The challenge of the one-take
Early on in the production Rickard, our director, decided that it could be an interesting challenge to create the movie using a one-take camera. This means there will be no visible edits during the entirety of the 7 minute running time.
Moves and pauses
Creating a one-take camera in a movie is no easy feat both in terms of the complexity of the production and in telling the best story with the way the camera moves and points.
A movie production is much easier to deal with, when you can split the film up into several smaller, more manageable parts. So we decided that we would of course create edits to make the production more manageable for our selves. The edits would just be invisible for the viewer.
Without visible edits the camera would need to move a lot around the environment to capture and keep up with the story and the characters' movement. So we created a rule set, to make sure the camera always has a reason to move (i.e. following a character's movement across a room).
It was also important to create contrasting pauses, where the camera would remain still and give pause in the otherwise high tempo of the storytelling.
The SpaceNavigator
As anyone who has used a normal mouse in Maya will know, it is near impossible to make smooth real time camera moves using only the mouse. Of course one can always create movements manually. You just place the camera, set an animation key and program how the camera should move.
It would be fantastic to somehow create a system, which could let the use fly the camera around in an intuitive way, and at the same time record the moves...
In our research we came upon the SpaceNavigator 3D mouse, which is a sort of joystick for 3D programs like Maya and Max. You can push, pull, tilt and drag the controller around the 3D environment. creating smooth and natural movements.
During the creation of our camera ride, we found out that the SpaceNavigator has some problems moving in sharp and small curves. Making real time recordings of the SpaceNavigator's moves also proved a bit choppy. So we ended up using the device as a fine guideline for maneouvering the environment, while classic keyframing of the camera turned out to be the way to go.
Here is the link for Space Navigators homepage:
http://www.3dconnexion.com/3dmouse/spacenavigator.php
Camera placement and blocking
The majority of the camera ride had to be created using key frames. This proved to be a long and extensive process, which would gradually be improved and revised over several months. We had several people involved in this: the animators, our photographer and editor all contributed.
The framing and timing of the camera is essential to what should be included in the production. Animations and props only need to be included if the camera can actually see them. But at the same time our camera is moving a lot around so most of the props and lighting has to be in place all of the time.
To have a better idea of the exact movements we put in previs versions of the characters, environments and props. These would serve to block in the camera, so we knew what would be visible in the framing. Later the animators would replace the early versions with more refined models, animations and mocap data.
Also as we had a one-take camera that should later be cut up into smaller scenes, we needed to have an exact take before we dívided the scenes up.
Creating invisible cuts
After the camera ride had been created we needed to find the right times to split up the scenes, so our animatiors could take on each animation part on their own.
As the cuts should be invisible, this would preferably be where no animations would be active and even better the camera would be still at the same time.
In the end the film was split up into 16 scenes, that start and stop at exact frames one after the other. Among these it was impossible to avoid a few shots, where animations in the background would have to be imported to another scene to create the right transition.
The final lay-out
With the final lay-out of the camera done and the frame count of the different scenes laid out, we created a number of references to the main camera ride and gave each animator a scene, where parts were locked and other parts were free to manipulate.
This way the animators could manipulate the 3D characters and props inside the scene, but not the frame count, the environment or the camera move.
After each scene was animated effects would need to be added as well as a few compositing tricks. Finally all the scenes would hopefully be brought together into a seamless whole.
|
