Statement


The only thing I miss about working in the darkroom is the image slowly appearing on the exposed paper after I put it in the developer tray. The hours of picture taking followed by the drudgery of loading the film with sweaty hands onto stainless steel reels in a damp, totally black basement culminated in that magical moment, when I could actually see what I had only visualized when I released the shutter.

In 2005, after a year or so of the instant gratification of digital imagery, I began recreating that “darkroom moment” by combining multiple shots into one larger, “stitched,” image. As I take the individual pictures I create a grid of rows and columns in which each individual shot overlaps a bit with the shots on either side as well as with the ones above and below.

While making such a picture grid, the final image is unseen and wholly imagined. Unlike the darkroom days, when at least you saw the complete image in the viewfinder, you don’t see the complete stitched version until you push the merge button in Photoshop, and then depending on the number of individual shots and the amount of pixels captured by the camera’s sensor, you may have to wait up to fifteen or twenty minutes. However, when the merge does finish, the visual impact is just as magical as when the image stared up at you through the developer, and as an added benefit, you don’t have the smell of all those chemicals.

Shooting Technique

When shooting a picture grid, the camera should be centered on both the pan and tilt axes, and it’s helpful to use a pano head marked with degrees on both axes, because then you can memorize how many degrees to shift the pano head for each length lens and eliminate having to check either the viewfinder or the live view. Utilizing the degree markings will thus make your grids more uniform as well as speed up the shooting process.

To avoid parallax when I first started creating grids, it was necessary to also compute the nodal point–the point where the light crosses within the lens–and to place that point at the center of the pan and tilt axes. Finding the nodal point is a cumbersome and tedious process and it has to be computed and then recorded for each lens, or in the case of compound lenses, for each significant length. Using this nodal information when shooting requires a third rail on the pano head to move the camera back and forth. Over the years improvements in merging software have lessened if not eliminated the need for this additional complication and I haven’t worried about nodal points for the last six or seven years.

One technique that will help alleviate problems when you merge your images, especially ones that have a strong vertical orientation, is to only tilt the camera in one direction during the shooting process. Your pano head should have a bubble level on it, and the bubble should be in the center of the circle before you take any shots. Once you have everything set up, you don’t want to cross the 0 point on the tilt scale–the point where your lens is perfectly horizontal. You can shoot multiple rows from 0 down or multiple rows from 0 up, but if you shoot above and below 0, your merge will have bulge on either side the 0 row, which even if you are quite accomplished at post-processing manipulation, will be hard to eliminate in a visually believable fashion. You can get away with crossing the 0 orientation by a small amount, but the further you go past it in either direction, the more problems you will encounter when merging. Thus, if you are shooting multiple horizontal rows, set up your tripod so that you are either tilting the head only up or only down. When shooting buildings this often means that you have to set your tripod so that your camera is fairly close to the ground.

Merging The Shots

Initially, I used third party software to merge the individual shots, but for the last ten or twelve years I’ve only used the merge functions in Photoshop–there are two, one in Camera Raw and one in Photoshop itself. The one in Camera Raw is better because it gives you more control over the process and blends the color and contrast in a more uniform fashion, however it has pixel limitations while the merge function in Photoshop does not. A workaround for Raw’s limitations is to separate your individual shots into smaller groups and then do multiple Raw merges which you can then either merge using Photoshop’s merge function or merge manually in Photoshop.

Whether you use the Raw or Photoshop merge functions, your merges will seldom be executed perfectly, and will require some manual intervention. Overall, you will often have to warp an image to create level horizontal and vertical lines. This is especially true for cityscapes or images of buildings, i.e., any merge which has well defined horizontals or verticals, but it also happens in landscapes, because the automatic merge function simply can’t get all of the overlapping detail to fit together properly. When these problems occur, you have to select the misaligned areas, place them on a separate layer, and warp, clone or otherwise manipulate them into conformity. Then you can feather mask them back onto the merged layer.

There will be times when the automatic merge function will fail altogether. In this situation you can either try merging the individual shots into multiple smaller merges, and then work these merges together to create the final overall image, or you can try warping your individual shots to line them up before your initial merge. As you accumulate experience, with enough effort you will be able to merge any grid together, one way or the other. Successfully merging images is an acquired skill not an inherited talent.

The Benefits of Stitching:

More Detail–Larger Prints

Despite the potential difficulties and extra effort, the benefits of stitching multiple shots into one image are substantial. For example, I shoot uncompressed raw with a Sony a7r Mark 4, and each shot is 6,336 pixels by 9,504 pixels. If I use a 25mm lens and stand 16.5 feet from my subject, those pixels, all 60,217,344 of them, are covering a rectangle that is 23.5 feet wide by 15.75 feet tall, which equals 1,136 pixels per square inch. If I shoot from the same distance with a 100mm lens I would need a five by five grid (five rows with five shots each) to cover approximately the same size rectangle, assuming a ten percent overlap between the images. However, with the 100mm lens I now have 1,219,401,216 pixels covering that 23.5 by 15.75 feet, or 23,000 pixels per square inch–over twenty times the number I got with the single shot using the 25mm lens. There will be a lot of information in those nearly 1.16 billion extra pixels. If you think about it in regard to print size, the comparison is similarly lopsided. If I print the image taken with the 25mm lens at 300 dots per inch, I end up with a print that is 31.6 inches wide and 21.1 inches high. If I print the stitched image at 300 dots per inch I end up with a print that is 143 inches wide and 95 inches high, in other words I have a print that is 4.5 times the size with exactly the same level of resolution and orders of magnitude more detail.

