# How much more useful are ever higher pixel densities?

Started Feb 18, 2013 | Discussions thread
Rule of thumb

I'll take the liberty to propose a rule of thumb: the "practical resolution" of a camera is defined by -1/(#MP), where larger (i.e. less negative numbers) are better. For example, a 10 MP camera has a practical resolution of -0.1. Of course I'm using the term "resolution" very loosely, since a number like -0.1 is meaningless by itself. I don't suggest this as an absolute scale, but as a relative scale to compare resolutions.

For example, when going from 6 MP to 12 MP, the difference is -1/12 + 1/6 = 1/12 unit gain in practical resolution. When going from 16 MP to 24 MP, the difference is -1/24 + 1/16 = 1/48 unit gain in practical resolution. So in practice, an increase from 6 to 12 MP is four times more important than an increase from 16 to 24 MP.

The advantages of using the negative reciprocal are debatable, but significant. The other two commonly used functions for evaluating MP increases are linear and logarithmetic. To the newbie, a 2 MP increase is always a 2 MP increase, whether it's from 4 MP to 6 MP or from 18 MP to 20 MP. This is why point-and-shoot manufacturers keep raising the number of MP in their cameras even though there is little gain to be had. To more savvy people, it is all about the percent increase, so an increase from 4 MP to 8 MP is the same as an increase from 8 MP to 16 MP. In mathematical terms, they consider improvements to be equivalent if they have the same increase in the logarithm of the number of megapixels.

However, I suggest that this is not the case. Is going from 4 MP to 8 MP really the same as going from 8 MP to 16 MP? 4 MP is fine for printing at small and medium sizes, but looks terrible when blown up large. 8 MP is a far better resolution, and 16 MP provides definite but incremental gains over 8 MP. In fact, my formula suggest that going from 4 MP to 8 MP (-1/8 + 1/4 = 1/8 increase in practical resolution) is the same as going from 8 MP to infinity MP (-1/infty + 1/8 = 1/8 increase), and I argue that this is justified. Does it really matter that you can blow up some bug on a car to 300 pixels wide? Once a certain level of MP has been achieved (even when we don't take the diffraction limit into account), you just don't gain anything, even by doubling or tripling the resolution.

Of course, my view is most applicable to traditional applications of photography. Gigapan images clearly have their place, though I'd say they aren't quite the same as normal photos. When viewing a normal photo, you go "wow, great photo, and great sharpness too!" When viewing a gigapan, the first thing you do is zoom in really close and go "wow, look at all the fine details it captured!" But when you're not pixel-peeping at a gigapan (where doubling the pixels makes for double the "wow"), even if you're printing gallery-sized, my rule applies. Sure, 8 MP won't give you a great 20 x 30 print. But 4 MP? Complete crud. 16 MP will look quite nice by comparison, but the way the human value system works, it's more important to be not bad than to be good.

Complain
Post ()
Keyboard shortcuts: