Capture of stereo panoramas using Ricoh Theta.



Capture of stereoscopic panoramas using 360º cameras.


Sin-título-2360° video cameras use a fisheye optics of more than 180° vision. For this reason we can use them as if they were DSLR cameras with fisheye lenses.

This makes production considerably cheaper. The photographic quality we obtain will depend on the camera used, but its stereocopic quality will be exactly the same as that of professional cameras.




Nodal Alignment


To use these cameras as if they were DSLRs, the first thing to consider is to align the point of no movement of the RIG, with the entrance pupil of the 360º camera.

This value varies depending on the camera used, but generally, we will find the entrance pupil to three millimeters behind the lens of the optics.

If we make this adjustment to make 2D panoramas, we eliminate the parallax error, and using stitching programs such as PTGUI we will obtain a perfect panoramic, even in very small scenarios, such as bathrooms or storage rooms.




Adjusting the parallax


Parallax is the misalignment of the camera with respect to the vertical axis of rotation.

The greater the parallax, the greater the stereocopic effect. But in the same way, the higher the parallax, the greater the sewing error. Therefore, it is very important to adjust this value correctly.

Parallax tables are included with the product.






The catch

rotOnce the parallax is calculated, we will make a capture of 60º, that is to say 6 photos around.

In 2D panography only three or four photos are needed, but in 3D format, knowing that we are shooting outside the nodal point, we must make a greater number of photos, so that this error is distributed among the 6 photos.

And if we want to reduce the parallax error even further, we’ll take more pictures around it. If instead of making 6, we make 12, the sewing error is reduced by half.




How it works

lateralIt’s very easy, but it’s hard to see.

In the lateral GIF we see the capture positions 1 and 4. It is really as if we were using two cameras, because the pairs are configurating as the capture is made.

Logically this technique is only for interiors without movement, if we want to capture scenes in movement will be necessary to use two cameras.




The rest of this paper is for sale. If you are interested contact the publisher.

Continue reading: