December 14, 2013 2 Comments
How often do we pick up a top-of-the-line, state-of-the-art DSLR camera like the Nikon D800 or D800E only to be disappointed with the result?
I am not talking about the times when we look at the image scratching our heads while wondering what on Earth we were thinking when we pressed the shutter? No. That is the Purgatory of Composition, Light and Colour which many of us visit all too often. No. This time, I am talking pixel peeping!
It has happened to all of us to expect a razor sharp image chock full of fine detail with excellent micro-contrast and great colour rendering, especially when using a modern, high pixel count camera. Instead, we get so-and-so details, slightly fuzzy edges and maybe funny colours in high contrast transition areas. The camera was on a tripod that could hold the Keck Telescope steady and we used a wireless remote and mirror lockup, so camera shake cannot be the cause.
We then go online on photography forums and vent our anger by letting everyone know how overrated that $4,000 camera is and how that darned manufacturer blew it again!
In reality, the cause of many of these failures is much simpler: lenses, lenses, lenses. If you do not believe me, let’s have a closer look at how the Nikkor AF-S 85mm f/1.8 performs. This lens is by no means a dud. In fact, it is one of the sharpest, highest rated lenses for the Nikon F-Mount. I used DxOMark to compare how the lens performs in the lab on three different cameras: D7100 (24MP DX), D800 (36MP FX) and D600 (24MP FX). I picked these particular cameras for the following reasons:
- the 24MP D7100 DX crop sensor camera has the highest pixel density (smallest photo sites) and no OLPF (anti-aliasing filter);
- the 24MP D600 has the same number of pixels as the D7100, but is a full frame camera; the sensor has twice the area of the D7100 and the photo-sites are larger
- the 36MP D800 full frame (FX) camera has a higher pixel density compared to the D600, but lower than the D7100.
Let’s have a look at the DxOMark Score comparison below:
Two things stand out right away:
- Although the D7100 and D600 are both 24MP cameras, Sharpness for the D600 is 19MP, but only 15MP for the D7100; why? sensor size? pixel density? Read on.
- Although the D800 is a 36MP camera and both the D800 and D600 have full frame sensors, both have the same Sharpness reading of 19MP; why?
- The D800 is a 36MP camera; how come we only get 19MP (perceptual)??!! What gives??!!
Considering only the sensor, one would expect the D7100 sharpness to be much closer to the D600 than it is. After all, it is the SAME lens we are using. Instead, while the perceptual sharpness for the D600 is close to the sensor resolution, for the D7100 we get only 15MP perceptual resolution out of the 24MP sensor. Even more disappointing, the D7100 does not have an AA filter, which, in theory, should allow the camera to capture more detail.
Another expectation is that the sharpness measurement for the 36MP camera would be significantly higher than the one for the D600 with only 24MP and closer to its physical resolution. Instead, it performs well below the expectation and identical to the much lower resolution D600.
Clearly, we are missing something. Before we draw any conclusions, let’s interpret the accutance chart for the same camera/lens combinations:
Here is the big surprise: comparing the D600 with the D800 we can see that the behaviour is identical, but the D600 is out-resolving the D800 slightly. Say what??!!
When analyzing how the D7100 fares, we find that in the center of the frame the accutance for the D7100 comes in lower than that of the full frame D600. As we move towards the edge of the sensor, however, the D7100 beats the D800 and almost matches the D600. When used with the D7100 the AF-S 85mm f/1.8 shows no astigmatism and almost no drop in sharpness across the entire frame! Surprising??!!
OK, one more diagram before we draw our conclusions (last one, I promise ;~) ):
Whoa!!! Totally different picture!!
The same DxOMark Score comparison but this time using the Carl Zeiss Distagon Otus 1.4/55 ZF.2 prime lens.
- notice the difference in Sharpness: values for ALL cameras are VERY CLOSE to the physical MP count!
- notice how close the Sharpness values are for the D7100 and D600; it appears that the DX vs. FX and the pixel density controversy are red herrings … a lens matched with the camera can take advantage of the high physical resolution of the sensor;
- notice that the 36MP FX sensor performs better from a Sharpness perspective that the 24MP FX sensor, AS IT SHOULD!
So, now the obvious conclusion:
- The Carl Zeiss Optus 1.4/55 ZF.2 out-resolves ALL modern FX and DX cameras; we expected no less from a $4,000 prime lens designed specifically for the new breed of high resolution full frame cameras!
- Although the Nikkor AF-S 85mm f/1.8 is an excellent and modern lens and works great on the D600, it is out-resolved by today’s very high pixel density sensors (36MP FX and 24MP DX no AA)
- Everything else being equal, the higher the pixel density of the sensor, the better (sharper) lens you will need in order to take advantage of those extra pixels; as densities get higher you need better lenses; this is even more critical for smaller sensor cameras (read Nikon 1, m4/3, etc)
- Smaller sensor cameras use only the central part of the image circle produced by lenses for full frame cameras; since lens performance for most lenses (except some very, very expensive ones) drops as you move away from the centre, DX and CX cameras use only the best part of the image circle.
It is common knowledge that we need excellent lenses in order to fully utilize the high pixel counts of today’s cameras. What is not apparent, however, is just HOW GOOD those lenses need to be. As I mentioned at the beginning of the article, the AF-S 85mm f/1.8 is an excellent, modern lens. However, it takes a $4,000, two pound monster NORMAL PRIME lens to take FULL advantage of those pixels.
Does this mean you have to mortgage your house and hire a Sherpa porter to carry your gear around??!! Not really. When you post your dog’s portrait on Facebook, it will not matter whether you used the Otus or the Nikkor. Only you will know what the pixels look like. The key messages are, IMHO:
- don’t buy MegaPixels; the D800 will net you nothing in terms of IQ compared to the D600 unless you have the lenses to go with it (and shot discipline, and workflow, and computing power, and budget….)
- make sure the lenses match your camera performance; the Nikkor AF-S 85mm f/1.8 is a perfect match for the D600 and quite good on the D7100 too; you won’t get the huge bump in IQ you’d expect when moving to the D800, though; on the flip side, the Otus brings the best out of the D800 and D7100, but it will be almost wasted on a 12MP Nikon D3.
- make sure your expectations match your camera/lens combination; the fact that the 85mm f/1.8 cannot squeeze every pixel out of the D800 does not mean you cannot get spectacular photos that will satisfy the pixel peeper as well; it will do so for under $600, which is a lot less than what the Otus costs.
- don’t shy away from smaller sensor cameras: they have advantages that might just be the ones that matter most for you; in our comparison, although not exactly apples-to-apples (85mm vs 55mm), the 24MP DX D7100 with the Zeiss Otus has the same 21MP perceptual resolution as the 36MP FX D800 with the 85mm f/1.8 Nikkor.
As sensor resolution for FX and even DX cameras creeps into medium format (MF) territory, we are quickly hitting new limitations: lens quality, diffraction, workflow, cost, etc. It will take a new generation of lenses, like the Zeiss Otus, designed specifically for the high resolution cameras, to ensure we fully reap the benefits of those densely packed sensors.