18 December 2008

DXO Mark

Anyone seen the new DXO Mark sensor ranking tool? The one that 'measures' the sensor quality before any in camera processing is applied?

Am I the only one to find this a COMPLETE waste of time? What use is the RAW camera sensor data to anyone? Will it affect your buying decision? If it will, you need to stop and think first. While it's nice to know how good the sensor is according to a third party, what can you do with the information? Nothing. Nothing at all. Why? 'Coz the in-camera processing is incredibly important!

If you had the best sensor in the world, poor in-camera processing will ruin the images. Sure having a good sensor helps, but it's the combination of sensor and processor that makes the image. Looking at either of them in isolation is just a waste of time.

No, I'm not going to get on my soap box about the relative merits of the EXPEED vs DIGIC processor. Suffice to say, comparing sensors is interesting but not the whole story. Try looking at the whole picture.

Go on, comment with the words "disgruntled Canon fanboy"! I dare you :-)



jordax said...

Disgruntled Canon fanboy! ;)

But in all seriousness, thank you for drawing this interesting site to my attention. I disagree with your comments though - I think DxO Mark does have a place and some value.

I assume DxO aren't ripping the cameras apart to do the tests and are, in fact, basing their tests on the raw files generated by the camera. As such, these resemble the raw files that many photographers will use when shooting. The importance of the image processor is relevant if you're shooting JPEG, but if I'm shooting raw, it doesn't matter so much as the processing I apply after the event. For that matter, even if the processor does have some input to the raw file, DxO Mark is still relevant assuming their raw files have passed through the same processing that mine would do.

So the question is, is it relevant to analyse the quality of the raw image data? Sure it is! To make up an example, it's all very well to say "The 50D offers great high-ISO images if you apply the noise reduction", but if a similar camera can generate images that don't need noise reduction, surely that's a win? Also, if I'm shooting raw I will processs the images on the PC afterwards. Today I'm using Canon's DPP, but tomorrow I might use Photoshop. If the image only looks good if processed using Canon's algorithms, I think that makes the camera output of lesser value than an image that's better to begin with and which will look better through any processing workflow.

I shoot raw for the high-bit-depth output and increased latitude in exposures after the event. I have an interest in putting low-noise images into that editing process.

Now, is raw file quality the whole story? No. There are many other factors in choosing a camera, including general feel, handling, focus speed, shooting speed, ... Any purchasing decision will be a combination of these and a compromise where need be. Was my 30D the best image quality when I bought it? I don't know, but it was certainly "good enough" and felt better in the hand than anything else I could afford.

So anyway - I doubt DxO Mark will decide my next purchase entirely, but it certainly is worthy of some consideration.

And as a side note, I was interested to note how similar the 50D scores to the 40D, especially when the 40D scores better than my 30D.

Martin said...

Disgruntled Canon fanboy! ;)

MMMM - DxO Mark

Well - I am sure it is of some use to some geeks who make it their business to do so

Me - I'd sooner be taking photos and enjoying the capabilities of the fine Canon cameras that I have purchased.

A couple of tiny percentage points here and there are not relevant to me in the real world when there are so many other user induced variable.

I see more differences as a result of the RAW converter I use.

Good post :-)