Zooming In

The above example illustrates another benefit of stitching. Robert Capa, the renowned World War II photographer, admonished, “If you don’t like your pictures, move closer”. When you use a 100mm lens instead of a 25mm one, you are effectively moving 4 times closer and the resulting image will thus pack correspondently more punch. My rule of thumb is to always use the longest lens possible with two caveats: depth of field issues (which I will discuss below) and computer processing power. The benefit of stitching is that you have more pixels to work with, however at some point those extra pixels shift from being a help to a hinderance. That point is reached when your file size makes processing your data too cumbersome to continue. I therefore choose a lens that will give me no more then twenty to thirty images, although I have stitched as many as forty-two (an image of the Flatiron building in New York City).

Customize Your Form Factor

When you limit yourself to single shots you are constrained by the format of your camera, even though the scene you are photographing most likely doesn’t quite conform to that ratio of width to height. So you move to a shorter lens to get everything in, or you crop to get unwanted things out. Neither option is optimal. However, when you shoot multiple images, you can size your grid to correspond with the scene you are trying to capture–stitching allows you to choose the format that fits the image, nothing more, nothing less.

Include Only What's Importantt

Especially when shooting in an urban setting, it is often difficult to capture a scene without extraneous elements–one car finally moves out of the image but then another drives in from the opposite side. When you are zoomed in taking multiple shots it is much easier to isolate exactly what you choose to include because your range of view is much smaller. When shooting cityscapes I always shoot the background first to set the scene–no people, no cars, simply the built environment. Then I shoot the things I wish to place on that stage, whatever I feel fits my conception for the overall image. If you are already merging a number of pictures into one, it’s not much more work to mask in objects that amplify your idea.

Match Individual Exposures to Conditions

It is often the case that different areas in an image would benefit from different exposure settings. You can accomplish this alteration with masks in post-processing, but Voilá, when shooting multiple shots you can alter your settings for each individual picture. If you have a hot spot, increase your shutter speed, if you have movement, raise your ISO and shorten your exposure time–the possibilities are almost endless.

Depth of Field

Depth of field is what distinguishes scenes shot through a lens from scenes viewed with your eyes. When you look at an object, your eye/brain combination does not perceive any areas that are out of focus–it never happens. Yet, when you photograph that same object through a mechanical lens, most of the time some area of the image is out of focus. At times this shallow depth of field is an artistic choice, but even when that is the case, it is often an artistic choice that is dictated by the limitations of the camera/lens combination. In fact, after the many years in which pictures have dominated our culture, we have come to strongly associate shallow depth of field with what it is to be a photograph.

I have a strong bias toward images that are in focus throughout–where the foreground, middle and background are all in focus, and the vast majority of my images reflect this bias (except for the times when I make an “artistic” choice). People often comment that my prints have a “painterly” quality. That’s not because paintings are normally filled with minute details, it’s because paintings rarely contain areas that are visibly out of focus in relation to other areas within the painting.

As you have probably already figured out, wide depth of field is another benefit of merging a grid of shots into a single image. Even if you are using a long lens, which has a correspondently shallow depth of field, shooting in rows helps keep everything in focus because each row is usually in the same plane, and you can adjust the focus when you shift to the next row. If the distance from the camera is not consistent within the row, you can also adjust the focus before taking each picture.

Focus Stacking

There is an additional technique that is useful when changing the focus with each individual shot is not adequate to maintain a sharp focus throughout the image, and this technique is especially useful in macro photography–focus stacking. In focus stacking you have to use a tripod, but instead of shifting the camera after each image you make multiple shots from the same position. In this photo stack, the first image is focused on the part of the scene that is closest to the camera, and with each successive shot you focus a little further out until you have captured everything that you want to be in focus. You then process the stack with software (I use HeliconFocus) that creates a final image from the in focus areas of each of the separate pictures in the stack.

Most people use focus stacking as a standalone technique, but I use it to replace the individual elements in the grid. For example, I compiled an image from a three by six grid (three shots in each row and six rows). I was shooting the reflection in a puddle alongside a curb with a 400mm lens that was ten feet from the center of the puddle–the depth of field for a 400mm lens at ten feet is just slightly less than four inches, about two inches on either side of the focal plane. I thus had a grid of eighteen elements, but each element was the result of a focus stack made up of from four to seven shots–the final image was the combination of 101 individual shots which could only have been produced with the combination of stitching and focus stacking.

Video

I have recently begun to make 4K ultra HD videos (3,840 pixels by 2,160 pixels). Large stitched images with tens of thousands of pixels work well in this format. If you convert them to Smart Objects in Photoshop, you have ample room to pan around, to zoom in and out, creating considerable movement that doesn’t require dollies or cranes.

Conclusion

I have described key elements and benefits of the stitching process and the images on this website illustrate the points I’ve raised. If you have further questions, let me know.

Heath Paley



Contact information:

By email: povheathágmail.com

By phone: 207.756.9512


Heath Paley
Portland, Maine


Represented By: