I took a quick video on my iPhone 4, and tracked it in Blender. It appears to have tracked quite well. Note that I have some tracks in the foreground and some in the background:

Example of tracks in movie clip viewer

However, after solving, Blender seems to think that all the tracks are basically planar, at a roughly fixed distance from the camera:

enter image description here

I used the tripod solve mode, because the camera isn't moving (just panning/rolling) in the shot, as suggested here. Is this correct?

I did select the "iPhone 4" preset, but I didn't change any of the other settings.

What can I do to get the scene to be tracked correctly?

  • $\begingroup$ As for the orientation. Try using the orientation tools to determine floor and origin. $\endgroup$ – user1853 Nov 23 '14 at 3:11
  • $\begingroup$ @cegaton Hmm…I hadn't noticed that, but it looks like you're right. Do you know any workarounds? (I tried without tripod solve and it's really jumpy.) $\endgroup$ – wchargin Nov 23 '14 at 3:11
  • $\begingroup$ @cegaton I tried setting Origin and X–Y axes, but it didn't turn out quite right (at an odd angle, and I'm going to be doing simulations, so that won't work). I also tried to use Floor, but the floor was placed perpendicular to how it should be, because the tracks aren't correctly solved. $\endgroup$ – wchargin Nov 23 '14 at 3:12
  • $\begingroup$ @cegaton Thanks very much! It didn't occur to me that you could reposition the camera even though it has the Camera Solver constraint. So the key, then, is that the Camera Solver constraint is relative, not absolute? (If you post an answer I'll accept it.) $\endgroup$ – wchargin Nov 23 '14 at 4:02

If the camera is not moving (you're not walking or otherwise displacing the camera to have some parallax) the only way to solve the shot is using tripod solve. Normal solve will only give you errors.

It's very hard however to accurately get depth on a tripod solve.

From the wiki:

tripod can be used for footage where the camera does not move and only rotates. Such footage can't be tracked with a generic solver approach, and it's impossible to determine the actual feature points in space due to a lack of information. So this solver will solve only the relative camera rotation and then reproject the feature points into a sphere, with the same distance between feature and camera for all feature points.

Once the scene is solved you might not get the right orientation using the default tools (like floor, wall or origin) but you can set manually by selecting the camera in the 3D viewport and rotating it to make the plane and cube match the orientation of the real images (remember to set the original footage as background image). All the trackers are parented to the camera, so they will follow it as you move it along.

What you want to get is something like this:

sample file here enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.