Hi,
I agree with Wayne that DxO measures only some aspects of image quality. In my view they essentially measure noise characteristics, of which DR is a part.
Wayne likes to print large, so other aspects of image quality like resolution, microcontrast (which I assume to be MTF near Nyquist) is important to the type of pictures he makes. Those parameters are not taken into account by DxO.
Best regards
Erik
Hi Erik,
The reason that resolution is not taken into account is because it's so dependent on lens quality, as you must already know. I think it is generally understood that all sensors with a higher pixel count have the potential to deliver higher resolution when used with a sufficiently sharp lens at an appropriate aperture. Do we really need someone to scientifically verify that fact?
Of course, it is true that a particular sensor which has a marginally greater pixel count than another, may not deliver the expected marginal increase in resolution because its AA filter is marginally stronger than the AA filter in the sensor with the slightly lower pixel count.
I suppose it would be possible for DXO to measure the strength of the AA filter in all its cameras tested, and give a ranking in terms of percentage of resolution in relation to the theoretical Nyquist limit of the sensor.
I suspect the reason they don't do this is because the rankings of such results would be so subjectively contentious. Consider all the lengthy discussions on this forum about the benefits and disadvantages of the lack of an AA filter. If DXOmark were to provide precise information on the strength of the AA filter in each camera it tested, how could it be determined whether or not a weak AA filter which causes more aliasing is more desirable than a stronger AA filter which more successfully prevents aliasing?
I guess they would have no choice but to exclude such results from their rankings, and such results would become just an extraneous piece of information, useful to some perhaps but not to many who wouldn't know if it were a good thing or a bad thing.
If one considers the current categories of sensor performance that DXOMark address, such as ISO Sensitivity, SNR, Dynamic Range, Tonal Range, Color Sensitivity, all across the full range of ISO settings, there is surely no doubt or confusion about which result is better, or which is more desirable.
I've never heard anyone complain that the Dynamic Range of their camera is too great, or the color too accurate, or the noise too low.
As I write this, I'm reminded of the great contoversy surrounding comparisons between the Canon 1DsII and the 5D which was released later and was the first affordable full-frame DSLR.
The pixel count of the 5D was slightly less (12.7 versus 16.2 for the 1Ds2) but in practice the resolution of compared images sometimes seemed to be so close as to be almost nonexistent. To explain this, it was suggested that the 5D had a slightly weaker AA filter than the 1Ds2.
Having been involved in that controversy myself, in the days when the Rob Galbraith forum was active, I find it interesting to now view the DXOMark comparison of the 5D and 1Ds2.
One can now understand the reasons for the controversy which took place before the DXOMark site existed. These cameras really were very close in performance.
What we find is that in all the parameters which DXO address, the 5D and 1Ds2 are neck and neck, except with regard to DR at high ISO. At ISO 1600 the DR of the 5D is a full 1/2 stop better than the much more expensive 1Ds2 at ISO 1600. A 1/2 stop is worthwhile, and I'm referring to comparisons of equal size images.
If we compare relative performance at the pixel level, the 5D is marginally ahead in all parameters, except its DR lead exceeds 1/2 a stop, as one would expect.
Sites such as DXOMark are tremendously useful because they help us dispel some of our subjective illusions about camera performance which we may have presumed from the price of the product and general advertising.
Those who had already bought the 1Ds2 before the 5D was released would naturally tend to be outraged at any suggestion that the much cheaper 5D was not only the equal of the 1Ds2 in most IQ departments but actually superior in at least one.
Without the benefit of the DXOMark results, an enormous amount of hot air was generated trying to justify the superior image quality from the 1Ds2, when in fact the real reasons for buying a 1Ds2 instead of a 5D would have been for practical issues not addressed by DXOMark, such as superior waterproofing, sturdier construction, greater number of predicted shutter actuations before maintenance, and autofocussing capability at F8 instead of the maximum F5.6 for the 5D, which has implications when using teleconverters.
The main attraction of the NEX-7 is not that it has better image quality and higher resolution than many other cameras on the market, but that it is significantly lighter and more compact than other cameras of similar resolution and IQ.