Doug Kerr
Well-known member
I am delighted to see the opening of this forum devoted to panoramic photography.
I thought this might me a good opportunity to remind those interested in basic multi-image panoramic photography of a fundamental technical concept, often misunderstood.
In basic multi-image panoramic photography, in order to avoid parallax shift between adjacent images (a cause of anomalies at the "seams"), the camera should be rotated (either in azimuth or altitude) around an axis passing through the entrance pupil of the lens.
There are many tests for empirically determining this location. However, in most cases, we can determine it by visual inspection.
The entrance pupil is, by definition, the virtual image of the aperture stop as it appears from in front of the lens.
If we peer into the front of a lens, and "see" the aperture stop diaphragm, we do not (normally) see it in its actual location along the lens axis nor with its actual size (due to the effects of lens elements in front of it). What we do "see", in fact, is always precisely the entrance pupil of the lens.
Thus if we visually judge the apparent position of the aperture stop iris (in terms of its depth from the front vertex of the lens), that will be the plane though which a panoramic shift axis (vertical or horizontal) should pass.
Of course, we often hear that the proper location for the rotation axis is through "the nodal point" of the lens. Our first clue that this is incorrect should be that there are of course two nodal points in the typical lens - which one is meant? Well, neither.
Some who learn that the entrance pupil is the critical location are curious as to why apparatus they have for testing "for the nodal point" gives a workable answer for the location of the pivot axis. The reason is that this apparatus actually tests for the location of the entrance pupil!
There is a rather different pivot apparatus, used in optical laboratories, that tests for the location of the second nodal point of a lens. It may be the superficial similarity of these two test rigs that is responsible for the misconception about "the" nodal point of the lens being the proper location of the panoramic pivot axis.
In a panoramic camera with a continuously rotating lens (having a vertical slit behind it) and film held stationary in a cylindrical frame, a different consideration comes into play. Avoidance of parallax shift indeed still calls for the lens to be rotated about an axis passing through the entrance pupil. But avoidance of shift of the image across the film (a cause of "smearing") requires that the lens rotate about an axis passing through the second nodal point. And that is the compelling consideration.
Parallax shift is not avoided by that choice of pivot axis. But such a camera, by virtue of the "slit", has an extremely narrow horizontal field of view. Thus any point on the "object" is essentially only recorded while the camera is at a specific point in its rotation. Accordingly, there is no opportunity for parallax shift to cause "doubling" or "smearing" of the image, and is in fact of almost no consequence.
In our third mechanical modality, a continuously-rotating camera with continuously-moving film, it is again the entrance pupil of the lens where the pivot axis should be located.
A further discussion of this topic will be found in my technical article, "The Proper Pivot Point for Panoramic Photography", available here:
http://doug.kerr.home.att.net/pumpkin/index.htm#PanoramicPivotPoint
I thought this might me a good opportunity to remind those interested in basic multi-image panoramic photography of a fundamental technical concept, often misunderstood.
In basic multi-image panoramic photography, in order to avoid parallax shift between adjacent images (a cause of anomalies at the "seams"), the camera should be rotated (either in azimuth or altitude) around an axis passing through the entrance pupil of the lens.
There are many tests for empirically determining this location. However, in most cases, we can determine it by visual inspection.
The entrance pupil is, by definition, the virtual image of the aperture stop as it appears from in front of the lens.
If we peer into the front of a lens, and "see" the aperture stop diaphragm, we do not (normally) see it in its actual location along the lens axis nor with its actual size (due to the effects of lens elements in front of it). What we do "see", in fact, is always precisely the entrance pupil of the lens.
Thus if we visually judge the apparent position of the aperture stop iris (in terms of its depth from the front vertex of the lens), that will be the plane though which a panoramic shift axis (vertical or horizontal) should pass.
Of course, we often hear that the proper location for the rotation axis is through "the nodal point" of the lens. Our first clue that this is incorrect should be that there are of course two nodal points in the typical lens - which one is meant? Well, neither.
Some who learn that the entrance pupil is the critical location are curious as to why apparatus they have for testing "for the nodal point" gives a workable answer for the location of the pivot axis. The reason is that this apparatus actually tests for the location of the entrance pupil!
There is a rather different pivot apparatus, used in optical laboratories, that tests for the location of the second nodal point of a lens. It may be the superficial similarity of these two test rigs that is responsible for the misconception about "the" nodal point of the lens being the proper location of the panoramic pivot axis.
In a panoramic camera with a continuously rotating lens (having a vertical slit behind it) and film held stationary in a cylindrical frame, a different consideration comes into play. Avoidance of parallax shift indeed still calls for the lens to be rotated about an axis passing through the entrance pupil. But avoidance of shift of the image across the film (a cause of "smearing") requires that the lens rotate about an axis passing through the second nodal point. And that is the compelling consideration.
Parallax shift is not avoided by that choice of pivot axis. But such a camera, by virtue of the "slit", has an extremely narrow horizontal field of view. Thus any point on the "object" is essentially only recorded while the camera is at a specific point in its rotation. Accordingly, there is no opportunity for parallax shift to cause "doubling" or "smearing" of the image, and is in fact of almost no consequence.
In our third mechanical modality, a continuously-rotating camera with continuously-moving film, it is again the entrance pupil of the lens where the pivot axis should be located.
A further discussion of this topic will be found in my technical article, "The Proper Pivot Point for Panoramic Photography", available here:
http://doug.kerr.home.att.net/pumpkin/index.htm#PanoramicPivotPoint
