Panoramic photography is as old as photography itself. Stitched panoramas were being produced during the American civil war in 1864. Many specialised film cameras were produced over the years and more recently specialised digital cameras have appeared.
Professional digital panoramic cameras are very expensive and so stitching of images taken by a ‘conventional’ digital camera is the best way forward for serious amateurs. Recent models which produce a panoramic image from a video ‘sweep’ appear to be limited to HDTV resolution, producing an image only 1080 pixels high, sufficient for projection but not sufficient for a decent print.
How you shoot images for stitching depends mainly on the subject matter and light levels. For landscapes in good light (where a hand-held shot might normally be used) a tripod is not necessary, provided there are no objects closer than about 50 metres. Objects close to the camera can cause parallax problems if there is significant lateral movement of the camera between shots (adjacent images will not match well and will be difficult to stitch). Parallax errors can be completely eliminated by using a tripod and a panoramic head. This allows fore and aft adjustment of the camera position so that in panning it rotates about the entrance pupil of the lens.
When using a tripod the axis of rotation should be vertical; this can be checked with a spirit level but a final adjustment can usually be made by looking through the viewfinder. For example, in photographing the interior of a building, the centre point of the viewfinder should follow a horizontal line on the wall at the same height as the camera as the camera is panned. In photographing a landscape, two or more features at the same height can often be identified by reference to the Ordnance Survey map.
My favourite panorama lens is a 24mm prime. My digital SLR has a 1.5 crop factor (Though widely used, ‘crop factor’ is a misnomer – nothing has been cropped) which means that it is equivalent (in field of view) to a 36mm focal length lens (24mm x 1.5) on a 35mm format camera. When used with the camera vertical (i.e. in ‘portrait’ mode) the field of view is 38°. When using this on a tripod I rotate 30° between shots, ensuring a consistent overlap. When shooting hand-held just make sure that images overlap by about 20% of the width. Having said that, modern stitching software is very tolerant of variations in overlap etc. I don’t always carry the 24mm prime lens and I have had good results from a 17-70mm zoom at its widest setting.
The advantage in using the camera in ‘portrait’ mode is that you get a larger vertical field of view at the expense of more shots to stitch to cover a particular horizontal field. If you are aiming for quality it is best to avoid very wide rectilinear lenses, particularly zoom lenses at their widest setting. This comment is clearly less applicable to ‘professional’ lenses than to those at the ‘budget’ end of the market.
When I stitched my first panorama in 2002 I used Panorama Tools, a suite of small programs for remapping and stitching developed by Professor Helmut Dersch. These are now much easier to use via the Graphical User Interfaces (GUI) PT Assembler ($45) or Hugin (pron. ‘Hoogin’) which is open source and is free of charge.
The End Product
Having stitched a panorama, or preferably before, we should consider what is to be the end product. Is it to be a projected image, perhaps six or eight feet wide but limited to 1400 pixels wide? Or a print restricted by some competition rules to a width of 50cm? Or a wider print restricted only by the number of pixels in the digital file (and perhaps our budget)?
With a digital SLR with a sensor of say 6 megapixels, probably 3000 x 2000, used vertically to create a 180° panorama in six shots, we would expect the final image to be about 10,000 pixels wide. (With no allowance for overlap, this would be 6 x 2000 = 12,000; allowing for overlap 10,000 is reasonable). I chose a 6 megapixel camera for this example to make the numbers easy; more typically we could consider a 10 megapixel camera, which would give a width of 12,910 pixels, almost ten times the number of pixels in the width of our projected image – what a waste!
If we print our panorama (from our 6 megapixel camera) at a resolution of 200 pixels per inch (a sensible minimum) the result will be 50” wide x 15” high (a 10 megapixel camera will allow an increase of 30% on each of these dimensions, to 65” x 19.5”).
If you are producing a print for a competition you will usually be limited by the maximum mount size of 50cm x 40cm. With this restriction, our example print will have a maximum size of 50cm x 15cm, an aspect ratio of 50:15 = 3.33:1. This is about as far as we should go, anything long and thin, at this size, will not impress competition judges. Note that at this size our resolution will be about 500 pixels per inch (from our 6 megapixel camera) – ample by any standards.