I’ve had a few opportunities lately to help people edit their photographs where they wanted to combine two photos into a composite and were worried about the relative sizes being proper, especially when the camera settings and/or the scene were not identical.  Based on these experiences, I’ve created a couple of scenarios to introduce certain concepts.

## A Safe Selfie

Suppose you need to add part of a wild animal behind you – to make a safe selfie, if you will.  Most of your composite shots, where two objects are moved around in an image with plenty of other size reference points, fall in this category.   Generally, combining subjects is a two-part process:

### Use A Vanishing Point To Resize One Subject For Changing Distance

Now you can use vanishing points to maintain the correct sizes as you move your objects into place.  We’ve already explained that process in Using The Vanishing Point To Keep
The Size Right When Moving Wildlife Around
.  I would like to point out that as long as your object stays the same distance from the camera, or in the same focal plane, you can move it up, down, and all around without changing size.  If you move it closer to the camera, it should get larger.  When you move it away, make it smaller.  Once you resize it for its new distance, you can again move it up, down, and all around within that new focal plane at no extra cost.  Also, once you find the horizon in your picture, it doesn’t matter which point along that horizon line you use as the vanishing point; all of them will resize your object correctly.  Pick a point that is conveniently off to one side far enough to make long enough construction lines to give you some precision when changing size.

## A Beach Scene

I also helped somebody with a beach scene that invoked two simpler special cases of the resize problem.  The base or background image was a wide-angle beach scene and the photographer wanted to add objects that they took with a zoom lens at the same scene that same day.

### Floating Objects

The photographer’s intent, in this case, was to shoot objects floating on the water near the horizon with a strong zoom lens and add them to the picture so that they looked closer.  An object floating in the water is restricted to a specific plane in such a way that its distance from the horizon is directly related to its distance from the camera (within a camera’s normal field of view), which is the determining factor in that object’s relative size.  As long as the horizon is in the picture, this is no problem.  Whether you add that object at its original pixel size (as magnified by a telephoto lens) or even if you scale it further in Photoshop (by holding down the shift key to preserve the aspect ratio as you move a corner of the selected border while using the Move tool, for example), as long as you keep the horizon of the added object directly on the same line as the horizon of the background, the invisible construction lines from an invisible vanishing point will ensure the size and placement of your object are in agreement.  If the horizon is not in the picture, then you need to look for other size references and handle as in the first general case discussed above.

### Flying Birds

Birds (or other airborne objects) are even easier.  Their position is unrestricted and, more importantly, there are no other size references in view so there is really no way of knowing how large the object really is or how far away, meaning that if it were a ball, there would be no way for you to tell if it is a large ball far away or a smaller ball up close.  If you are familiar with the object and know how large cooper hawks are, for example, then your brain will automatically assign the hawk an appropriate distance based on its size when trying to make sense of the picture.  You can put that hawk just about anywhere and the viewer won’t know the difference.  Obviously, if you put a pigeon in a hawk’s talons then each would act as a size reference for the other and at least their relative sizes would have to match.  If they were not touching (or near enough to imply an interaction), there would be no such restriction.

## Other Considerations

### Lighting

There are other positional clues besides size to think about.  On a sunny day, an object’s shadow provides positional information, namely the object’s relationship to the sun, which must be consistent throughout the image (for best results).

### Perspective Changes

When you shoot someone’s face with a wide-angle lens from a close distance, it will not look the same as when you shoot the same face from far away with a telephoto lens.  An example of this is shown in the fourth image from the top at Choose the Right Lens to Make Flattering Portraits (the only image that’s in color). I’ve seen some experts blame this on lens distortion (as the guy in this otherwise great video at Focal Length for Storytelling – How Lens Choice Affects Your Images, but I don’t consider that lens distortion.  There is such a thing as lens distortion, but in this case, the subject’s nose really does look bigger and the ears really do disappear behind the cheeks if you were to close one eye and look at that person from 3″ in front of their face.  I call that a perspective shift and it is strictly a matter of angles and geometry, not lens issues. The “distortion” occurs when you take that image out of context by changing the perspective, which happens quite noticeably when you move an object from very far away to very close (or vice versa) in your image or if you take a 180° panorama, for example, and print/display it small enough to cover only 15°.  This perspective shift is virtually impossible to correct in Photoshop, so don’t go too wild while moving things around in your picture.  (Interestingly, it is by a lack of any perspective shift that you can catch somebody who created a reflection in their picture by just adding a flipped subject in post-processing.  The explanation of this tangent to today’s topic would require a separate article, however.)

As it turns out, I am not the first to call out the lens distortion theorists. See Daniel Baker’s blog Face distortion is not due to lens distortion.

Well, that’s about everything I know on this subject.  Please feel free to contribute your own hard-earned understanding of this issue for the betterment of photography in the comment section below.  Thanks.

