Experiments seem to always generate more questions than answers and then more experiments.
The purpose of yesterday’s tests was to see where the Least Parallax Point occurs at varying focal lengths of this zoom fisheye lens. Probably the clearest result from these tests is that PTGui, and in particular, PTGui Pro, Version 9.1beta1 (released August 2, 2011) allows an amazing degree of latitude afforded through its improved automatic control point generator for fisheye images. This is illustrated in the two 8mm examples where the upper rail settings on the NN5 varied by nearly 5mm, yet both stitched with very good results – freaky!
A couple of things to keep in mind: these tests were conducted in part with the idea of studying the effectiveness of pano head tuning using the method outlined in the article Optimal Stitching Point and accordingly, was primarily focused at the nadir and as you’ll note in three of the the assembled panos above, that there are holes along the horizon. The reason: it didn’t really matter for these tests if there was sufficient overlap/ coverage along the horizon or not. For determining the proper coverage/ overlap of images, be sure to check out the very nice Pano Calculator Christian Bloch and Frank van der Pol have put together. The other thing, these tests used the unedited RAW CR2 files straight from the camera with, again as I’m sure you’ll note in these examples, all of the chromatic aberrations – something typically and easily removed prior to loading into PTGui through Adobe Camera Raw.
Interesting Observations Along the Way
Getting the lower rail tuned first is a must and easily accomplished using the camera inverted and simply visually aligning it with the center of rotation. I’ve included a little snippet of the camera plate in each of the six exhibits as a reminder that the lower rail settings used here reflect its use.
I also played around using other techniques utilizing the pano head in different positions normally seen during its routine operation; namely, when the camera is looking at the horizon and looking at the nadir. This proved more of a distraction away from the lens and more attention on the Nodal Ninja 5′s construction combined with perhaps other issues intrinsic with the Canon 40D – who knows? None the less, I’ve included those observations.
As a variation on John Houghton’s plumb line method outlined on his website, I carefully constructed a template that fit on the lens and then ensured its axial alignment with a drafting triangle. Flatbed scanners like the Epson Perfection 1640SU are great cameras for beautifully capturing detail at high resolutions and what I used to first scan the lens cap and then draw x and y axis through its center in Photoshop. Careful measurements allowed for the correct diameter of the hole to be cut out. The final image was then printed and spray mounted onto a piece of mat board.
Another idea was to employ the Live View feature on the 40D while the camera was in its normal orientation for shooting the nadir. This proved to be dubious at best for reasons unknown. How much of the unexpected offset there is from the Live View’s center and the center of rotation because of the lens, camera and /or the Nodal Ninja is the subject for later experiments… maybe, but probably not mine.
Again, while having the camera aimed at the nadir, and in an unlit space, I used an LED type flashlight to project the little square seen in the camera’s view finder onto the surface of a piece of paper using the lens to focus the projected image. Why the little projected square is offset differently from the lower rotator screw’s center as seen using the Live View method mentioned above is another puzzle for another time. Both methods were originally thought of as independent means in finding the exact setting for the lower rail but fail to even come close.
Getting the upper rail tuned was initially, an astonishingly simple accident. With the camera aimed at the horizon and still wearing the plumb line card mentioned earlier, I first noticed that the card failed to rest perfectly on the lower rail even though I had very carefully measured the distance from the center of the upper rotator screw. So, in studying that further, I began sliding the camera along the upper rail until the red line was coincident with the upper rotator screw. Then in studying that relationship, I noticed that the screw itself lacked symmetry… another puzzle for another day! But happily, that initial location of the camera on the upper rail proved perfect for the 8mm focal length
Much appreciation goes out to Michel Thoby for his pioneering and inspiring work with fisheye lenses and their use in panoramic photography and to all those who frequently and kindly provide support on the PTGui Google group list.