This is a description of how I create the model railway panoramas.
First of all, it is important to recognize that the images produced are designed for model railway enthusiasts as a backdrop for their models, so the photography is about trying to produce an image that will enhance a model railway scene. Secondly, the final image needs to be capable of printing up to a large size (up to 12 foot long) without an unacceptable loss of quality. Some of the panoramas on the model railway website are composite pictures of up to 45 images; this means that the images are high resolution (at a resolution of 300ppi they can be printed up to approximately 72inches by 12inches).
The lens I use is a Canon TS-E f2.8 (Tilt and Shift) with a Canon EOS 5D mark II. I’m not going to go into a lot of detail here but refer you to two websites which contain the most helpful information I’ve found about Tilt and Shift lenses, which are the Cambridge In Colour and the EOS Documentation Project Websites. In essence the lens mimics ‘View Cameras’ in that the lens orientation can be adjusted independently of the Digital Sensor (instead of the film plane); the lens can be tilted (angled away from the sensor) and/or shifted (moved up or down, or side to side). These camera movements allow for extra control over perspective and depth of field, and so can be useful for some landscape, product and architectural photography.
The lens can also be used to create panoramas by shifting the lens from side to side. I don’t tend to use the tilt mechanism for the panoramas because it adds another layer of complexity and potential for inconsistency. The lens has no auto focus, but the focus points can still be used as they will light up if the shutter is half pressed down when focus is achieved. I tend to utilize hyperfocal focusing (Cambridge In Colour has a very clear explanation of hyperfocal focusing).
To create the panoramas, the camera is set on a level tripod and the first picture is taken with the lens fully shifted to one side, the second with the lens in the middle and the third fully shifted to the opposite side which gives a wide overlap for stitching in the post processing stage. In fact, taking 2 pictures with the lens fully shifted either side will give a sufficient overlap. I then pan the camera and take the next set of pictures, making sure that there is overlap between the sets. I then repeat the process up to 5 times. If the tripod head platform is level (or the camera consistently level horizontally) then the post production stitching process should not be problematic. There are tripod heads designed for taking panoramas, but I use a heavy Pan and Tilt head which has 2 spirit levels at right angles, to get a level platform.
So far so good, but I also add a further complication in that I use HDR. I often shoot the panoramas when the sun is high in the sky-to avoid exposure differences across the horizon, so the exposure in the sky at one end of the panorama looks similar to the other. There is still potential for exposure problems, e.g. between sky and ground, and in summer particularly, heavy shadows and HDR helps me get round these challenges.
My first step in post production, is to process the images using HDR Photomatix software and I usually batch process as there are a lot of images. I go straight to Photomatix, I don’t convert the raw file, to minimise inconsistencies occurring between images. I will use one of my own Photomatix presets which gives a fairly neutral result and lowish contrast and apply that to all images. I will save these files as Tiffs, not psd, as I will want the option of opening the file In Camera Raw.
The next step is to take images 3 at a time into Photoshop and use Photomerge. I click on the ‘cylindrical’ setting in the Photomerge layout menu which, by trail and error, I’ve found works best. I then flatten the image, and will often find there are small areas round the edge of the image which are blank. I select with the Magic Wand Tool, and use the ‘content aware fill’ introduced in CS5 (Edit>fill>Content Aware), which works consistently well. It’s possible, of course to do more than 3 images at a time, but I’ve found this way works best even if more time consuming. I’ll eventually photomerge the ‘photomerged’ images to produce the final size of panorama I want. So you can see how the final picture could be composite of up to 45 images.
As I have kept the files in tiff format, I will be able to open the files in Adobe Camera Raw, which can be especially useful to make basic adjustments but also to alleviate noise and colour casts which may be present (the noise reduction panel in Camera Raw in CS5 works well). I can do any final adjustments in photoshop.
Working with these size of files is very memory intensive; not only do they eat up storage space, especially with the working methods I’ve adopted, but unless you have a lot of Ram it will be very slow. Fortunately, my computer can support a 64bit version of Photoshop CS5, and with11GB of ram installed, I am able to process these files reasonably quickly.