|
I have dabbled with cylindrical panoramas on and off since 2002, using a Horizon 202 swing lens film camera. But when I aquired the Canon EOS 10D, I stopped using film completely, including the Horizon. I still have it, but it has not been used.
Panoramas still held a fascination, and I tested various programs for stitching multiple images into a complete panorama, but none of them had the ease of use that satisfied me, and the photographic hardware I had was not really suited for the task.
As time went on, the 10D was exchanged for a 1D MkII, and I got a proper tripod ball head. And then some questions from a good friend triggered more research into creating panoramas by stitching, with a heavy bias towards DIY panoramic hardware.
There were several pages on the net describing how to use a ball head upside down, in order to use the ball for easy leveling, and the panorama base for turning the camera. A macro rail made it easy to position the entry pupil on the tripod axis, and a simple L-bracket allowed for vertical pictures. It worked well enough, but it only allowed single row panoramas to be created. But by using a macro rail, I could use any lens I wanted, and still be able to position the entry pupil correctly. But then I would have to remember to reposition the macro rail if I changed lens.
 |
For full sphere panoramas, only the widest lens I had was appropriate. The next step was making a rig that allowed the camera not only to be turned horizontally, but also vertically around the entry pupil. Wood is cheap, and easy to work with, but now the camera became a problem: A 1DMkII is a very heavy camera, and needs a very strong support in order for the movable arm to stay put without slipping. The first attempt failed miserably, so for the second rig much sturdier materials were used, including an oversized knob for tightening the movable arm.
One benefit of this approach, is that the correct distances for a particular combination of camera and lens must be used, which means that there are no adjustments to be made or forgotten.
The first experiments with this rig showed that the inverted ball head had some drawbacks. For one, the knobs controlling the head tend to sit at the bottom of the head (which is now the top), blocking out more of the downward view than necessary. The scale on the panorama plate faces downwards, and even if additional marks were placed on the panorama base, there was still the need to peek under the rig for some of the shots to see the index mark on the head and align it with the marks added to the panorama base. This may be a 'shortcoming' of my ball head, the control knobs are placed next to the panorama base. Other heads with the knobs placed differently may allow adding an index strip around the head, easing the correct positioning of the camera. And finally, converting the ball head between panorama use and ordinary use was cumbersome and something one thought about twice before performing.
 |
So I finally caved in and bought a Manfrotto 300N panorama base. The ball head can now be used the correct way up, and the panorama base has very precise click stops (from 4 to 72 stops in one rotation). This precision, along with acurate marks on the rig for horizontal, up and down images, allows easy creation of high quality panoramas. And as an added bonus, the knobs of the ball head are now far enough away not to block extra the downward view. See the now obsoleted Large DIY QTVR Panorama Head.
As mentioned, the 1DMkII with a 17-40/4.0 is a very heavy camera, way too heavy for panorama use (it is very difficult to hold it properly for a nadir shot), and the number of pictures required (3 rows of 8) made timing difficult if the day was cloudy with the sun coming and going behind clouds. A wider lens was needed. I contemplated the EF 15/2.8 fisheye, but that would still require quite many shots, and the Sigma 8/4.0 was not reputed to be very sharp at the perifery. But Sigma was coming out with a new fisheye, the Sigma 8mm F3.5 EX DG. When it became available locally, I had read preliminary tests that showed that this lens was a lot better than the one it replaced, so this lens was purchased.
 |
Alas, the combo was still too heavy for comfort. The Canon EOS 400D had been introduced, and at the first price drop I got one for panorama use (and for travel and backup, of course). Now a smaller DIY QTVR Panorama Head had to be made.
After about 1 and a half years with the 400D, it was given away to my brother in law, and was replaced by a 40D. The 40D has live view, and I used this to verify the accuracy of the focusing scale on the Sigma. It turned out to be pretty bad, and this is how the correct setting of the focus barrel can be determined.
The software I ended up using, is PTGui for stitching. Version 5, which was introduced September 2005, can now generate control points on its own, something I find very useful. The exception being a partly cloudy day with trees blowing in the wind - in this case large numbers of control points are generated on clouds and leaves that have moved between the shots.
PTGui creates equirectangular images, which need to be converted to cube faces for use in Quicktime movies. I use Pano2QTVR to generate the cube faces and the final movie. If you use Pano2QTVR, make sure that you also have Panorama Tools installed, the built in converter is not very good.
I use Picture Window Pro for editing the panoramas, including removing the tripod at nadir. See the tutorial on how to use Picture Window Pro to edit panoramas.
The panoramas from this year's May 17 celebration were edited with Picture Window Pro.
Return to Øyvind and GI Tryti's home page.