Avatar as we all know is a milestone in filmmaking which in its journey created, introduced and defined many technical and creative firsts. The phenomenal amount of work that has gone in making the film will be discussed and re discussed for many years to come.
At a recent Autodesk press briefing, Nolan Murtha, Digital Effects Supervisor, Lightstorm Entertainment and Shawn Dunn, Head of Layout and Animation Technologies, Weta Digital got talking about Avatar and shared some valuable insights into the behind the scenes of the film.
Virtual Camera
For shooting Avatar a virtual camera was build which enables one to see the simulated CG world when focused on a real environment. "Virtual Camera is really like taking a monitoroff of the computer. We use our best motion building software which has acquired 200 million methods. So we would send our video feeds from motion building operator‘s box up to the monitor which your camera is holding. And the virtual camera has several controls; you can zoom in and out and he (James Cameron) could offset himself in the space. He could shoot with the actors without having to be right next to them. He could move himself in the 3D space, he could move up and down. We could scale his movements up to 50-1. So for his every foot, he‘s actually moved 50 feet in the virtual world. So he could fly along with one of the banshees or the gunship just by taking a few steps" Nolan said.
For shooting Avatar a virtual camera was build which enables one to see the simulated CG world when focused on a real environment. "Virtual Camera is really like taking a monitoroff of the computer. We use our best motion building software which has acquired 200 million methods. So we would send our video feeds from motion building operator‘s box up to the monitor which your camera is holding. And the virtual camera has several controls; you can zoom in and out and he (James Cameron) could offset himself in the space. He could shoot with the actors without having to be right next to them. He could move himself in the 3D space, he could move up and down. We could scale his movements up to 50-1. So for his every foot, he‘s actually moved 50 feet in the virtual world. So he could fly along with one of the banshees or the gunship just by taking a few steps" Nolan said.
‘With the Virtual Camera we could scale his (James Cameron) movements up to 50-1. So for his every foot, he’s actually moved 50 feet in the virtual world’- Nolan Murtha.
Advertisement |
 |
"We have a steady cam mode so we would simulate the steady cam mode and handheld the camera. We had a crane mode so you can go up from a tree and we can drop you down right where you are needed to land with someone around. You can do a lot more things which you can‘t do with a live action camera, obviously, since we have full control in the CG world. So when we are in full CG lens it‘s much easier and then we would have to tie it up with the live action camera, the 3D camera" Nolan explained.
When asked about the drawbacks of this technology and custom made equipment Nolan says, "Actually in some cases it worked really well and in some cases it just falls apart like when you are either shooting outside in the sunlight, there are various problems that come up with the live action stuff"
Technicalities
Shooting Avatar was not at all an easy task as there were two cameras, one recording the CG world and the other recording the real environment. Elaborating it Nolan says, "We had to combine it in real time. So with the CG version of it you use the monitor. The space is huge so it‘s the same as taking your monitor and plugging in your computer, the only difference was we did it with a 50 foot cable so he (James Cameron) can walk around with it. But with the live action camera, the fusion camera, we actually have to put motion capture markers on the big camera and match the position of the lenses. And then we would put live action cameras all around the place at the set and we would recreate all of the action sets as the digital copy so we have the exact copy aligned. And if someone walks behind the table so we had digital tables so their legs get cut off. We did lots of maths like that."
Shooting Avatar was not at all an easy task as there were two cameras, one recording the CG world and the other recording the real environment. Elaborating it Nolan says, "We had to combine it in real time. So with the CG version of it you use the monitor. The space is huge so it‘s the same as taking your monitor and plugging in your computer, the only difference was we did it with a 50 foot cable so he (James Cameron) can walk around with it. But with the live action camera, the fusion camera, we actually have to put motion capture markers on the big camera and match the position of the lenses. And then we would put live action cameras all around the place at the set and we would recreate all of the action sets as the digital copy so we have the exact copy aligned. And if someone walks behind the table so we had digital tables so their legs get cut off. We did lots of maths like that."
‘we would put live action cameras all around the place at the set and we would recreate all of the action sets as the digital copy‘- Nolan Murtha
Facial Expressions
Facial expressions and motions were one of the crucial things which were done brilliantly in Avatar. Shawn gives the credit to the facial system team working for Avatar. "We did some official mocap for King Kong and it worked well depending on the scene. But for avatar it worked amazingly well and one of the reasons for its success is that we have a very talented facial team who built a really great facial system that will accurately portray the actors facial. The important point in facial animation is that the character should look like the actor and it was really tricky. To get things like lips moving when the person is talking so that the lips not just open and close. And making the eyes move around giving the eyes movement. It‘s a long process making sure that you get the details right".
Facial expressions and motions were one of the crucial things which were done brilliantly in Avatar. Shawn gives the credit to the facial system team working for Avatar. "We did some official mocap for King Kong and it worked well depending on the scene. But for avatar it worked amazingly well and one of the reasons for its success is that we have a very talented facial team who built a really great facial system that will accurately portray the actors facial. The important point in facial animation is that the character should look like the actor and it was really tricky. To get things like lips moving when the person is talking so that the lips not just open and close. And making the eyes move around giving the eyes movement. It‘s a long process making sure that you get the details right".
