As many of you know, I’m a strong advocate of getting everything right in camera, including cropping. A summation of my stance might be:
Right place (perspective)
Right lens (crop)
Right time (moment)
Optimal settings
Obviously, you can’t always get that 100% right in the field. It’s a matter of how close you can get that determines how strong the image you end up with will be.
One pushback I get a lot, particularly now that we have high megapixel count cameras, is that “you can always crop.” Sure, but that means that you’re not in the right place, you’re not using the right lens, and it’s possible that you’re not at optimal settings (particularly if you’re using an automated metering method). The closer you can get to being in the right place and using the right lens, the easier it is to get the other two things right, as well.
However, things have changed a little bit in modern times due to the things that we can do in post processing. I’ve found that there are a few times when I might intentionally crop somewhat looser these days, and it’s worth identifying what motivates that:
- Perspective correction. Particularly with wide angle lenses and large objects (e.g. buildings), there’s a good chance you’re tilting up. Doing so will give you a perspective distortion that you may want to correct. Parallel lines will converge as you tilt up, and we have excellent software tools that will make them parallel again. The only problem with that is that if you don’t frame wide enough, you’re going to get edge crop problems. Simply put, if the camera is pointed up and down from horizontal and you have parallel lines that need to stay parallel, frame wider than necessary.
- Distortion correction. Similarly, a lot of modern lenses are not well corrected for linear distortion. Wide angle lenses tend to provide considerable barrel distortion these days, while telephoto lenses will tend to produce pincushion distortion. Those pesky parallel lines will become curves if not corrected. Most wide angle lenses these days are slightly wider than their stated focal length, because they’re assumed to be corrected by software either in the camera or in post processing. But that same processing can start to clip information at the corners yet might not resolve the entire distortion, so again, you should probably frame just a tiny bit looser than usual to be safe. The good news is that mirrorless cameras are WYSIWIG, so you generally see the correction in the viewfinder and can frame more precisely. But DSLR owners don’t have that benefit at all, and need to be careful about corners.
- Stitching. If you like to make wide photos via panoramic stitching, taking images with more overlap is probably better than less (up to a point). Somewhere between a quarter and a third of the frames you take for the sequence should overlap, slightly dependent upon the lens you’re using and how much vignetting/distortion it has. The more distortion and vignetting, the more overlap you should use.
- Negative space. Back when I produced more work for editorial (magazines), I tended to take the images the way I saw them, but then also would frame a similar image that had plenty of negative space on which a magazine (or ad) designer could put headlines, intros, text, logos, page numbers, and more. This is a tricky, tricky compositional problem, as if you don’t think like the designer(s), your images don’t get used. You had to think about subject flow, eye flow, page flow, and more. For instance, I found that a strong composition on the left with plenty of negative space on the right (or vice versa) often became an opening image for an article as a two-page spread. I have to admit that this was a strategy that I had difficulty with mastering. I’m focused. Negative space is less focused.
But all of the above unfortunately once again begs the question of how many megapixels do you need?
Let’s start with my initial four variable premise (in bold, second paragraph). Assuming you’re right, right, right, and optimal, figuring out how many pixels you should be using is simple: simple consider what’s the largest output you’re likely to produce.
These days I have people give me all kinds of answers to that: iPad, 4K or 8K screen, 19” short-axis print, and so on. Okay, let’s just take those four:
- iPad — an iPad Pro 12.9” diagonal is 2732x2048 pixels. The 11” is 2388x1668 pixels. That’s 5.6mp and 4mp respectively. Every current camera can deal with that. You don’t need a huge megapixel count.
- 4K screen — here we’re at 3840x2160 pixels, or 8.3mp. You still don’t need the top megapixel counts.
- 8K screen — this doubles both directions to 7680x4320 pixels, or 33.2mp. We’re approaching the top common pixel counts (45-50mp) while retaining some cropping ability.
- 19” print — at 300 dpi that puts us at 48.8mp. We’re now pushing many top APS-C and full frame cameras to their limits. You don’t want to be cropping the source input.
Note that there are aspect ratio issues that come into play with those statements. The 4K/8K screens are 16:9 aspect ratio, while our cameras are configured as 3:2 or 4:3 aspect ratios. So you may need a few more pixels than just suggested, depending upon how your capture aspect fits your output aspect.
Still, in case you didn’t notice, our highest need is that 19x29” print, and that pixel count is pretty darned close to the higher end cameras we’re using today (45mp, 50mp, 61mp). Right place with right lens nets you a pretty darned good large print. I rarely stray above that, and usually only because a client requested it (I’ve had a few large poster requests over the years).
The problem that occurs when you crop that 45mp camera significantly is two-fold: you lose the ability to output that 19” print at 300 dpi, and thus you technically are scaling up noise (loss of usable dynamic range). This was my problem with the initial 61mp Sony (A7R Mark IV): it was already struggling to provide me everything I wanted in the dynamic range department at full size; any cropping made things worse. (The A7R Mark V seems to be much better in this regard, even though it's a very modest underlying improvement.)
So what happens if you’re a “significant cropper”? By that, I mean that you’re simply not in the right place with the right lens, so you’re using 10%+ cropping in post processing to get your final image properly composed. Essentially the same thing that I just complained about with the Sony: you lose the ability to print max size at 300 dpi, and your usable dynamic range suffers. Adding pixels with current technology tends to only fix one of those things.
Personally, I’m happy with the 40-61mp sizes of the top recent APS-C and full frame cameras, though 40mp APS-C and 61mp full frame pretty much require you to answer my four points at the top as right, right, right, yes.
And now we can get to this article’s title. Where do you stand? (1) Are you in the right place? (2) Are you using the right lens? (3) Are you taking the image at the right moment? (4) Are you using optimal settings? Add to that: (5) do you really know what your desired maximum output really should be?
If you can’t answer yes to all five of those questions, buying a higher pixel count camera isn’t likely to solve any problems for you, but actually increase your problems.
So, for each question you answered no to, figure out why and what you need to do to reverse that answer. It’s not likely “buy a new camera."
