7
\$\begingroup\$

I'm considering buying a new camera and my two candidates are Nikon D7000 and Nikon D7100. I heard that 24Mpix sensor needs very good lens to produce sharp images. Considering that I currently have this:

  • Nikkor AF-S 60mm f/2.8 G ED N
  • Nikkor AF-S 35mm 1.8G DX
  • Nikkor AF-S 70-300mm f/4.5-5.6 G IF-ED VR
  • Nikkor AF-S DX 16-85mm f/3.5-5.6G VR

would it be better to stay with D7000 and its 16Mpix or are these lenses good enough for the newer model?

Also in general how much the higher sensor resolution affects the image quality with ordinary lenses?

\$\endgroup\$
2

3 Answers 3

17
\$\begingroup\$

Ironically, I just did the math for this in another thread. Image quality is a convolution of all imaging system factors. The resolution of the lens or the resolution of the sensor are not independent factors...they are factors that convolve to produce the final "system resolution". If we use your two cameras as examples, we can calculate the resolutions of both, at both the ideal aperture and a common diffraction limited aperture (say f/11). No matter how you slice it, when you improve the resolution of a SINGLE component, the final system resolution improves. You will never reach the theoretical maximum of the best component, but resolution...and final IQ, WILL increase...at all apertures...when you improve any component, and increase the most when you improve the lowest common denominator.

To put that another way...diffraction can never cause a higher resolution sensor to perform worse than a lower resolution sensor. The notion that diffraction is an IQ killer is completely false, and I'll prove that fact with some math.

First, a few basic facts. The actual IQ of your images is the result of total system resolution. Total system resolution (lets use spatial resolution, measured in line pairs per millimeter) is determined by deriving the total system blur, which is calculated by taking the Root Mean Square (RMS) of the resolutions of all imaging system components [1]. For simplicity, we will only factor in the resolution of the lens (at a given aperture) and the resolution of the sensor.


To get the "blurs" of each sensor, the simplest way is to divide the width of the sensor by the number of columns. That would give us the pixel pitch. Since modern DSLR's use a Bayer design, which use an RGB color filter array (CFA), we can't get the maximum theoretical resolution the pixel pitch alone would allow. To accommodate for the sparse sampling, as well as the lower spatial resolution of red and blue pixels relative to green, I like to assume 68% resolution. To get the blur at 68%, just multiply the pixel pitch by 1.32 (yes, LARGER blur circle.) For the two cameras at hand, we have:

  • D7000: 23.6mm / 4928p = 0.0048mm/p (4.8µm pixel pitch); 4.8 * 1.32 = 6.3µm blur
  • D7100: 23.5mm / 6000p = 0.0039mm/p (3.9µm pixel pitch); 3.9 * 1.32 = 5.2µm blur

Lens resolution can be a very complicated beast. When aberration limited, computing the PSF (point spread function) has to factor in a wide variety of potential aberration types, at varying weights. That is very complex, so for the purposes of discussion, let's assume that we are working with diffraction-limited lenses. A diffraction limited lens is one that is producing the maximum resolution physically possible, limited only by the function of diffraction (the spreading of light as it passes through the aperture.) It is easier to make diffraction-limited lenses at telephoto lengths than at wide angles, so let's assume we have a diffraction-limited 70-300 when used at 300mm. Given that, we then have the following maximum resolutions from the lens:

  • f/5.6: 123lp/mm, or a 4.1µm blur
  • f/8: 86lp/mm, or a 5.8µm blur
  • f/11: 63lp/mm, or a 7.9µm blur

Total system blur is the RMS of the lens and sensor blur, which is computed as:

TSB = sqrt(lb^2µm + sb^2µm)

To convert blur back into system spatial resolution (SR) as lp/mm, you take the reciprocal of the total system blur and divide by two (to get line pairs rather than lines). The full formula to compute the whole thing at one is:

SR = 1l/(sqrt(lb^2µm + sb^2µm) / 1000µm/mm) / 2l/lp

If we plug the numbers we have for both cameras and all lenses into this formula, we get:

  • D7000 f/5.6: 66.5lp/mm
  • D7100 f/5.6: 75.5lp/mm
  • D7000 f/8: 58.4lp/mm
  • D7100 f/8: 64.2lp/mm
  • D7000 f/11: 49.5lp/mm
  • D7100 f/11: 52.9lp/mm

From the numbers above, you can clearly see that despite the fact that the D7100, with a higher resolution sensor and a lower diffraction-limited aperture than the D7000, it still performs better. AT ALL APERTURES! Diffraction can never produce worse results with better components. At f/11, both cameras would be thoroughly diffraction limited, and the D7100 can still produce better results. The math can be extrapolated further...to f/22 or even f/32. The improvement of the D7100 over the D7000 will diminish as you continue to use a smaller aperture, however at no point in time will the D7100 produce WORSE results than the D7000.

