Tracking From The Beginning

Just recently, Dan Brockett wrote about automation in edit and motion graphics. He talked about scripts, intelligent software and plug-ins. One piece of software mentioned was Mocha Pro by Boris FX.

Mocha Pro is a tracking tool popular among visual effects specialists, motion graphic artists and editors. Boris FX recently released Mocha Pro 2019, so I’ll explain the concept of tracking and how it has changed. Next time I will go through many of the new features in Mocha Pro 2019. Following that, I’ll point out an often-ignored Mocha Pro configuration.

Why track?

In the beginning, the reason you tracked a scene was to overlay an “object” on the scene in order to make it move as if it were there during the original shoot. This “object” might have been type or a blur filter to obscure a name-brand logo or it might replace an image on a TV.

Tracking a shot was also the method you used to stabilize a shot. You might need stabilization to remove a bump in a dolly move or to help a handheld shot when the camera operator had a little too much caffeine.

The method behind tracking was all about picking a small group of pixels in a frame and collecting data about how those pixels moved from frame to frame. Without the quality of capture (and the algorithms) that we have today, it was important that the small group of pixels be unique (and visible) across the full duration of the scene.

If all you needed was to track position (horizontal and vertical movement) you only needed one tracking point. This assumed that the camera didn’t change position—it just panned or tilted—and there was no change in lens focal length by zooming.

When tracking a change in scale (which could also mean depth, since moving closer to the camera means getting bigger), then you would need two tracking points. And if you needed to replace an image on a TV you’d need four tracking points to track the four corners of the screen—which led to the term “corner pinning” that you see in some editing and motion graphics applications.

I mentioned stabilization earlier. A byproduct of tracking is the ability to take the motion data from one or two tracking points, invert it and then reapply the motion data to the footage in order to smooth out any movement. This is an example of how the term “tracking” has grown to mean a lot of different visual effects techniques.

In the early days of tracking, software really needed those pixels to be obvious and unique. If they weren’t, the tracking would fail and start to drift. Then you had to manually adjust the trackers frame to frame. If a tracking point left the screen or was covered up by a foreground object, success became even more difficult. When these things happened, the “automatic” part of tracking disappeared. You ended up tweaking the tracking frame by frame. If not done with care, the hand adjustment added a lot of chatter to the movement.

The difficulty in identifying obvious and unique pixels gave birth to all the weird tracking marker designs you may come across if you search for “tracking marker” on the web. The designs were all about making unique patterns in the scene.

But that was then. I haven’t recommended anyone use a tracking marker for quite a while. While I’m not as demonstrative as Mathew Merkovich is in this Vimeo video, I do agree with what he’s saying.

Tracking has come a long way from those days. And one example of that is Mocha Pro, which doesn’t track points but planes.

Next time, planar tracking and how Mocha Pro 2019 makes things easier.