The new "Where amazing happens" spots feature famous NBA playoff moments starring Larry Bird, Kobe Bryant, Dr. J, Magic Johnson and Shaquille O'Neal. Empty stadiums are gradually filled (and emptied) with players and spectators as the iconic moments play out, Ian Failes speaks to Brickyard VFX about the task of working with such iconic clips.
Brickyard VFX completed the visual effects for four spots promoting the 2009 NBA Finals Championship Series. The spots, conceived by agency Goodby Silverstein & Partners and directed by Jonathan Dayton and Valerie Faris of Bob Industries, feature famous NBA playoff moments starring Larry Bird, Kobe Bryant, Dr. J, Magic Johnson and Shaquille O'Neal. Empty stadiums are gradually filled (and emptied) with players and spectators as the iconic moments play out. Visual effects supervisor Geoff McAuliffe tells us how the spots were done.
fxg: The spots gave me a strong sense of nostalgia about the NBA. Was that a pretty clear intention going in?
GM: Definitely. One of the things the guys really wanted to come across in these spots - and we looked at tons of clips before we picked these ones - was that we wanted this emotion to come over you. This place we were showing was like hallowed ground where something really special happened. We really wanted to get that emotion in there and I really did think we achieved that. For us, working on it for so many bloody weeks, you can lose sight of it, but it really does seem like people are affected by them. You do get a real feeling from them.
GM: We went through stacks of clips, just old footage. Usually they had multiple camera angles too, because obviously it was shot live with cameramen all over the stadium. There could be three or four angles for each shot. But, boy oh boy, it was hard to get nice shots that would work for what we wanted to happen. The other thing is, the resolution on them was terrible. It was all video resolution, and we had to slow down the shots so much. So the resolution was a real bloody issue. But eventually we were able to go through the process, pick some shots. Obviously we liked some that the NBA didn't necessarily like. Jonathan Dayton and Valerie Faris were involved in helping us pick shots and in giving us some direction about which ones worked, which was great. So between all of us and the agency we were able to whittle it down to four clips.
fxg: What format were they in?
GM: They were all just NTSC video clips, on Betacam. It was really crappy footage. One of the clips had been shot on 16mm film, which was amazing - that was the Dr. J spot. That was the best quality one. The others were really, really terrible. That was the biggest problem, because we knew going into it that the spots could never be perfect. So we had to rotoscope everyone out and you couldn't even tell what you were rotoscoping.
The background people were just a smear. You watch it and your brain says, 'Oh, they're basketball players on a court. There he is, throwing the ball...'. But really, the ball isn't a ball that you can cut out. It's just a brown smear. You could actually see all the audience through the smear. So when you cut it out you can see bits of audience in the picture and it looks horrible! But we knew this going on, and we knew we could never spend the time on these spots getting them perfect. We knew we'd just have to work as much as we can on them until the deadline, and we'll just have to drop our pens then.
fxg: How were the spots conceived? Were they always going to be famous plays with the players appearing in and out like that?
GM: That was exactly the concept. It's funny - we did some spots last year which were the split screens - the two sides of the theatres. At the same time as we were doing tests for them, we actually tested this same concept. We weren't really sure then that it was going to be famous plays or just other ones. The one that we tested last year was Kevin Garnett when the Celtics knew they won last year. There was an excellent clip of Garnett screaming his head off in the stadium. Basically, we emptied the stadium and had Kevin Garnett just standing by himself and then had everything form around him.
That was the first test we did last year. But when we went into it this year, we knew it was going to be exactly as the clips play out - just these really special plays that people knew very well. That was the crux of the whole campaign. It was these temples where these really special moments happened that people have remembered. It's crazy. I had a client come in and he saw me working on one of these spots and he knew every word of the commentary from the spot, because it's such a famous moment. For me, I don't really know that much about basketball, I have to admit, being from Britain. I didn't really know how special some of these moments are for people. They're a big deal to some folk.
fxg: Once you'd chosen the plays, what sort of work was involved in starting the shots?
GM: The first thing we did when we worked out which clips they were going to be - we had to do a previs. We very roughly rotoscoped all the guys out and we had to change the camera move at the head and the tail of the shot, because the camera moves on all of these spots were all handheld. It's crazy. It's always following the ball and the player. So for the head and the tail of each spot we had to ease the camera, as if the cameraman was in an empty arena and he was walking through it, and then suddenly something starts to form before his eyes, and then he follows. The camera moves at the head and tail of the shots are completely manufactured. We had to work out how to best take the real camera motion and then add these slow camera move clips at the head and the tail to make the spots work. We used previs until we got the right formula to work for these moves. We also built a very basic greyscale stadium behind for the previs. We worked out the fake camera moves in Flame.
Part of the previs stage also involved slowing the shots down as much as we could. We had to work out how slow the shots would be, and the timing for everything. It had to be pretty perfect in the previs before we started any work, just because there was so much hand work in rotoscoping, eventually painting and then tracking to this new camera move that we'd built. So the previs' were very important.
fxg: What was the next step?
GM: Once we got the previs bought off on, the first thing we'd do was to get our team to rotoscope the players as best they could. That involved spline matting around the players and giving the Flame artists some mattes that they could use to do the composite. At the same time, we had to make sure the stadium was tracking correctly. A couple of guys spent quite amount of time doing that.
