DXO mark (1 Viewer)

Here is something for you techno freaks.....And nice to see Nikon is ruling the roost at the moment.

http://www.luminous-landscape.com/essays/dxomark_sensor_for_benchmarking_cameras.shtml

I daren't read all that in case something important is pushed out of my brain while I try and understand it.

Ok perhaps later when I've woken up a bit. Interesting to see the level of thinking going into these things!

TobyH

It'd be surprising if Nikon didn't do well in DxOMark tests, given that their methodologies are inherently skewed in favour of Nikon's approach to sensor design.

It's a nonsense for DxO to claim that theirs is an objective test of IQ at the sensor level: no such thing exists, because IQ at the sensor involves a hell of a lot more than what DxO measures.

For example they take no notice of increased resolution in their tests - and that unequivocally has a positive effect on IQ.

And they don't account for Nikon's trick of clipping RAW data in the blacks, which automatically "chops off" noise data that would otherwise appear in the shadows, lowering the noise floor and resulting in better noise and DR results than would be the case if all the Nikon sensor's RAW data were presented to DxOMark.

So their DR tests are debatable too. The Canon 40D was - until relatively recently - widely accepted across the review test sites as having "best in class" DR: yet according to DxO, the Nikon D200 has better DR than the 40D, which - having owned both cameras - I can assure you is utterly preposterous. The D200 is simply not in the same league compared to the 40D where DR is concerned.

They also claim that some Nikon crop bodies and the Pentax K-5 have better resolution and DR than some of the best medium format backs/bodies available today, like the Hasselblad, Phase One and Pentax 645D units.

Incidentally, if you really place store in DxOMark, you'll find that the Pentax K-5 has overtaken the rest as Crop Camera King Of The Hill, where DR and noise are concerned, despite the fact that it has in essence the same sensor that the Nikon D7000 contains. Makes you wonder how an objective assessment of IQ at the sensor level can put one camera over another when they have the same sensor...

And about the K-5, it is known that this camera applies NR to its RAW data from 3200 ISO and above, which fact DxO completely ignores in its high ISO noise results - this will be true of the D7000's sensor too.

Is that fair to cameras like those from Canon which don't muck about with the RAW data?

Finally (although there's a lot more where this came from): according to DxO, the Nikon D90 is a better low light (high ISO) camera than the Canon 7D.

Utter, utter, BS of the first order - as anyone who has seen the results from these cameras will know all too well.

Still, the LuLa article you've linked to has some value - at least it states unequivocally that it's sensor size that matters most in IQ terms; that high pixel density on a crop camera need have no detrimental impact on IQ, especially at high ISOs, and that it can actually be an advantage; and that noise per pixel is no indicator of sensor performance.

There. I feel better for that!

Keith off the soap box and feeling better......

That's a great summary and fully respect those people that fully understand this stuff and you appear to be one of those.
Agree with you that D200 (still have one) was not as good as the 40D and having access to a D90 (Daughters) and seeing results from 7D I'd have to agree again (hard to swallow as a Nikon user) this isn't based on a bench test but results you see day in day out

My take on all this is the fact that mere mortals like myself struggle to maximize he equipment we have, I know that I can do better and I know that the weakest link is me and not what some test bench indicates

Nikon or Canon or Pentax etc they will all preform admirably given the right person behind the shutter release and equally they can all make us look stupid when we get it wrong
DSLRs are great we can delete at will and this means that in some respect we may accept failure more readily, nothing does it for me than a RAW file being pened up and only cropped and sharped with a little adjustment to he histogram.

Me I don't care what it states in a test, as long as it keeps me occupied and I can slowly improve I'm happy

Hi Steve,

my real problem with DxOMark is the fanboy fodder it provides: there's actually loads of really useful information in there, but people misinterpret and misquote the "headline" data in a way which completely ignores its context, just to score points. (Not saying Swainsons is doing this, of course! )

There's no doubt that the D7000 and Pentax K-5 are fantastic cameras (as they should be, being two-year-newer technology than say, the 7D), and they rightly deserve their current position at the top of the DxOMark "hit parade": but when the reasons why aren't fully reflected by DxOMark's tests because they ignore things like RAW NR and black offsetting (which sensors from other companies don't do) it can't be considered to be a properly fair, properly objective testing regime.

In a sense it doesn't matter how you get to the end result - the finished, high quality image - which is why DxOMark is ultimately pointless, because IQ isn't by any stretch of the imagination just (or even mainly) to do with what comes off the sensor - a great sensor is pretty useless if the camera can't focus worth a damn, or if its buffer is too small, or if you haven't got the FPS you need for fast BIFs.

Don't forget either that to a huge extent, sensor differences can be equalised into irrelevancy with good conversion and PP decisions.

That's why it rankles when DxOMark is wheeled out to "prove" that this camera or that company is ruling the roost.

In fact - as hinted at by the LuLA article - you can pretty much sum up everything useful that DxOMark tells us about digital cameras by saying "newer cameras are generally better than older ones".