Rick, I used both K10D and K20D for over a year. one thing that really peeved me is that the K20 is marketed as an evolutionary upgrade to K10. That's very, very far from the truth. Especially when it comes to dealing with the images in post processing. The K10D is a CCD based camera, with an oversized A/D converter to sift out as much dynamic range as possible. The noise characteristic of this camera is very "grain-like" and reminiscent of film. Much in contrast with the CMOS-based K20D with its more reticular, gritty looking noise characteristics.
I came to appreciate the K20D anyway, though, because its noise is easier to reduce in post processing than is K10D noise. I use NoiseNinja as plugin to Photoshop, btw, and for the record I think the DxO tests are measurbating BS. :-) The K-7 has essentially the same sensor as has the K20D. However the image quality is improved over K20D. Not much, but the devil is as usual in the details. Literally. :-) On my trip to Antarctica, I got the opportunity to compare my own files to raw files from various Canon cameras, and what struck me most was the amount of detail preserved in the K-7 raw files. The noise reduction with Canon is a lot more brutal, clipping quite a lot of detail in the shadows. It's a matter of taste whether the Canon look is "plastic" or "clean", but I know for myself that I much prefer to have the detail recorded, and tweak the noise vs. detail ratio myself. Besides, the post-processing noise reduction software has a lot more data power to draw from than the in-camera processing, and I suspect the algorithms are more sophisticated as well. In comparison with K20D, I believe the K-7 is better at discerning between shadow detail and noise. No idea if the lab tests take this into account, but I find the K-7 raw files (I shoot DNG, btw) very pleasing to work with. I also think the K-7 is an improvement in exposure accuracy over K20D. I tend to get histograms leaning more to the right with K-7. Where I would lift the shadows in a K20D shot, I end up reducing the highlights with K-7. Yet I rarely get burned highlights, so the overall effect is one of less noise all together. So in short, don't get hung up on those tests. The K-7 is one heck of a performer, and the best tool out there for your K-mount glass. You'll need a few hundred shots to get used to the new sensor, but trust me that you won't look back afterwards. It's a significant step up from the K10D's strategy of masking noise as "grain". Jostein 2010/1/3 Rick Womer <rwomer1...@yahoo.com>: > Right now, I am not planning to replace my K10D with a K7, because the latter > does not seem to have significantly better low-light performance. By > reputation, the K20 is about 1 1/2 stops better than the K10 in low light, > and the K7 is about a half-stop worse than the K20 (thus a stop better than > the K10). > > That seemed simple, until I began looking into low-light performance test > results. > > The noise testing on dpreview used only jpgs until their review of the K7, > which compared RAW noise in the K7 and K20. For chroma noise, a score of 10 > (y-axis) corresponds to an ISO of about 2800 for the K20, and about 2000 for > the K7; this agrees with others' observations that the K20's low-light > performance is about a half-stop better. > > When one goes to dxomark.com, things get confusing. Even though the > low-light performance of the K20 is reputed to be about 1 1/2 stops better > than the K10, the low light ISOs are 639 and 522, which is only about a 1/3 > stop difference; and the overall scores (blending low-light ISO, dynamic > range, and color depth) are almost identical. The K7 sensor has a lower > overall score than the K10 by 5 points, having 1 stop less dynamic range and > almost the same low light ISO. > > Which leads me to think that the K10 and K7 sensors' low light performance is > almost the same, and only slightly worse thana the K20's. So where does the > K20's reputedly better performance come from? Firmware? > > Comments or explanations? > > Cheers, > > Rick > > > P.S. I am very intrigued by the dpreview results on the Kx sensor, which (in > RAW) has a chroma noise score of 10 at an ISO of 6400. No dxomark testing > yet. If a similar sensor found its way into a K7-like body, they'd have a > sale. > > > http://photo.net/photos/RickW > > > > > > -- > PDML Pentax-Discuss Mail List > PDML@pdml.net > http://pdml.net/mailman/listinfo/pdml_pdml.net > to UNSUBSCRIBE from the PDML, please visit the link directly above and follow > the directions. > -- http://www.alunfoto.no/galleri/ http://alunfoto.blogspot.com -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