If you use all of the lenses you listed on both cameras, every single one of them will produce better IQ on the D7100 than on the D7000...assuming ideal circumstances. Now, ideal circumstances may not exist. If you have unsteady hands, causing a lot of camera shake, use a less than perfectly sturdy tripod, photograph on a sturdy tripod in the wind, etc. you will introduce additional factors that cause blur. Wind and shaky hands and all that cause blur regardless of the resolution, however if you have a higher resolution camera, the benefits of having that extra resolution will be diminished or eliminated due to the external factors that add blur. Assuming you do use a sturdy tripod, and do everything else you can to minimize these external factors that affect blur, then the theory clearly indicates the D7100 is a better camera...regardless of the lenses used! :)


It should be pointed out that at wide apertures, when lenses are aberration limited, the same general basis for image resolution remains true. Assuming that at f/1.8 on the 35mm lens produces a blur circle that is 15µm in size thanks to significant aberrations, that is much larger than the blur circle at even the entirely diffraction-limited aperture of f/11. However, the math works out the same either way. Plug 15 microns into the formula, and the D7100 will still produce better results than the D7000. Lower-quality lenses tend to be more aberration limited, and at more apertures, than their high-quality counterparts. A $500 lens will produce more blur at all apertures than a $1500 lens, which in turn will probably produce more blur than a $2500 lens. Only when you get into the $5000 to $15,000 range for lenses do they really start to approach "perfect", or near-100% diffraction-limited behavior at all apertures. The new Canon Mark II supertelephoto lenses, the 300mm & 400mm f/2.8 L II's and the 500mm & 600mm f/4 L II's, are very close to "perfect" lenses, even at their maximum apertures.

So, if a higher resolution camera performs better than a lower resolution camera regardless of the lens, why get a better lens? To maximize the potential of the whole setup. A better DSLR body will always do better...both with "crappy" lenses as well as a $12,999 EF 600mm f/4 L II IS. To maximize the potential of a better camera, get better lenses. To get the best IQ possible, get the best lenses possible.

\$\endgroup\$
4
  • \$\begingroup\$ Glad I was able to help. :) \$\endgroup\$
    – jrista
    Mar 12, 2013 at 15:50
  • \$\begingroup\$ @jrista Thanks for you nice answer. Could you please add some sources to your article? (Link to your previous answer, you are mentioning. Source that Total Blur is RMS of Component's blur.) Also it would help if you could spell out RMS and CFA once. And finally, that factor 1000 in the formula is not necessary and confusing -- maybe make it explicit that you use it to convert units? \$\endgroup\$
    – Unapiedra
    Mar 12, 2013 at 18:33
  • \$\begingroup\$ @Unapiedra: I've added units and clarification of acronyms. As for the RMS formula...that is just a SIMPLE APPROXIMATION, and is by no means highly accurate. To get an accurate measure of total system resolution, you would need to apply far more complex calculations to the PSF, point spread function for the lens, and a far more accurate sampling map to determine blur contribution from the sensor. One would also need to factor in the sensor filter stack (IR Cut, Low-pass), as well as the demosaicing algorithm. \$\endgroup\$
    – jrista
    Mar 12, 2013 at 18:47
  • \$\begingroup\$ Getting an accurate measure of system resolution is very complex. Using the Total Blur approximation (which I actually picked up from ctein) is a simple way to demonstrate in rough terms the benefits of a higher resolution sensor or a lens with a better MTF. \$\endgroup\$
    – jrista
    Mar 12, 2013 at 18:48
4
\$\begingroup\$

While true, the statement you make is misleading. For clarity, it should read as:

A higher-resolution sensor needs higher-quality lenses to delivers its maximum image quality.

What this means is that it is possible for a lens to be perfectly good on a 16 MP camera but produce an image that looks soft on a 24 MP one. However, if you resize the 24 MP image to 16 MP it will look at least as sharp as the image taken by the 16 MP camera.

In other words this difference is visible at 100% but when comparing the same print or display size, the extra demands of the camera on the lens are not visible.

NOTE: This assumes the same sensor-size, as is the case of the D7000 and D7100.

\$\endgroup\$
1
  • 5
    \$\begingroup\$ Also said differently: a higher resolution camera does not make a lens worse than a low resolution camera. \$\endgroup\$ Mar 11, 2013 at 20:40
2
\$\begingroup\$

You'll be no worse off with 24MP sensor. If your lenses can resolve as much as the 16MP sensor, but not as much as the 24MP sensor, then you'll essentially have "wasted" pixels - i.e. resolution power that can't really be used. But you'll only see the difference if you pixel peep, and you won't get a worse image from the higher resolution sensor, you just won't be able to take advantage of the higher resolving power.

As for the lenses you mention, I have seen excellent results with the 70-300 VR on a D800 of 36MP, even at 300mm, so I believe that lens, and surely the razor sharp 60mm should be just fine with 24MP.

The other two lenses are also pretty sharp lenses in the centre. The corners may be a bit soft at wide apertures, especially the 16-85 at 16mm. I would imagine that you will get some benefit from the 24MP sensor even with those two lenses, at least in the centre, if you use at the optimum apertures and nail the focus.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.