around it. Or you can mark out a certain area of geometry as something to attract them ‚ or maybe block them - so when they come up to the police barriers they'll stop, and go into their idle animation, or something like that.

When you animate a hero character, you have direct control over the animation process, whether by using key-frame animation or motion capture. However, when animating a crowd of 10,000 people, there is no way you can physically animate each one, so you have a process whereby you indirectly influence groups of these 'agents', and then let the computer create a simulation of what they'd do. That's what our tools give us. When it comes down to the actual animation of the crowd of what the little characters are doing, that's built up from lots and lots of different little clips of motion capture data, which are being blended by the crowd system at the appropriate time to tell the little characters what they should be doing.

Have you built a detailed V character?

Thrain: Yes, as part of our process we are building the digital Vs for the crowds, so we did build in a high level of detail so that we were covered for all eventualities. The crowd software can then reduce the amount of detail in the characters to fit the appropriate shot, because if you've got 10,000 guys you can't have them all with the same detail as a hero character in the foreground.

Did you do any 3D work on the mask?

Thrain: We did do the mask - you don't really see it particularly clearly in any of the crowd shots, but if you zoom down to one of the characters in the crowd, they have a very accurate representation of the mask. We created that separately, and then we put it onto the character. We actually had a high resolution digital scan of the mask, right down to the little dents in it, because we didn't know whether we'd be doing a digital V. It was a possibility, so we put quite a lot of work into getting that mask looking quite accurateÖ if you look really closely in the crowd shots you'll be able to see little white masks down there.

VICTORIA STATION

What is your biggest sequence from a 3D point of view?

Thrain: The biggest sequence would definitely be the Victoria Station sequence, which involves quite a number of different things. Probably the biggest thing is that we're putting in some computer graphic knives. In about three or four shots knives need to fly through the air, which would've been either physically dangerous or impossible to do in real life. In addition to the knife itself, there are quite a lot of shots in the sequence where there's an effect being added to what V does with his knife. It's a slightly surreal effect, but it visualizes his superhuman speed. The shots are generally in slow motion and the knife leaves a trail like it's cutting through the air, which is again being created in 3D. We've tracked where the knife is going through space, along with the camera, and then created a program that plots that track through space and creates a trail behind it with various animated textures on the surface to give it the look they wanted. That is then taken and composited by the compositors on the plate, and they also add quite a few different effects to it at that point as well.

Where does 2D end and 3D begin for the knife sequence?

Thrain: The knife trails actually start in 3D, and are then manipulated in 2D and combined with the live action plate. The compositing ‚ the 2D side of it ‚ is generally about bringing together elements that already exist. For instance, having elements like live action, miniatures, or computer graphics, and blending them hopefully seamlessly. So we [in 3D] give the compositors the trail elements that they mix into the live action and treat in such a way so that it sits in there looking natural. They also add more 2-dimensional effects, but it does start with a 3D element that has been tracked in. What we render out of 3D at that point, in this particular case, doesn't look like how it looks in the end because it's a bunch of different layers that are then combined in a creative way in the compositing software and mixed back over the live action, as opposed to something like the knife which, when we re-render it in 3D, looks like a knife. In that instance the knife pretty much goes on over the live action how it looks when it comes out of the 3D software.

How difficult was it to accomplish the look of the trails ‚ something that is almost intangible?

Thrain: I think the trails were probably the biggest creative challengeÖ with the knife, we knew what it looked like, there's a close up of a gun later on and we knew what that looked like, we knew what the crowds looked like because there was plenty of live action reference for that. The trails were something that had never existed anywhere, and it's a fairly unique looking effect, so we had quite a long process where we tried a lot of different looks. Initially in 3D ‚ and 2D ‚ we combined many renders of the knife itself, seeing how that looked with trails of different kinds of particle effects. What we eventually came up with was quite influenced by some of the early concept art we were given, which is a little bit more like something from a manga film or a comic book, with the lines they draw after a sword slash, but taken into a space where it looks like it exists in the real world as well. Like it's something that is affected by the light in the scene and the other things that are going on in the shot.

What was the process of getting the leather and metal of the real world knife onto your screen?

Thrain: With the knife we started out with a laser scan of the real object, so we had a very detailed representation within the computer of the actual size and shape, which we then rebuilt into a more efficient representation inside the computer, and textured it from photographs of the actual prop. We used that as a basis for texturing the actual surface of it, and from there it became a matter of dealing with it with our shading and lighting software. We did research into different kind of shading models for metals and found some shading models based on real world measurement of the way light is caught on metal and various other surfaces, so we were able to get quite a nice realistic feel.

