Flying in a CG World
By Tom Bremer
Issue: Volume 37 Issue 2: (Mar/Apr 2014)

Flying in a CG World

Planes are cool. They look cool, they do cool things. They can go super fast. Some have guns. It's every boy's dream to be a fighter-pilot at some point. So, if it's such a fascination to so many people, why can't we get the motions correct in CG re-creations for film and television?

I was once one of those people who thought that making a CG airplane fly around in a piece of software would be a cakewalk. "Just put a keyframe here, a keyframe there, and bank the plane. How hard could it be?" Well, let me tell you, it's a lot harder than it looks.

I started in this business about 10 years ago; my first job was animating characters at a small company in Minnesota called Wet Cement. But when my friend in Los Angeles called me about a job involving animating airplanes, I jumped at the chance. I moved to LA to work at a studio called Radical 3D on the TV show Dogfights: Greatest Air Battles for The History Channel, and became the lead animator on the show for the two years it was on the air. After that, that same friend called me to Zoic Studios to work on a show called Jericho, which had a flight sequence the studio needed help with. Shortly after, I had learned that ILM and Lucas were working on Red Tails, and they came to Radical 3D for the flight animations in the film thanks to the accuracy of the Dogfights series. Later, I was hired by Disney Toons Studios to help out on its animated feature Planes, because they, too, were unhappy with their initial digital flying. I ended up going back to Radical 3D, this time as VFX supervisor to complete a show called "Why Planes Crash" on MSNBC as well as a film called Red Sky (directed by Mario Van Peebles).

Weighty Matters
If ILM and Disney were having trouble with digital flying, there must be something special about it. So, what is it?

Well, there's a long and a short answer. I will start with the long version. The main issue that most films and TV shows have when it comes to virtual flying is weight. Weight, weight, weight. People see these machines flying at Mach 2 doing barrel rolls and loop de loops, and they see how effortlessly they do it. The problem is we forget how much power is actually behind these accomplishments.

For example, an F/A-18 Super Hornet can go from zero to 60 mph in .8 seconds, but it weighs over 32,000 pounds. So, obviously it's not because its super lightweight, but it actually has two GE F414 jet engines kicking out 22,000 of pounds of force each. That's a lot of kick, so it's easy to see why one would think these things move quickly. The problem comes when we have fast-moving airplanes over short distances. These things have a lot of mass and a lot of speed, and any maneuver is going to require a certain amount of lag. So, when they do these maneuvers in a short distance, the planes end up looking very lightweight, too fast, and very unrealistic.

On every plane project I have worked on, we always made the planes fly at real speed and at real distances. This meant that at least the scope of the maneuver was reined in to at least a realistic starting point.

AERIAL ANIMATIONS require skill and knowledge for realistic results.

But, I may be getting ahead of myself. Before we even started animation, we required the plane to have a control rig. We usually had a three-locator hierarchy, the first for X,Y, Z translation, the second for heading and pitch rotation, and the third for bank rotation. Under that, the plane was parented to the bank locator. This simple rig gave us so much control over the motion of our aircraft that we never had to fight competing curves again.

Having this simple setup is great and all, but it's only part of the rig. The locators need to be placed where the plane would realistically pivot in flight, which isn't necessarily the center of gravity. It's actually center and back, just before the flaps on the main wings. I have seen a lot of animations where people put the pivot point back on the elevators near the tail, and that throws off the entire look. Too far back and it will look like your plane is doing a wheelie, and too far forward will look like it's doing a nose slide.

Another problem I see in nearly every film and television aerial sequence is another simple yet overlooked aspect of flight. When people animate these planes, they tend to plot the basic motion curve and then orient the nose to point right down the curve, as if the plane were a car on a roller coaster locked onto a track. This is fine for certain light maneuvers - maybe an easy bank to the left or a gradual climb. This rule doesn't hold up, though, when the plane is moving exceedingly fast and makes abrupt maneuvers, like in a dogfight.

Due to the mass of these huge machines, they need time to adjust their trajectory and, therefore, shouldn't be able to turn on a dime. The point of my story is that when going 1,000 mph and needing to abruptly climb or bank left or right, the plane's mass will keep it moving forward despite the fact that the elevators or ailerons have pointed the nose elsewhere. If this is unclear, think of it as a car drifting around a corner: The car has all this forward momentum and then it's suddenly changed to the right, but the car's mass is still moving forward until it is overcome by the force and grip of the tires on the road. It's the same in the air. These things turn and actually slide in the air until the forward momentum is overcome by the thrust of the engines and lift of the wings. So, the act of just prematurely rotating the heading or pitch in a turn or dive in order to make it slide adds a sense of weight that you ordinarily wouldn't have. But that said, the curve has to have that lag built into it to be believable.

Another major issue plaguing this realm of animation is when artists have planes flying perfectly straight and level. That just doesn't happen. There is always some sort of air current or pocket of less-dense air that we fly through - a concept we know as turbulence. This kind of stuff causes shifting in altitude, speed, and rotation. Knowing that there are air currents and such, there should always be some element of random motion in either the rotation (heading, pitch, or bank) or translation. That said, it is super easy to go overboard to where the aircraft starts looking as if it were light as a feather, rather than a 30,000-pound hunk of metal.

Roto Moves
Helicopters follow pretty much the same rules as planes do, though there are some differences. The pivot point should be at the center of the main rotor, as everything else hangs underneath it. The cyclic of a helicopter, which is the main control stick, manages the pitch and bank of the main rotor; this, in turn, is what causes the helicopter to move in different directions while still facing forward. Once that rotor moves, everything follows - it's that simple. If the pivot was anywhere else, it wouldn't look correct.

It also needs to be centered to the main rotors because the engine and rotors add a certain amount of torque to the entire craft. If there were no tail rotor, the main rotors would rotate one way and the body of the helicopter would spin the other way. As a pilot flies a helicopter around, he uses the throttle to speed up the rotors, but as he does that, the engine exerts more torque on the craft. The pilot constantly has to fight that torque by using the rudder pedals, which control the tail rotor blade pitch. This creates more or less lift, but only laterally (since the tail rotors are mounted vertically), and controls the heading of the craft. If anything needs heading fluctuation as it is flying around in CG, it's a helicopter. But again, these changes take time to move the mass of the aircraft and counteract the other momentum forces.

Honestly, I see some things on the screen and think, "Who OK'd that? Have they never seen a plane fly before?" I don't know, but maybe I'm oversensitive since I've been animating airplanes off and on for nine years. I could go on and on about this, but I will cut to the short answer: The animation just needs to look good.

I go by a little mantra when it comes to CG and VFX, and that is there is a difference between being real and looking real. What I mean by that is, yes, maybe during a mission the military will refrain from using its navigation lights on the jet, but by adding it to a final comp, it increases the realism, since that's what people see all the time when they look at planes flying at night.

What I want you to take away from this piece is that flying is hard, and you need to look at references, but almost more importantly, just know how a plane flies. Without that understanding, you most likely will get subpar results. But, in the end, the ultimate goal should be to make it look good - whether it's accurate or not.

Tom Bremer has more than nine years of experience as a visual effects artist, working on movies such as The Hunger Games, The Amazing Spider-man, and Disney’s Planes. In 2010, he received a Primetime Emmy Award for Outstanding Visual Effects in a Series for his work on CSI: Crime Scene Investigation. He is currently an instructor at The Digital Animation & Visual Effects (DAVE) School in Orlando, Florida.