## How We Digitally Stretch Our Gallery Wrap Edges Before Printing

As we discussed on the Services page of our website, we digitally “stretch” our image before wrapping it around the edge of our gallery-wrapped canvas images. Here’s how we do that:

Our gallery wraps are either 3/4” thick or 11/2“. On the thin ones, I usually take the 1/4” strip along the edges and stretch it to 1″, thus having an extra 1/4” to wrap around to the back side to cover for variations in the printing and stretching processes. On the larger ones, I take 1/2” and stretch it to 2″ (thus leaving 1/2” on the back). I wouldn’t stretch the image more than four times its original size, but you could go less. To do that, you would effectively be taking a wider margin to wrap around the side.

As an example, if I want a 12” x 18” image stretched around a 11/2” frame, I would crop the image to 13” x 19”. Then, after putting guides 1/2” in from each edge and another guide right on each edge, I would increase the canvas size 3” in both dimensions to get 16” x 22” with the image centered.

1. Click Image ⇨ Canvas Size…
2. Put a check in the Relative Box
3. Make Width and Height 3 Inches
4. Make sure Anchor dot is in center of the grid
5. Hit OK

I would then use a scale transform to digitally stretch the outermost 1/2” to 2” wide, filling the canvas.

1. Make sure Snap is checked in the View Menu
2. Use Rectangular Marquee tool to select the 1/2” strip between the guides along one of the edges
3. Click Edit ⇨ Transform ⇨ Scale
4. Place the mouse cursor over the little square in the middle of the outer edge of the selected area and drag to the edge of the canvas
5. Hit the check mark to finish the transform
6. Repeat Steps 2 through 5 with the 1/2” strips along the other three edges

(Actually, I first do the four corner squares separately, but since only a small bit along the edge of those squares has any chance of being seen, you could include them in either the horizontal or vertical strips (or even both)).

Then I add a blank (transparent) edge around the image representing the canvas I need for stretching the canvas around the frame by increasing the canvas size by double the required margins in both dimensions, the same way we did above. That margin would be at least the width of the moulding along the bottom (1″ for the 11/2” moulding we are using now) and enough extra to get a grip with the canvas pliers (for me that’s at least 3/4“). That would make the image’s final dimensions at least 191/2” x 251/2“. When I am finished, I add layers with cut lines, fold lines, staple lines, positioning marks for the hanging hardware, etcetera, but that is a personal matter beyond the scope of this article.

## Making “Eclipse Over Long Pine Key”

Our latest image, Eclipse Over Long Pine Key, of the solar eclipse in August in the Everglades was by far our most complicated yet. While spending hours and hours overcoming challenges in post-processing, I wondered if I was wasting my time – would anybody even be interested in the results and was each of these steps really necessary, or was I just over-thinking a problem again. You be the judge.

# Another Brilliant Idea

Although I wrote a blog article with suggestions for taking eclipse photographs with either your smartphone or camera, we had no plans to take any ourselves. Then, about twenty hours before the start of the event, I got another ‘brilliant’ idea – the concept for a multi-Gigapan panorama and time-lapse image of the solar eclipse.

What I pictured were images of the eclipsing sun as it soared barely over the tops of the skyscrapers in downtown Miami. This image would be taken from a balcony somewhere between the tenth and twentieth floor so you could capture an interesting street view in the foreground. Of course, it would be a Gigapan to give plenty of detail (we’ve been influenced and inspired by fellow Miami photographer Robert Holmes), but one panorama would not be enough. Ideally, as you took each of your carefully placed sun shots, you would need to shoot an area of the city directly beneath it so that the constantly changing light intensity and the buildings’ shadows would change in synchrony with the sun. The sun photos would be taken with a camera with about 16-stops of neutral density filter. For ease of execution, it might be better to have a second camera for the Gigapan unit. Each associated Gigapan panorama would need to have enough overlap with its neighbor to be able to stitch them together and go high enough to capture the sun’s position so you could correctly add in the better sun shot later, and give enough headroom for the sun’s path.

# The Real Work Begins

## Stitching The Panoramas

Photoshop does pretty well putting together smaller panoramas, but bogs down as the number of images grows (for small panoramas, Canon’s PhotoStitch, which came with the camera, does an even better job). Gigapan Stitch, which comes with their motor drives, does pretty well on the larger panos, but I usually use Kolor’s Autopano Giga because it has more choices in projections, better control options, and does better at eliminating ghosting (which happens when things (including even trees and branches) move between shots). I’m still climbing the learning curve, which added to the time needed to stitch together all seven panoramas. I believe there must be a way to combine the stitched panoramas into one large image with Autopano, but I couldn’t find it in time for this project, so had to warp and stitch the individual panoramas together by hand. Unlike my imagined cityscape, the ground location of the camera and the intervening lake make the concerns about the shadows much less significant. Because of that, and the fact that each of the panoramas was larger than strictly required, I was able to cover the field with only two of the seven panoramas, but then added the last panorama – the one with clouds.