The other thing that contributed to that was that when the movie was being shot, Dan Glass the Visual Effects Supervisor, also made sure he captured some nice images of the set that are a combination of multiple fish eye lens images, which we then turned into what we call an environment map. That is an image that creates an environment in which the knife needs to sit into - it's like a 360ƒ representation of the live action plate with all its different levels of light and dark. So when we put the knife into the scene, it's lit by this special image and also reflects that special image, and it gives us a pretty accurate representation of what a metallic object would look like.

How difficult were the decisions on how fast and tight the knife spins, along with how high and low its circles are?

Thrain: There were a lot of different versions of animation done for that, so we could get exactly the feel that they're after. There are a couple of shots ‚ one that went into the first trailer where you see two knives flying through the air next to each other, describing a certain kind of pattern in the trails - that was quite deliberate, to have this almost helix pattern evolve from the slightly different speed of the rotations of the knives. There's another shot where V throws a knife almost directly at camera, and it flies past and hits one of the Fingermen in the head. Again, there was no knife in that plate to start with, and the plate was very different. Initially, we came up with the best animation of the knife that was required by the client, and then retimed that whole plate to match to the animation of the knife, so that the camera move almost became a synthetic shot, just so we could get the exact kind of animation on the knife that was required.

So the knife animation was a completely different process to animating the crowd?

Thrain: With the knife you would call the kind of animation we're doing with that key framing - where an animator sits the different positions of the knife explicitly in space. With the crowd, because we're dealing with so many individual things, you can't realistically sit there and key frame all the little guys, so we do that through a process of simulation where you're influencing all of the agents inside the computer. Computer graphic animation tends to fall into two camps a lot of the time: that is key frame animation versus simulation.

How long did it take to create the knife as a synthetic, digital object?

Thrain: We knew it had to be seen very, very close, it's iconic of the character, so we probably spent a good month of one person's time taking that, remodeling it, texturing it, and getting it set up so it looked good on the screen. There was a bit of further tweaking once we actually got shots through and put it in the shot and in the lighting conditions of that shot, so for what seems like a very small object, it can take quite a while.

What are the gunshots you're working on?

Thrain: There are two shots in the Victoria Station sequence, towards the end, where one of the characters is firing a gun at V and he runs out of bullets. So we have two synthetic shots that are inside the gun chamber where you're seeing the bullet rotate around. You're looking down the barrel of the gun while that happens, and then in the second shot you're right inside the barrel and there's no bullet anymore. Those are computer graphic shots that graphically show that he has run out of bullets. It's a slightly surreal abstract way of looking at the gun, and putting you inside it. A live action plate was shot with just V walking towards the camera and we put that over the top, basically.

OTHER VFX

Have you worked on any shots that might not be apparent that they've gone through VFX?

Thrain: There's at least one shot in the Victoria Station sequence that has a computer graphic knife that I think you'll find very hard to spot, because it's not one of the hero ones where it's flying at you. We had to put a knife into V's hand because there was a shot where V stabbed one of the Fingermen with his prop folding knife and it fell out of his hand and potentially spoiled the shot, but we kept the shot and in 2D removed the knife falling out of his hand and put in a new knife into his hand as he continues through the rest of the shot.

We've got another shot that we're working on at the moment that involves slow motion rain, which will probably look like some kind of effect, but I think what we're doing to the plate itself won't. We had a very wobbly crane camera down to Natalie Portman's [Evey's] face and it's become an entirely synthetic shot. We're replacing the background behind her and we're taking a few frames of her, creating almost a morph of what she's doing to get exactly the right combination of camera and action and what's happening with the CG. We're throwing away the original camera move now, and recreating it as a combination of 2D and 3D take, because we've been given the direction to create a totally linear smooth and slightly surreal kind of feeling with the camera. We follow a raindrop in close up into the shot and it falls away from us, and meanwhile there's a lot of other rain falling down as well, it's going to be quite a nice shot, I think.

That has evolved over the course of time, partly because of limitations on what you can shoot. It would be very difficult to get a real camera to travel smoothly, have live action rain falling, and get the right camera speed all at the same time. In this case, because of the weave and the camera and the timing changes they want, it has become something that's easier to recreate it as an effect, rather than trying to keep what was there originally.

Does it affect your work, having more frames in the shot now?

Thrain: Not really. It's got a bit of a slow motion speed to it now, something like 125 frames per second, and I think it was originally shot at about 48 frames, so we're slowing it down further by doing quite a large move in that time, because we're falling with raindrops, which are falling only a little bit quicker than us.

