RTPVid wrote in post #13927580
Since you brought up DR, the DR straight off the sensor is only of interest to the camera designer, not the photographer. The photographer wants to know the DR of the resulting image. This is true of most of their measurements. Unlike with film, we do not have access to the signals coming off the sensor. We only have access to the images written to the memory card.
Since the signals from the sensor are ALWAYS processed by every camera before they are written to the card (even so-called RAW is not REALLY RAW, as in unprocessed), and since the different camera manufacturers make different choices for each of their camera models wrt how the sensor data is processed and under what conditions, the measurements of the sensor are, not to be repetitious, irrelevant.
It doesn't take too much investigation by someone who has experience with more than one camera to identify DxOMark data that seems even contradictory to real results.
Well, actually what we want them to analyse is the raw data that comes from the sensor in a particular camera - that is what we want to use to evaluate which body to buy/use, and an objective set of measurements on that data is precisely what will be useful to us. It's semantics really - what we have a choice over is not the sensor alone, but rather the data written to the card from a sensor/processor combo. We are highly unlikely to be able to use the DR information coming from the sensor - unless we are in the business of developing processors of our own to combine with the raw sensor output
For film, we have to develop the exposed film to be able to analyse the image, so the developing conditions are akin to raw conversion (different developers can produce different images) and printing would be the equivalent of post processing. To your point, DxO overtly addresses the issue of what, if any, "processing" is done before the data are written.
As we do not have access to intermediate outputs on the sensor, DxOMark measures RAW images — the very same images that can be accessed by photographers who use cameras that shoot in RAW.
From DxO's testing descriptions
Certain manufacturers embed a small part of the processing directly in the sensor, which means that some degree of processing occurs before the RAW image is sent to the RAW converter. In this case, measurements for these "pre-cooked" RAW images can be biased by this processing.
To avoid any potential impact on our measurements, DxOMark always tests all cameras to detect any pre-processing of RAW images. A processed or pre-cooked RAW image has different characteristics from a genuinely unprocessed image. To some extent, these characteristics enable us to walk back the processing and reconstruct the original image to perform unbiased measurements, and we always inform the user about models with embedded pre-processing.
Even though they add the disclaimer "to some extent", the semantics still apply - we get what we get from the sensor/processor combo into the "raw" file and it is upon those data that we would base a purchase decision if it were to be made based on objective testing. I disagree with some of their decisions on "normalization" of data they present (especially normalizing the data supposedly representing a lens's properties by normalizing to FF even if the lens was tested on a crop sensor) The converter and postprocessing choices are quite individual, as are the "metrics" someone might use to do their own (eyeball) comparisons of different camera models. In particular, the weightings of different parameters (some of which can be measured objectively) by different photographers and their tastes, preferences etc will likely vary quite a bit. The concept of rolling all the numbers for color depth, DR and high ISO noise all into one number is particularly unhelpful. Should anyone rely on DxO alone? That would be very unwise indeed, but that doesn't translate into "DxO tests are irrelevant for all photographers"!
BTW, I wasn't meaning "you" in particular in the technical comment, indeed I would expect someone with your background to be able to assimilate data from a range of inputs and combine them to provide the "best informed" view of the situation.