Taking Handheld Panoramic Images
4. Coit Tower
Handheld Panorama
As further evidence that handheld panoramas can be successful without
worrying about parallax, take a gander at the following image.
I was
sightseeing on a sunny day in San Francisco and wanted to capture the
view from Coit Tower. If you are not familiar, it is a cylindrical
building with windows all around. I shot four images from each window,
left, right, up, down. And then stitched with Autopano Pro. Is there
parallax? Hell yes. Are there stitching issues? Uh huh. But the 44
images capture San Francisco in the round. At full scale it is over 100
million pixels. The stitching errors aren’t visible until you zoom way
in, and there aren’t too many of them. Honestly, the biggest
problem is the reflections in the glass windows. When I have some time,
I will finish this one up, and clone out all the reflections. I
promise, it looks good enough right now poster sized.
This is the worst stitching error, and it doesn’t affect much more than
the bridge and one pier.
Handheld Panorama Conclusions
While keeping your camera level and rotating it consistently will help
you produce a panoramic image, they are not strictly necessary. If at
sometime you find yourself tempted to take a panoramic series, but
don’t have any fancy equipment handy, fire away. Keeping the camera
level will decrease the amount of scenery lost in the final crop, so
practice your moves and plan ahead. Parallax is less of an issue if
objects are farther away. Landscapes work great, but interiors or
crowded markets might be overly ambitious. If this article has gotten
you interested in making panoramas, you might like our article on
Autostitch, a free automated stitching software.
Resources
More images by Jook Leung can be found at http://360vr.com/jook/
or http://360vr.com/tealounge/.
Panoguide
has lots of information on theory, technique, and results
Here are a couple 720 degree panoramic images from the world wide
panorama project: #1,
#2.
VR pano examples:
Tokyo VR is a
great collection of images, including night scenes.