Godfrey DiGiorgi wrote: > Dario said: > >> .. Furthermore, I think that many prospect users are not so >> interested in a >> sensor test. They are interested in camera performance. For those >> wanting >> to shoot RAW exclusively - and provided that DPReview or someone >> else would >> ony test RAW performance - you could look at the Nikon D80 or Sony >> Alpha 100 >> test for knowing what to expect from the K10D. > > Not true. Each manufacturer adds their own twist to the sensor > configuration and supporting electronics. Like the K10D's weak > antialiasing filter and 22bit ADC. The results, while roughly > comparable, aren't identical even in RAW capture.
Of course, I completely agree. And we can go even further on that hairsplitting. That's not the point. I wanted to highlight that disassembling a camera is not a good way to see how it performs. In order to test camera performance (or at least for giving a useful idea of its performance), you have to test it as a whole, including camera software and lens. Further on that, my point is that there isn't such a thing as a useful camera test, as anyone has his/her own questions, which also vary with time and situation. So you don't have to blame someone for not doing the test you want that day. Just try an see how much of that test can be useful and what cannot fit. This said, I find Phil's tests to be the closest thing to a useful test I've ever seen. A possible improvement could be testing each camera with the standard lens it comes equipped with AND with the best possible standard lesn for that bayonet. >> Even worse: which converter of choice for testing RAW files? Which >> release? >> Isn't that changing on an almost daily basis? How to get comparable >> results? >> Then, isn't it possible that a given converter or a given >> conversion (or a >> given tester!) would be biased toward a camera, or sensor, or >> whatever? I'm >> truly afraid that your suggested test procedure will result being >> far more >> debatable than testing a camera JPEG performance, which at least >> can be >> rather associated to a given camera. > > I see nothing wrong with testing using JPEG and/or RAW output data. Me neither. > Or with the "standard" lens provided by each manufacturer, although > that starts to be more of a system test than a test of how a > particular body works. OK, I've suggested doing both: consumer-grade zoom (the one the camera is usually sold with) and the best gem in that same system. So we could undestand if say a given manufacturer works better than its competitors on in-camera software and then caps the body with a bottle bottom. > Present the test data and the condition and > settings of the tests. Make it clear as to what is your opinion and > what in the review the test data consists of, and the set up for that > testing. Fine. > Don't test the camera on its default settings and then judge its > performance without taking into account what those default settings > are intended by the manufacturer to do, which is the usual point of > departure from reality. Departing from defaults will start an endless buch of tests to be done, to be explained and to be understood, far out of reach for any tester and 99%+ of readers, I'm afraid. > Testing and interpreting the results is a very complex game to > achieve useful, comprehensive and objective information with > something as complex as a modern DSLR camera. I couldn't agree more. Dario -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net