They were quite difficult because of the crappy state of the footage. It was all old, blurry footage so it took some time to track. Then the stadium was all modeled in Lightwave. A little of it was done in Maya too, but all the rendering we did in Lightwave. Some of the stadiums don't exist anymore, so we had to try and get old photos of them to set up the shape and the colour of the chairs and all that sort of thing. The Boston Garden, where the Larry Bird spot happens, doesn't really exist anymore. It's a whole different building now.
GM: Exactly. It was funny because we got all this old archival photographic reference and it was quite obvious that things got moved around quite a bit as it didn't match the shot that we had. Not perfectly - certain things did but maybe a lot of the chairs had been mobile because they were in different positions when we looked at the shot. We didn't have to get it 100 per cent right. We just had to get the key elements in the right place and then there was a little leeway with the groups of chairs. Just as we dissolve the stadium in and out, certain things had to match up when they dissolve through - the hoop, the position of the floor, for example. Luckily the chairs were being covered by people so we were able to cheat a little bit there.
fxg: How much is CG? Just the stadium? What about the floor?
GM: I guess you could say the floor was CG, but it wasn't done in Lightwave. It was tracked in. We did a matte painting of the original floor. So we used photographs of the real floor, but it was tracked in Flame using the camera move and then the real reflections added back on top. We had to replace the floor because of the extra camera moves that were added at the heads and tails of the spots.
fxg: How did you overcome some of the roto challenges from using the old footage?
GM: This is where it got really squirrely, because it literally was impossible to rotoscope out some of the background action. It's just completely a smear. There was nothing to cut around. We just had to tell the guys who were making the splines for us to contain the colour smears, the action, as best they could. Then we had to do some painting and keying after the fact. We got everything composited to a certain level, then we had to do some edge painting to fix the edges. There was no other way to do it, apart from a little bit of elbow grease with old-fashioned brush work and trying to key some details back in whenever we could. We could have spent weeks doing that for these spots. We could have just gone on and on and on, because it was really tough. But we had a deadline, so when that arrived it was, 'Stop painting, boys! Put your pens down'.
fxg: : I think I've heard someone say that you never really finish visual effects, you just have to part with them.
GM: That was exactly the case here. None of these spots are ones where you'd say, 'That is absolutely finished'. We could have gone on working on these for weeks.
fxg: Well, they still look great, I think.
GM: I think they look really nice. They get the idea over. They play out well when you see them. But all of us are like 'Dammit! If we could only spend a little more time...'.
fxg: So you've done the roto and you've got a CG stadium, what was the next step in terms of executing the final shots?
GM: The CG needed to be integrated to look more like the archival footage that we had. We had a nice pristine CG stadium that we had to muck up a bit. We had to add blurs and grain and defocuses, lens flares, distortions, interference - anything that needed to be added to blend it with the original footage.
GM: The players were all separated out onto different layers. We had separate matte channels for them. We had worked out all the timings at the previs stage, so we knew what the timings would be. So we were just following the previs at this stage. We tried a few effects for the turning off and on. The client first wanted us to use an effect to bring in the characters. We tried some different things but ended up going with an absolutely simple dissolve. It was the best thing. Any effect that we did seemed to ruin the whole bloody spot. It took your eye to the effect and took your brain somewhere else. The dissolve was the best - old-school, dead simple. You're unfazed by it. Everyone just has the same dissolve and you just enjoy the spot for what it was, so that's what we went for.
fxg: One of my favourite parts, in the Larry Bird spot, is where the ball's going high in the air and you see the lights in the roof and it's blurry with video artifacts and you're thinking, 'I know this is a bit CG, but it still looks old-school'. How did you decide which artifacts to keep in from the original footage?
GM: You know what, there were a lot more artifacts that we were originally going to keep in. The client made us lose some of them, so it's actually cleaner than the original footage. There were so many artifacts that they thought it was spoiling it a little. They let us leave some of them, but a lot we took out. There were sometimes these big streaks across the screen, big pops and bangs. It was nuts. It looked like it had been to hell and back!
fxg: What tools did you use to do the final compositing?
GM: The final compositing was done in Autodesk's Flame. We used a lot of different software packages, but it was mostly TinderTools, Spark plug-ins for Flame and Sapphire plug-ins too. We used Synth Eyes for tracking and Silhouette for rotoscoping.
fxg: Did you have any issues slowing down the footage and making sure it appeared quite smooth?
GM: Yes, we did. Again, that's something else that we could have spent forever cleaning up the slow-downs. You can still see some slow-down artifacts, if you look. To do the slow-downs, the Flame has a nice motion-based time warper now. We used a combination of that and Furnace Kronos to slow the clips down. There were quite a lot of artifacts and for some of the pieces we spent a considerable amount of time trying to repair them. At a point, we decided to move on and leave some of the artifacts in there.
fxg: But the point still gets across.
GM: These spots are not about technical perfection. That was pretty much impossible with what we were working with. We felt as long as we weren't spoiling the point of the spot, the emotion of the spot, then anything goes.
We've been a free service since 1999 and now rely on the generous contributions of readers like you. If you'd like to help support our work, please join the hundreds of others and become an fxinsider member.