Solving Camera Motion
In the Tools Region, jump to the Solve tab where you’ll find options to solve the camera motion that will ultimately be in your 3D scene. For example, you’ll see the Tripod option; if you filmed from a tripod, your footage won’t have much perspective information, so Blender will only calculate the camera rotation if you enable this option.
The keyframes selection is also important. Blender needs to select two frames of the footage that will serve as a base for calculating the perspective in the rest of the frames. Those two frames should be frames that include fairly different perspectives but have a significant number of markers in common between them. This way, Blender will compare the perspective shift of those markers between the two frames and use that information as a guide. You can activate the Keyframe option and Blender will select those two frames for you, or you can input them yourself as the keyframes A and B.
The Refine option is useful when you don’t have the information about the camera’s focal length, for example, or its distortion values (the K1, K2, and K3 parameters). So if you enable one of those options for Refine, Blender will estimate them for you.
Once you’ve made the appropriate selections, click the Solve Camera Motion button and look at the header. It will now display the error margin to the right after all of the buttons with the Solve error field (see Figure 13.3). Blender detects the difference between the 3D camera and the real camera’s perspective information it determined from the markers. A tracking Solve error of 0 is perfect tracking, but that never happens and there is always a small amount of error. Usually, the tracking will be acceptable if it has a value of less than 3, but the camera can have a slide effect at times (when placing the 3D objects onto the real footage), and under 1 is usually pretty good, with less than 0.4 or 0.3 considered to be a very good tracking.
If you go to the 3D scene, there is probably nothing going on and this is because there is still one thing you need to do to make this scene work. Here are the steps to follow:
1. Select the Camera and, on the Constraints tab of the Properties Editor, add a Camera Solver constraint.
2. Enable the Active Clip or select the clip’s name from the list. Now you should see the camera and a set of little points in the scene. Each one of those points represents a marker in the Movie Clip Editor.
3. Scrub through the Timeline now and you’ll see the camera moving (see Figure 13.4).
Figure 13.4 There is camera motion, but you still need to align camera so the 3D scene is in place over the real footage.