‘We merged the boom mike and a camera and placed it in front of the actors face. So it records their voice as well as their facial movements’- Shawn Dunn
According to Shawn, they merged the boom mike and a camera and placed it in front of the actors face. So it records their voice as well as their facial movements. "So the capture part became very easy you just put the markers on face and record it. Later is the difficult part when we have to align the markers and make it 3D. We then took it to the animators and the animators would work on finessing the motions making sure that the emotions of the actors and their facial expressions were presented clearly in the CG character".
Creatures and their motions
To record the motion of CG creatures Avatar team found a unique way. To record the motion path they made small gunships and banshee puppets and moved them around tiny mountain puppets. And then they would put the motion of a human flapping his arms on the wings. "And so what we would do is we record the flying path and then we would put the monitor in front of our guys face and he would look into the monitor and if the banshee is flying up then he would flap his arms pretty hard and he would bring his chest out and then he would duck and all. And so in taking the animated scene and giving it to Jim we would only take 45secs or a min to capture the creature and it worked. We did that for the fibre wolf as well. We had an acrobat who would run around on the floor. So we would improvise few things but it was mostly just to get the speed, the tempo and the time and in the meantime Weta would go and fix that", explained Nolan.
To record the motion of CG creatures Avatar team found a unique way. To record the motion path they made small gunships and banshee puppets and moved them around tiny mountain puppets. And then they would put the motion of a human flapping his arms on the wings. "And so what we would do is we record the flying path and then we would put the monitor in front of our guys face and he would look into the monitor and if the banshee is flying up then he would flap his arms pretty hard and he would bring his chest out and then he would duck and all. And so in taking the animated scene and giving it to Jim we would only take 45secs or a min to capture the creature and it worked. We did that for the fibre wolf as well. We had an acrobat who would run around on the floor. So we would improvise few things but it was mostly just to get the speed, the tempo and the time and in the meantime Weta would go and fix that", explained Nolan.
Ask Nolan why human giving motion to these creatures and he goes, "I mean obviously the human motion of flapping wasn‘t what you saw on the movie. So what our animators would do is look at bird motions and look at how creatures fly and they would acquire those in 2D. But this gave us enough idea to shoot the movie in Motion Builder and work in a quick pace and that became possible on the Motion Builder".
Composing a Scene
Shawn asserts, "It‘s not like moving your camera while recording a scene. What I mean is you have got a scene comprising of 15- 20 shots and Jim would change the environment for each shot. So if he wants a hand to come and touch the home tree he would shoot from different angles and then we would be like no it‘s not working move the home tree little closer. Digitally it‘s very easy because in virtual world you can move things. In the final environment world it‘s painful. In the scene when the gunship is hovering above the forest and looking down the home tree Jim wanted to show the ground, but if you notice the home tree is tilted. So then we created another home tree which was straighter, took the shot and stuck the picture perpendicular to the first home tree so you can see ground from the gunship hovering above the home tree".
Shawn asserts, "It‘s not like moving your camera while recording a scene. What I mean is you have got a scene comprising of 15- 20 shots and Jim would change the environment for each shot. So if he wants a hand to come and touch the home tree he would shoot from different angles and then we would be like no it‘s not working move the home tree little closer. Digitally it‘s very easy because in virtual world you can move things. In the final environment world it‘s painful. In the scene when the gunship is hovering above the forest and looking down the home tree Jim wanted to show the ground, but if you notice the home tree is tilted. So then we created another home tree which was straighter, took the shot and stuck the picture perpendicular to the first home tree so you can see ground from the gunship hovering above the home tree".
To which Nolan adds "Essentially there were three motion building cameras and they all were put together to create a single stereoscopic image. We would open the same file on two computers and render the right eye on one computer and left eye on the other computer. And the third box is where the operator will sit and control the other two. And for every single shot in the movie we literally went frame by frame and evaluated the stereo".
"We used about 106 cameras in the space of 25 m width by 12m breath and the height of 5 m. The motion solving was done by Giant Studious, motion capture providers. There software they used is called Real Time which they used to solve the skeletons. We would feed those skeletons over to Motion Builder and drive the puppets of real time and then that would go to Jim‘s monitor", Nolan said.
Procedure followed for most Scenes:
- Construct the environment
- Test it with dummy actors for positioning
- Shoot the scene with real actors
- Editorial process (maybe out of 10 takes they like 3 and delete the remaining 7)
- Custom-tailor the remaining scenes
- Feed the motion skeleton from Giant Studious to the Motion Builder
- Rendering
- Selecting the final scene
About outsourcing the work Nolan says, "We would work in morning with Paris and London because they are ahead of us. We worked with Montreal and Industrial Light Magic, San Francisco and in evenings we would work with Weta! We worked with 6 to 7 different countries on 3 continents. We would like to keep things in one roof like Weta but if the movie gets too big then we do outsource".
For Shawn working for Avatar was challenging and difficult and ‘religious‘. He says, "It‘s just that Avatar was taking a complete new level with much bigger environments, many more characters, and more animation". Nolan who is currently working with James Cameron on Avatar 2 admits, "Avatar was a risky movie. You don‘t know if people will like 10 foot tall blue aliens kissing in a weird glow in the dark jungle. It could be a disaster because it‘s not necessarily a big blockbuster movie formula".
Avatar‘s creative journey was not smooth and hurdles were tackled intuitively, technically and also by tricks and some cheating- technical one‘s. But the level of VFX filmmaking brilliance that Avatar brought to the viewers is yet to be surpassed or even matched.
No comments:
Post a Comment