VISUAL EFFECTS

How many people are in your team and how long have you been working on the project?

Thrain: In full production mode, we've been going for about six months. Over the course of that time our team has grown in size up to about 20 people at its peak - on the 3D side. From tracking and matching on through to animation, the technical development, lighting, and rendering. There are also a lot of people involved on the 2D side. This isn't a huge film from a 3D point of view, really. There are probably about 3o to 35 shots that are 3D, there's probably a lot more work in the compositing. I think there are about 150 shots altogether, and all of those shots will have some kind of 2D work in them.

In your team you have mentioned trackers, lighters and animators; does everyone stick to their job, or do they cross over?

Thrain: Most major facilities that do film work tend to have a more segmented pipeline, where people concentrate on what they do best, but that's not to say people don't have multiple skills; people do crossover. Tracking, for instance, is very important. We have a small department that concentrates on that because pretty much every time you want to put some computer graphics into a shot you need a camera that's matched to the live action camera. Whether that's static or moving, someone needs to do that, so we have people who specialize in working with the software that will track shots. In creating 3D models and animation, most people are fairly specialized, but we do have people who cross over the different tasks. On a big project it's often simpler to break it down into different tasks that can be done by people who are very good at them.

How has VFX changed since you've been working in this business, and where do you see it going in the future?

Thrain: I think it's definitely changed a lot over the past decade, from what I've seen. I remember seeing the very first computer graphics going to films and being quite amazed by a lot of computer graphics at that point. Apart from a few companies in the United States, visual effects were generally done on more of a broadcast level. Then Jurassic Park came along, and that made everybody realize how much could be done with computers; they weren't just a gimmick. From that point on, people really thought you could push the boundaries.

It's very difficult to imagine where we're going be in another ten years. There are a lot of things we haven't thought of yet, and I think it's almost impossible to try and predict that. There's a lot of work being done right now with digital doubles of people and things like that. I think you're going to see more and more of that, and people are going to push that further. Whether you'll see that goal of the entirely computer generated actor, I don't know, and I'm not sure there's a point to that.

Do you think filmmakers will continue to combine methods of shooting with VFX, as for the models on this film?

Thrain: Unfortunately I think we're going see more and more things that people once used to shoot move into the digital world, even if there's not necessarily a good reason for doing that. You see it more often now, partly because of budget reasons - it's got that much more expensive to shoot huge crowds of extras, so we're doing them digitally. And it's expensive to build big outdoor sets, so we do 3D matte paintings of buildings. These days people shoot scenes with huge cables and wires in them, because they know they can be removed later on.

At the same time, I think there is potential for that to enhance people's creativity. There's a lot of recreating sets entirely in 3D now, so that you can free up the camera in post and come up with new camera moves and things that weren't imagined at the time, or maybe weren't physically possible on set. If you take enough reference of the set when it's built, you can put it on a 3D model.

The actors aren't in that environment though.

Thrain: No, but you can always blue screen them in later, or make a CG person for some shots, or alternatively you could shoot it initially. I think there's a move to see more and more stuff being done in post and after the shoot.

Do you feel that even green screen is becoming less necessary?

Thrain: It really helps, because otherwise somebody has to paint a matte, and the quality is better if you prepare rigorously - the quality you can achieve is going to be better. But of course there's cost to that in flexibility and speed in which you can shoot, and you can't just change your mind on the day when you're there, and I think a lot of directors now want to take a broad kind of flexibility that you can't get on set. Within realms of possibility you can track almost any shot in 3D now, and put something into it, which even five years ago you couldn't, really. Someone would have to do it by eye, and try and match it up, but now there's some pretty good automated software that, with a bit of tweaking, does the job.

Digital effects is a technology that is improving, it is letting you do more quicker and more realistically. Having worked in the visual effects industry for some time now, when I see a bad shot I don't necessarily think that they've done a bad job, or that they didn't know what they were doing, because I know so many other factors can come into play that aren't related at all to any kind of technical thing. It could be time constraints, it could be unusual requests from directors or other people involved in the process. A lot of times you might animate something and you'll do it to what you know would be a physically accurate representation, but people aren't always after reality. Even some of the moves of the knife in V - we looked into the way physics of knife throwing actually work, and V doesn't always throw his knife the way a real knife thrower would throw it. He does it in a way that looks cool and cinematic and works on screen, but doesn't have a lot of bearing to reality, necessarily.

Thanks Thrain.