Adding Up the Pixels (plus, JPEG sucks!)

November 19, 2016  •  Leave a Comment

We'll start with a simple bit of mathematics (hang in there — the key word is simple):
20 x 12 x 100 = 24,000.

That's "20 feet" x "12 inches per foot" x "100 pixels per inch" = "24,000 pixels": the number of pixels needed to span a single image at the back of your 20-foot-wide trade show booth.

Here is such a booth, created for Donaldson (makers of filtration systems for helicopters and many other types of vehicles or applications).

The number of pixels, top to bottom, is 8 x 12 x 100 = 9,600. So, if you are contemplating a new exhibit graphic that is 20 feet wide and eight feet tall, those are numbers you'll want to keep in mind: 24,000 x 9,600.

Grab a camera that captures 24,000 pixels by 9,600 pixels and stab that shutter button, right? You don't have a camera that captures 230 megapixels? Now what, you ask?

First, I'll note that while cameras exist that capture very nearly enough pixels for a 10-foot-wide booth graphic in a single shot, most photographers don't have that kind of camera (they run about $50,000). It's the ideal solution, but most of my clients choose to make do with some other, less expensive, approach. Here are a couple of those.

Blow it Up
This is the most common approach to filling a large area with pixels — use software to just make more of them. This requires the computer to examine the extant pixels and extrapolate additional ones. It sounds easy and, for the person typing a couple numbers into a dialog box, it is. Easy does not mean perfect. Acceptable, perhaps, and the quality of the resultant image is highly dependent on the amount of enlargement and the quality of the original.

So, first, let me get the topic of image quality out of the way, and the most common source of poor quality — JPEG! Not to go into the depths of the topic, but rest assured: if you begin down the path of making a really large image, you need to begin with an excellent one. If the file type you are using is, or ever was, a JPEG then you might be severely handicapped. Here are two versions of a small section of a photo — one had never been a JPEG (always a RAW or TIFF) while the other one made a stop through JPEG-land. Each was then enlarged by 400% using the same method. If your original image looks like the latter, I urge you to find another because it's not going to get better just because it's bigger. Seriously. With my short rant about the evils of JPEG out of the way, the path is simple enough. Use an app to digitally "enlarge" the image. I use several methods, depending on where in a workflow I am — 
Photoshop does a good job, though I usually take care of the enlargement earlier in the process, at the same instant I am converting my RAW file into a further-editable format (almost always TIFF).

Another app for adding pixels, also available as a plug-in for Photoshop, is 
Perfect Resize from onOne. With default settings its results are comparable to Photoshop's, but with attention to its many controls the results can be superior, again depending on the details and quality of the original file.

Let me be clear — enlarging does not result in an image as clean and detailed as one shot at the proper size but, like I wrote above…only those willing to pay get all the best pixels.

Piece it Together
This method is possible if the subject can sit still for a bit, as it consists of photographing overlapping portions of the subject and "stitching" them together in the computer. The image below is the result of 16 images, side by side, and is 20,550 pixels wide without any enlargement. Blow it up 17% more? No one would notice a smidgen of difference. These panoramas also can be shot in columns, one above the other, or in any combination of columns and rows to cover a static subject or to fit a space. (I've shot landscape panoramas comprised of over 100 images that would print to over 90 feet wide!) Obviously, the technique will not work for moving subjects, like air-to-air aircraft or factory floors with objects or people in motion across multiple frames.

If you're thinking of shooting a panorama, do a little research to learn the technique and, perhaps, look into specialized equipment that can lead to easier, more accurate, stitching. Since the image, small as it is in this article, doesn't do justice to the full-size image quality, here is a detail, of half of the spinner — looks like that spinner needs some elbow grease and spinner polish.

A fast computer helps, as the individual images you shoot do not just click together like a jigsaw puzzle, but more like a puzzle with malleable pieces which must have their proper shape discovered before they will fit with the other, also malleable, pieces. It's a bit of magic how the software takes care of the many factors (can you say "geometric distortion," "parallax," and "vignetting?").

Photoshop can take a stab at the stitching, with limited controls, while I usually turn to AutoPano Pro from Kolor. There are other apps, too, so search the web for panorama software to find alternatives to suit your budget and skill level.

Nothing is as good as having the requisite number of unadulterated pixels when it comes to putting images in front of clients and prospects. The subject should be the single focus of a viewer's attention. If they are distracted by artifacts or soft details or noise (we didn't even talk about blowing up a noise-filled image!), then they are thinking about something other than what you intend, and it's also a ding to your branding efforts.

Start with the highest quality image you can make or afford, then do the best you can to make it work.

About the Hyperlinks
I want to point out that I've provided hyperlinks in the article to ease your search for additional information about a company or product. No one is compensating me for these; I'm just being a nice guy.


No comments posted.
Photography is a massive field. Aerospace photography less so. In these writings I share stories and tips.
January February March April May June (2) July (2) August (1) September October (2) November December (1)
January February March (2) April May June July August September (6) October (15) November (2) December (2)
January (2) February (2) March (2) April (2) May (2) June (2) July August September October November December