Camera Upgrade Resolution Expectations
What can you expect from a lens when you change to a higher resolution camera? For crummy lenses, don’t expect improvement. Good lenses, though, are another story.
Let’s compare the Nikon D7000 (16 megapixel) against the D7100 (24 megapixel) cameras. The D7000 sensor is 3264 X 4928 pixels, while the D7100 is 4000 X 6000 pixels. Both sensors are the same size, at 15.6 mm X 23.6 mm.
Resolution change is evaluated in a linear fashion, so it’s the change in the ratio of either the short or long edge measurement of the sensor, such as 6000/4928 = 1.22, or 22 percent more with the D7100. Doesn’t sound nearly as sexy as 24MP/16MP = 1.5 or 50% bump, does it?
If a lens has good enough optics, then you would expect your MTF50 measurements would therefore increase by about 22% by upgrading from the D7000 to the D7100. But wait, there’s more! What about that removal of the OLPF (optical low-pass filter) on the D7100? At the risk of moiré in some pictures (never actually observed by me in any of my pictures yet except for shooting a “Siemen’s star” target) you can squeeze a bit more resolution out of a sensor when you stop fuzzing-out the image by getting rid of that low-pass filter.
Low-pass (e.g. don’t pass high-frequency light/dark transitions) filters stop those nasty color patterns you used to see watching TV when some guy would wear a pin-stripe suit. If the pinstripes were spaced at the same interval as the camera sensor elements, it really messes up the image. The OLPF filters work by de-focusing the image a little bit, and therefore ruins those fine details in your carefully-focused picture you paid big bucks to get. You get the lost resolution mostly recovered by using the “un-sharp mask” in your favorite image-processing program. It always seemed a real hoot that “un-sharp” was used to get “sharp”.
Another way to get rid of moiré is to use the "moire removal" feature in many modern image-processing programs designed specifically for that task. In any event, moiré is a “don’t care” for most people and most photographs.
After that major digression, let’s get back to lens resolution and camera sensors. We’re expecting about 22% improvement in MTF50 measurements, not counting the OLPF filter removal. This assumes the lens resolution isn’t already maxed-out on the D7000.
MTF50 results for Sigma 150-600mm lens on D7000.
MTF50 results for Sigma 150-600mm lens on D7100.
Using the above Sigma lens results, we get a MTF50 high-reading change from 40 to 48, or about 48/40= 1.2 = 20%. On the low end, the change is from 26 to 30, or 30/26=1.15 = 15%.
Next, let’s check out my best-resolving lens:
MTF50 results for Nikkor 85mm lens on D7000.
MTF50 results for Nikkor 85mm lens on D7100.
Using the above Nikkor 85mm f/1.4 AF-S lens results at f/4.0, we get an MTF50 high-reading change from 54 to 68, or 68/54 = 1.26 or 26%. On the low end, the change is from 46 to 52, or 52/46 = 1.13 or 13%.
If you have a pretty good lens, you can expect to get nearly the theoretical maximum resolution change. It’s unclear (no pun intended) the contribution of the OLPF filter removal to overall resolution, but I’m getting around 26% overall (camera sensor plus OLPF removal). The OLPF removal must be at least 4% improvement.
Corners aren’t quite as rosy of a scenario, but still a substantial improvement.
BTW, don’t abandon the un-sharp mask when post-processing just because you don’t have the OLPF filter. You’ll still see a huge bump in sharpness.