Stitchery

From the mailbag:

Hi Jim,

I have been following your landscape photos and they are absolutely brilliant. I saw you mentioned something about stitching and was wondering if you could detail your method.

Reason I am asking is that I am going to get rid of my A7r in preparation for the M2 and will only have my A7II and I have holidays planned. Most of what I do is either landscape/architectural. And I am sure I am going to miss the A7r for those trips.

Occasionally, I stitch three or four images together. That, however, is the exception. The rule, for me, is to stitch between a dozen and over 100 images. Therefore, when I usually do and what will suit you is probably quite different. As I understand your question, you’re looking for a way to extend the 24 megapixel resolution of the a7II into the 36 megapixel or somewhat above range. If you normally make images in landscape orientation, you can get to that kind of image by rotating the camera into portrait orientation and making a single row of three or four images with a lot of overlap.

If all the elements in your picture are distant — being defined as further then 50 or 100 focal length of the lens you’re using — you won’t need to rotate the camera around the entrance pupil, or even, in many cases, put the camera on a tripod. I do handheld captures all the time, and it works very well. If you do have objects in the scene that are closer than 50 or hundred focal lengths, it would be best if you purchased a tripod mount that would allow you to locate the camera so that a line drawn perpendicularly through the entrance pupil of the lens passes through the rotational axis of your tripod head. If you’re making single row stitches (or single column stitches) the equipment can be fairly minimal. If you want to make multi-row stitches, the devices for position in your camera get more complex. My favorite supplier for this kind of equipment is RRS. They’re not cheap, but their gear is very well designed.

There are two main difficulties with stitching.

The first is motion. If your subject is moving, the stitching software is likely to become confused, and you’ll end up with “ghosts” in the rendered image. These can often be dealt with by manually controlling the details of the stitching, which the more advanced stitching software programs allow you to do. But, if there’s much subject motion, you may end up with a set of captures that cannot be successfully stitched.

The second is distortion. Most stitching programs allow you to choose from either a small number or a dizzyingly complex array of rendering projections. That’s all well and good, but if the stitching software has misunderstood the geometry of the scene from the captures, any geometrical distortion will carry forward into the rendered image, no matter what projection your choose. This is not normally a problem in landscape photography, but can easily become a problem in architectural work. Again, the advanced stitching programs allow you to control the details of the geometry of the scene. That’s the good news. The bad news is that there is usually quite a learning curve involved in taking advantage of these controls.

Okay, let’s assume that you have a good set of captures. What’s the best way to stitch them? There is stitching software built into both Photoshop and the latest version of Lightroom. It is rudimentary, but convenient. If I were you I’d give it a try, especially the Lightroom version, which is quite easy to use.

If you’re ready for more advanced stitching software, I can recommend two programs. The first is AutoPano Giga, which I use more than any stitching program. The second is PTGui. AutoPano is better at making the choices for you automatically. PTGui is good if you want to roll up your sleeves and wrest control from the program.

I would be remiss if I didn’t mention that you can also make captures for stitching with multiple exposures with a tilt/shift lens and not rotating the camera. There are those who believe that that technique offers better control of distortion.