|History is important because we learn from our
past. Many of the methods I have developed are partly based on my
early experiences stitching uncorrected images together. Solving the
problems to make something work when it's not supposed to provides
invaluable insights into solving problems when something doesn't work when
it is supposed to. This 'brief' history of my panoramas will hopefully
provide you with an appreciation of the experience behind this tutorial.
My foray into the world of interactive panoramas began around the same time
as the introduction of QuickTime VR panoramas. I didn't have the software to
create QTVR panoramas but instead I would stitch together a series of images
and then slice them up to separate files with some overlap and convert the
sequence to a QuickTime movie. Manually scrolling through the frames
would provide a crude panning effect. Apple then released their free
MakePanorama Tool and I was in business.
The image above was shot with a 24mm lens on a 35mm SLR and then scanned
in using a slide scanner. A normal tripod head was used, with the camera in
a portrait orientation. No geometric correction was applied to the images
and stitching took between 5 - 7 hours to combine 12 frames. I had tried
Panorama Tools but didn't have the patience and gave up.
I then progressed to my Hasselblad SWC which had a similar field of view
to the 24mm lens, but being a square format, did not require tilting the
camera. To take advantage of the extra quality of a 6cm transparency I had
the film professionally scanned prior to stitching, still without any
The extra image quality of my medium format panoramas gave me the
encouragement I needed to sit down and nut out how to use Panorama Tools. I
finally worked a few things out and cut my stitching time down to around 4 -
5 hours. There seemed only one way to get faster... cut down on the number
of frames. It was time for a new lens and, in the process, a new camera.
Armed with an 18-28mm zoom and a crude "panoramic" tripod head I shot a
series of panoramas for the Ararat Project 2.
A week after stitching them together I decided to buy an 8mm lens and be
done with it. I had to do these spherical panoramas I had seen and the
18mm lens was just not wide enough to do a single row of images and a one
image patch. Three months and an "incident" with the importer later and my
lens arrived. I promptly went out and shot 10 panoramas that weekend, and
another 6 the next. In the meantime I had managed to actually stitch one of
them together. Having seen the familiar tri-star join at the top and bottom
of other panoramas I had already planned to patch these areas with an
additional image before shooting my first panorama.
My first spherical panorama was stitched using a
very crude image remapping and manual
alignment. It still took 4 hours to complete but it proved to be
remarkably successful. I also discovered some of the additional benefits of
using a TRUE fisheye lens for image capture. Only one problem remained. I
was shooting much faster than I was stitching so it was time to knuckle down
and learn how to use Panorama Tools properly.
The other challenge which seems to strike most people who start shooting
spherical panoramas is how far can you push the limits of where you shoot.
What's the point of being able to look down if there's nothing to look at.
And so it was off to the Grampians and straight to the edge of the nearest
cliff... or even better, the "Jaws of Death", a 1m wide strip of rock
jutting out from the top of a mountain.
The next weekend took me down to Portsea for an AIMBI workshop weekend
where I gave a small talk on web design/publishing. This is where I found
"The Big Swing" at Portsea Camp. This would surely be a great challenge.
Shooting some 15-20m up in the air without a camera strap and hanging from
the end of a rope there was no way I was going to rotate the camera around
the lens' nodal point... and there was no way to get rid of the rope so I'd
have to get myself in the shot as well... but then I'd risk losing a hand as
it would be behind the lens pushing the trigger. Seven images later....
Cheating was becoming easier as I learnt more about PTools but it had
also set me back a little as the techniques I picked up along the way were
not entirely correct, even though they worked. Be warned... some methods
work some of the time but can go horribly wrong at other times.
Landscapes provided ample opportunity for "cheating" but an architectural
interior would not be so forgiving. I got lucky again with my first
interior (the majestic Royal Exhibition Building during the International
Flower Show) but then had problems with a second panorama shot only moments
later with exactly the same set up. I couldn't get it to work after repeated
attempts and it sat unstitched on my computer for a couple of months
Thinking of an image as being rectangular and spherical at the same time
tends to make my brain hurt but after a few more panoramas I began to find
ways of getting consistently accurate stitches and greatly reducing the time
factor. Key to this was they way in which control points are selected and
armed with this I returned to my stubborn interior and got such an accurate
stitch that I used a 100 pixel brush on my eraser for the stiches. And this
without aligning the camera over the lens' nodal point!
Even though it didn't seem entirely essential I ended up making my own
panorama head for my tripod. It worked pretty well but I must confess to
buying a commercial panoramic product to make it a little more practical
;-) Once I had my camera properly aligned over the nodal point things
got a lot easier and a lot more consistent enabling me to experiment with
different optimisation techniques to try and find a fool proof method.
The percentage of panoramas that don't work first time around is now much
smaller, but a little bit of experimentation remains, particularly for
And as you get things working the "right" way, you learn more about how
to do things the "wrong" way. This time a roof top panorama shot
through the windows of the Rialto Observation Deck. Roof top panoramas
aren't as hard as they may seem once you get the technique working.
And finally of course there remains the challenge of shooting panoramas
in places where cameras normally don't go... kayaking. ;-)
I have a few ideas left to try along this theme, like shooting a panorama
while surfing a standing wave in a river. The more you learn, the more
possibilities open up.... where it will end I don't know... and if I did I'd
probably stop trying.