Here is the latest letter written by Mark Dubovoy of the Luminous Landscape:-
An Open Letter To The Major Camera Manufacturers - Luminous Landscape
Although I am not any of the camera manufacturers, I think I have something to respond.. :-)
First of all, there is an error at the beginning of the article, I believe. I think the fact is that *nowadays* the indicated f-stop values do have already taken account of all the light loss inside a lens and from its optical elements. Why? It is because if it is not the case, correct metering and exposure is just impossible, especially when modern (zoom) lenses that have 20+ glass elements are not uncommon nowadays, which just mean more than 40+ surfaces and physically light path boundaries/barriers in total! Well, even with the most excellent glass and state-of-art coatings and technologies, each surface can still have 0.x% light loss and with 40+ surfaces it will accumulate to a significant amount!
Btw, the lens by itself can be and should be seen as a black box actually. I firmly believe that nowadays the f-stop values indicated and shown are already the *effective* numbers. In fact, there is nothing by any means which is physically related, i.e., nothing about the true lens "diameter" to focal length ratio, which is indeed non-existent for the complicated lens formulae and designs adopted nowadays. So, after all, the loss and non-ideal factors have all been considered and taken into account, i.e., the aggregated light loss within a lens should have been reflected with the f-stop numbers already - that's my point.
But then, the light loss at the sensor could be another real issue. But yet once again, depending what you look at and where you look from, this may not be an issue again. Um.. well, the ISO value of the sensor is indeed not always be true! Everything is software adjusted nowadays! In fact, I just wish to ask: what is the true ISO value? I think the answer is: we just don't know and will never know! As I have said above, if you are viewing the whole camera (i.e., body w/ sensor & processor + lens) as a black box, everything can be compensated inside. Actually, as similarly pointed out in LL's article, a "slower" lens could be used and then the ISO to be increased and vice versa, which will yield the "same" exposure result (first putting aside the DoF concern). After all, the exposure value could yet be kept the same with the change of multiple variable at the same time. Under this case, there is NO test can be carried out to find out the "true" ISO value, indeed! It is because since everything is relative and every test used to find out the "true ISO" by the methodology of comparison on exposure times is just meaningless, IMHO. In other words, since all those tests in finding out the "true" and "actual" ISO speeds are done on the basis of assuming one or more of the variables to be correct and "trustworthy" in the very beginning, e.g., the f-stop and value.
Or, to ask the question more specifically, just think in another way and answer my this question: Why can't the testers assuming that the rated ISOs are accurate first and then to measure the "actual" f-stop, on the other hand?? This might really be the case, although it is more unlikely. So, it boils down to a basic question of logics for thinking which could shake the bases of all those similar testing methodologies, which is really worth for some more thoughts.
Do keep in mind that no two digital camera models created on Earth are made equally, neither do different sensors/imagers, non the image processors, and then nor the software written and the algorithms programmed into the firmware. So, after all, everything varies! But, only the end results and outcomes count! And, the result is the only thing that matters!
All in all, If LL is asking for a response from the "major manufacturers" for the "issue" and proposing to use a new value to represent the light transmission and energy, I would think that there is just no issue and everything has just been right already, although the problem is tackled in various different ways, as it might be. For some more food for thought, one more example is that with today's' lens correction function for correcting the vignetting of lens on digital sensors, i.e., one of the practical digital problems as pointed out in the LL article, WHAT "actual" ISO value should be defined for an image/picture that has been corrected via software, with the brightness value of (the pixels at) the corners are raised but the centre are kept. So? Just think about it..
In short, I am opposing to the proposition of adopting the new "T-stop" in order to "help" the current situation, which I think will just create more chaos than what it has been now - which a different sensitivity value IMO is just a non-issue in real-life, as the "true" ISO value in use is never known. As I have been emphasizing, everything is relative and all the variables are now inside a black box, e.g., even the Tv should not be trusted as it can still have variations so the exact exposure times should also be re-measured, say, in addition to the actual "real" f-stop value! After all, only the final result produced by the blackbox does matter, but anything else!
P.S. I am still glad to see that my 5D is yet the king of being having the most efficient sensor in it with the minimal light loss, amongst of all the current and past Canon and Nikon DSLRs as listed! :-) So, sometimes old stuff with more primitive structure do still shine! ;->