Since moving to our new office, the video department has much bigger and better studio space. Currently, we have 5 powerful PCs churning away and the setup is much more conducive for coordinating and maintaining a render farm. Using After Effects CS3 as our main video compositing software, we are able to make use of the Render Engine installed on the 5 PCs to make the rendering process much faster and streamlined. This allows us to be faster and more productive, spending more time on the creative aspects of each video editing and animation project.

Tomorrow we have a fairly involved video shoot which I talked about in a previous blog. We have an interesting shoot setup which should help in making the video flow smoothly when a user interacts with the website in which it is incorporated. Using Adobe Ultra keying software, we will be able to display a composite video of the live actress with a still shot of that same actress in a specific pose that will remain a consistent starting position for all the video clips. We are able to use the teleprompter screen to display this composite video which prevents clutter in the video studio. Directing and shooting a spokesperson video that relies heavily on website interaction is no easy task and the saying “you’re only as good as your talent” definitely comes to mind. Luckily, our network of actors are all professionals with extensive experience, so we know that performance is not an issue.

Match Moving is my word of the week, maybe even of the month and something tells me that I might not be the only one writing about it this week. The title of today’s blog is part of a larger mathematical calculation of match moving calibration which “attempts to derive the motion of the camera by solving the inverse-projection of the 2D paths for the position of the camera”. You say “What?”, I say, “Awesome.”.

Basically, any movie that you have seen which incorporates 3D animation and characters has used the process of match moving. It sounds weird but it is a very similar process as motion tracking, only that it doesn’t track on a 2D plane (X and Y) but a 3D plane (X, Y and Z) and accurately calculates the actual movement in space of the camera which shot the video footage.

The first step in match moving a piece of video is Tracking. Tracking involves an algorithm that follows a specific point or set of pixels in an image through a set of frames. Usually, a point is chosen because it represents an object in the image, something that has high contrast and is easily identified in the moving image.

The second step is called Calibration. This is where complex mathematics come into play and the hair on the top of my head shivers as the gust of “things I can’t yet comprehend” zooms by. To compensate from any errors in the equation that may create unnatural movement of the camera, optimization algorithms and bundle block adjustment are used (well duh!).

Mentioned in a previous blog, the DDA CMT project is going to make full use of match moving. The video will be strictly choreographed so that we can ensure the 3D elements (3d characters and 3d sets) are visible and the composition is correct when match moving is applied and the video composited. It really is going to be beautiful!