Incense

by Bart Arondson

submit your photo


Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

I have seen sites that use measure DSLR performances with Image QUality (in numbers). So I was thinking, given a D800 and a D7000, with the same lens, the same lighting condition and many other external factors, will it always be the case that the D800 will produce better images than the D7000? Or maybe even given a DX lens mounted on a D7000 with the FX lens equivalent mounted on the D800.

SIDE QUESTION: Will it always be the case that FX sensor cameras will always produce better images than DX sensors ones?

share|improve this question

2 Answers 2

up vote 6 down vote accepted

"Better" must be rigorously defined.
And then it can be rigorously argued :-)

For camera comparisons it is usual to make lens variations and effects as small as possible so that you are actually comparing the cameras. There will be some cases where a lens performs so terribly on an FX sensor and so well on a DX sensor that the D7000 will win BUT that is not the sort of lens that you will usually see in a test. Also, it is not usually the sort of lens a D800 user would probably care about when attempting to take good photos. And it is notionally possible that the best equivalent DX lens to a FF lens gives a better DX result - but this would usually be rare, as the FF cameras are the flagship models on which camera makers aim to produce superior camera + lens results.

One means by which you might "win" with the DX camera, which meets you criteria, is to use the same lens in each case under ideal conditions and to adjust subject distance etc to produce visibly identical images. In this case the DX camera is using the centre portion of the lens while the DX camera uses the whole lens. The effect on the actual image of factors such as dynamic range and given ISO signal to noise ratio are minimised by being able to use optimum lighting levels. Under such conditions the improved quality usually achieved by using the lens centre may well offset the DX sensors overall performance. [Example: I have an APSC Sony A77 with a 24 megapixel sensor and a DX Nikon D700 with a 12 megapixel sensor. In studio conditions with "good" lenses, for practical purposes the Sony would usually win. Take them to the circus and photograph a trapeze artist at work and the D700 would be utterly superior.]

Three measures of "better" which can be numerically expressed are

  • The ISO setting at which a camera can achieve a specified signal to noise ratio

  • The dynamic range achievable (in bits)

  • Colour depth (also in bits)

each under specified conditions.

Figures for most DSLRs and large format cameras are available from
the "DxOMark Sensor" sensor comparison site..

The above link takes you to the ISO evaluation table - which is where my book mark is set, as high-ISO noise performance is what I most care about when looking at new results, but the other tabs cover color-depth, dynamic range and overall result.

The D7000 has an 1167 ISO rating and the D800 has a 2853 rating. The difference of 2853/1167 = 2.44, which is 1.27 stops - making the D7000 an extremely good APSC performer.
[Stops = log_base2(ISO ratio) = log_base10(ISO ratio) / log_base10(2) ]

I personally find these evaluations suspect as the reality seems to be slightly different in most cases than the table suggests. They compare sensors with different numbers of pixels by scaling the sensor noise readings DOWN by a factor of square root(megapixels high/ 12). ie they use 12 megapixels as their standard sensor size and reduce the noise measured with large megapixel sensors to scale the results. This is equivalent to downsizing the large images with an algorithm that improves the signal to noise ratio by sqrt(mp/12).

My suspicion - based on both overall image appearance and on extensive pixel peeping - is that this is too generous to noisy high mp count sensors, i.e. when you scale your 36 mP D800 image down to their reference 12 mp size I suspect that it does not get sqrt(36/12) = 1.7 times less noisy as they deem.

So if I am correct (which is moot) the D7000 may be closer noise wise to the D800 than this chart suggests.

DxO weight their 3 scores to provide an overall rating. Some "overall" results are

  • D800 = 95
  • D4 = 89
  • D3s = 82 !!
  • D3200 = 81 !!!!!!!!!!!!!!!!!
  • 5D MkIII = 81
  • D7000 = 80
  • D700 = 80

I recently bought a D700.
I would not have considered buying a D3200.
That may say more about me than about the cameras :-).

I rather feel DxO's weightings may need looking at, but their overall rating is at least one available measure of "better" on which discussion can be based.
On this basis the D800 stands very clearly ahead of the pack.
But, the D7000 is still a very good camera.

share|improve this answer
1  
Just to be objective, one really needs to take DXO scores with an appropriate grain of salt. Their process may be algorithmic and consistent, but many do not thing its accurate. For those wondering if Canon technology is really quite as far behind Sony/Nikon technology, I highly recommend reading Roger Clark's site for some Ph.D. level knowledge from a guy who knows sensor tech like the back of his hand, and is renown for its use in some major scientific discoveries. –  jrista Sep 13 '12 at 18:52
    
@jrista - I'd not gained the impression that Canon was much if at all from DxO scores plus other available material. The DxO ISO score shows D800 = 2853 & 5DMKIII = 2293. The difference is log2(2853/2293) = 0.32 stops. {FWIW - about what the Pellicle mirror will take off the D800 sensor in the newly announced 36 mp Sony A99.} My suspicions re DxO scaling for mp difference plus comments from users who own or use both D800 and 5D MkIII suggest they are close to equal and that other aspects would make a 5DMKIII superior by my measure. –  Russell McMahon Sep 13 '12 at 20:59
    
@jrista - Roger Clarks's site is exceedingly impressive. Thanks for the reference. –  Russell McMahon Sep 13 '12 at 22:47
    
I think the only edge the D800 has is in DR at ISO 100...specifically. The lower read noise in the Exmor sensor would account for approximately two stops worth of improved DR. But its ONLY available at very low ISO. Even the higher Q.E. of the D800 doesn't fully compensate for its smaller pixel size, which is why the 5D III edges it out at higher ISO...but even then, physics simply dominates at and above ISO 400, so no camera can really "win" there. –  jrista Sep 13 '12 at 22:54
1  
Another point to note about DXO...they utilize "bonus points" in some circumstances. In an objective measurement comparing sensors, there really shouldn't be any such thing as bonus points...even if they are equally rewarded across all brands and sensors. The use of bonus points introduces an explicit factor of bias into their system. The only reason the D800 achieves a score of 95 is they just barely succeeded in achieving two levels necessary in some of the tests to earn those bonus points. The extra points mean ZIP, ZERO, ZILCH about the hardware, its pure subjectivity. –  jrista Sep 14 '12 at 3:00

How do you evaluate a "better" image? One example: all other things being equal, a crop sensor will give you a greater depth of field than a full-frame sensor, which means more things will be in sharper focus - in some circumstances, this may be considered a "better" image than one with less sharp focus. In other circumstances, you may want a shallower depth of field in which case the full-frame sensor will give a "better" image.

In most circumstances, you can probably stop down the lens on the full-frame camera to increase the depth of field and then increase the ISO to compensate, while maintaining the same signal to noise ratio in the image due to the better high ISO performance of the full frame sensor, but that requires some knowledge from the camera operator (or maybe just the camera itself).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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