Nikon D7000 Dxo scores

So Emil Martinec writes a paper on why Canon is better and imagine my surprise -- he shoots a Canon!

 

http://theory.uchicago.edu/~ejm/pix/20d/

 

And just like DXO, he has lots of charts and figures, but doesn't give you his raw data. Plus, he references five different sources where he has culled data and combined with his to get his results which of course show the Canon (in most cases) to be a more efficient sensor:

 

http://www.clarkvision.com/imagedetail/dig...mance.summary/;

http://astrosurf.com/buil/us/test/test.htm

http://home.comcast.net/~NikonD70/Investig...acteristics.htm

http://www.brisk.org.uk/photog/d3summary.html

http://www.openphotographyforums.com/forum...read.php?t=4784

 

And of course each one uses a different source of data, different cameras and different lenses. You CANNOT combine results from 6 different data sources and come to a valid conclusion. This is what gets scientists into trouble. My wife is a research scientist and any results used from another study which are not duplicated in an independent test cannot be combined with other data.

 

So tell me, how does this "prove" DXO is biased towards Nikon other than one guy seems to think so? Yes, Emil is a professor in the physics department at the University of Chicago, but I fail to see how this makes him an expert in the mechanics of photography.

What Emil says is realy a simple thing: Nikon quantize the low light levels differently then Canon.

Let me explain how this effects the DXO scores. First. Lets understand what this means. The sensor output is analog voltage that corresponds to the light it observes, and you have to quantize it to get the RAW numbers. Now, lets look at the dark level where ther is no light. the sensor voltage output corresponds to the thermal noise

and has a gaussian noise characreristics. you can quantize the levels such that the numbers will give you the gaussian shape (like canon does) and therefore the dark level is not the zero but a higher number, or you can quantize them suce that the dark level correspond to the zero number, like nikon does. This causes the noise level to be half the true level, therefor increase artifficialy the DXO score for the dynamic range for example, but you loose the symetric shape of the noise, so you can't efficiently use noise reduction algorithms.

That Emil Martinec is right you can see from the DXO graphs of the dynamic range. You can clearly sea that while canon graphs bends twards the low iso levels, Nikon graph stays streight.

And of course each one uses a different source of data, different cameras and different lenses. You CANNOT combine results from 6 different data sources and come to a valid conclusion. This is what gets scientists into trouble. My wife is a research scientist and any results used from another study which are not duplicated in an independent test cannot be combined with other data.

This is so true! Many scientists do not realize this. There is a growing trend to make all data available on the web for anyone to use. One can only imagine what kind of results this will eventually generate as "apples and oranges" data are selectively used.

The key is to generate data systematically. This is the advantage of DXO (assuming they are consistent in their methods) even if some of the results may be flawed. Of the cameras I have used I find their "Sports" (high ISO) ratings consistent with my experience. The dynamic range (DR) is more challenging. For example, I think there is more "headroom" in my 1Ds2 files compared to my D2X files. Is this reflected in their DR ratings? It is interesting to see that newer cameras are much better with respect to DR. Thus I am looking forward to seeing the results of the next generation of pro bodies.

If you reffering to my remark, then I can just prove it. All you need to do is read this paper by Emil Martinec:

http://theory.uchicago.edu/~ejm/pix/20d/po...D300_40D_tests/

And you'l see the reson why DXO scores of Nikon are higher.

It would be neat to see this paper updated. Last update was March 2008 so is a bit dated. It would be interesting to see where the new cameras rank in Table 9 and to compare it with the sports ranking done by DXO.

Sure would be nice to have a housing for the pentax system.

I second that... Pentax are going after the Outdoor Photographer market.. A quality Underwater housing will fit that image nicely. The Pentax DA 14mm f2.8 with a min focus of 17cm would be a sweet lens to take underwater...