William Robb wrote: > You don't need to be up on the technical details of digital cameras > to see the flaw in a test procedure that uses several different > lenses and conversion algorithms to compare sensor sharpness between > camera brands.
I'm not interested at all in comparing sensors. Does anybody shoot a sensor alone? Such a comparison could only make sense for a design team, prior to choose a new sensor for a new project. First, I'm interested in understanding the reason for a poorer than expected performance, and I'm convinced I got some explanation. Since there are so many people here continuing denying we have a sharpness/resolution problem, it is very difficult to go beyond this simple starting point. I'd be tempted to ask all those negating this, to kindly stay away for a while from this discussion, so that it can go further, but I understand that it won't be acceptable. Then, trying to understand which factor is responsible for a so so performance of a given system, I'm aware there are many risks and I'm far from considering a simple resize test (on an image obtained by a poor RAW converter) as a final proof of the best possible results one could obtain by the *ist D. However, it's at least a good evidence of two problems: 1 - Poor RAW conversion, making harsh unnatural edges. I'm still amazed when people here continue writing that the Pentax approach is the most film-like. Such a crap RAW conversion makes a mess of image rendition and makes it impossible to add the same level of unsharp mask other cameras/camera systems allow, because it will become too evident. As most should know, a proper unsharp mask is related to the print size. For getting good sharpness perception, the proper unsharp mask is the highest one which is not visible in the print. 2 - I agree the resize down/resize up process is rough, but can be a good indication of image data loss. If you loss no image data during this resizing, your original image contained lesser information than it should. Denying this is swearing black is white. > Nikon may take sharper pictures because of a better conversion > algorithm, or it may be a higher resolution lens, or it may be > something else altogether. So, do you agree that Nikon (and Canon, and Olympus, and Fuji...) take sharper and better detailed pictures? If so, we can go ahead with such a discussion, as most replies to my messages negate existence of such a problem. In my experience, I still have to see a picture taken with an *ist D (I tried 3 bodies and 15 or so lenses) allowing blow ups as large as other cameras (either P&S or DSLRs) apparently allow quite easy. Is it possible that all Pentax lenses (including DA, designed to that purpose) are dogs while most lenses by other manufacturers (with the possible exception of the 18-55mm EF-S) are excellent? Very very unlikely. So I believe the conversion algorithm and/or the anti-alias filter in front of the sensor to be the suspected factors. > If you want to find out who grows the nicest apple, don't throw a > banana into the test and say it's an awful apple. Yes, but... has the above statement anything to do with current discussion? I'm choosing an apple (an official Pentax image) among apples (aren't we discussing image quality here?) and I'm trying to understand something useful. Nothing less, nothing more. We can use other images too, to get confirmation/denial of my conclusions. This message is also intended as a reply to all others on the same subject, as I cannot spend so much time doing more here. My apologies to all other folks here, some of which had arguments worth considering. I never pretend to bear the truth. I only try to understand and when I find something interesting I like to share it with other folks that could be interested too. Now I have something else to do (sorry, I cannot spend all of my day on this list), but I have another idea under development. Wait and tremble :-) Dario Bonazza
