Your comment about the a900 and reflections.. I noticed this too. Some lenses seem to cut through reflections to allow light behing to come through. I found this with my Contax G1 and its Zeiss lenses, and the a900 and the 24-70mm Zeiss.

I think its very hard to decide what to do in terms of an upgrade. I did not find the new Nikon N 24-70mm f2.8 to have lower distortion than the 17-55 f2.8. But the Nikon 24-70mm f2.8 L seems to have less barrel distortion than its Nikon comptetitor.

My major concern is the 5D MKii’s AF performance. Does anyone have some objective comments?

Karl

In this article the viola was moved, the cameras were not all there at the same time and most important there were focus differences as you mentioned. Still you did a great job.

I have long been fighting with front and back focus issues – even back to my SRT 102. I had issues with my a700 and sent it back and it is now prefect. I love shooting at F1.4 or 2.0 and never realized how amazingly sharp the 50m F1.4 is until I got this camera back from Sony. But with the machinations of going thru all that, I came to realize that the indicated spot central realestate as inscribed on the finder does not match the actual area in which it works. The actual area is wider. Secondly and more important is that whatever size of the focus sensor – if there are solid objects in front of the desired object the focus mechanism will choose the closest item. The sensors focus mechanism views a solid plane in space which is parallel to the camera. In this case the strings are in front of the end of the fingerboard. And since the strings are not in the plane of the focus area, the closer end of the string would be chosen. The manufacturers program the sensor to set focus to the closest item when there are multiple items in the sensors view.

Canon claims its spot meter is 3.5% of the frame. If you do the mathematics a 3.5% area in the center of a rectancle takes up almost 20% of its horizontal length and height (.2 x .2 = .04 or 4% of the surface area). That is way too large to be a spot meter and a serious defect in the Canon. I suppose they do that so that the mechanism can work to a lower light level. I believe the spot focus area in the Sony is the smallestof the three and that is why it focuses more accurately and consistently than the other cameras. However, I cannot find any claim for the area from either Nikon or Sony.
Please check my response to your “Sony Alpha 900 and Nikon D3X raw file Noise comparison” on this site.

The alternative solution would be to have shot using the flash power to bracket the exposure, it can be incremented in 1/10th stop steps globally from the Skyport controller, but in this case only increased not decreased. I did make tests at other apertures between f22 and f8 and determined that f11 gave the best balance between sharpness, depth of field, and diffraction losses.

The only negative bias against the Sony present in this test is the use of IDC2 to convert a raw file which had 0 sharpness set in-camera. IDC2 translated this as ‘normal’ sharpneing and applies a fairly coarse USM-type sharpening, which shows in the lines down the instrument strings and the black haloes round the grains of rosin. Out of choice I would have used ACR 5.2 to convert the Sony file, because the result is far more detailed at pixel level, but this would be unfair to the Nikon as ACR 5.2 doesn’t handle DX3 files yet.

Everything is a compromise when absolutely identical software and optics can not be obtained.

The Canon test can not be repeated because the Canon has gone back home. However, Nikon is shipping me a 105mm VR Nikkor macro today, and it should be easy enough to set up the viola again and replicate the flash powers (2.3 backlight, 4.7 sidelight) and camera position. Sigma is sending me two 28-70mm f2.8 lenses to compare the A900 and D3X with supposedly identical glass.

What’s interesting about reactions to this test elsewhere is that each brand’s users consider it unfair to their own brand – which says more about brand loyalty than about the results!

David

You are using old film optimised lenses on the Alpha (for understood reasons of practicality).

Exposure was calibrated off the Canon, over exposure in the case of both the other 2 put them at a significant disadvantage.

As a result of the above, this test extremely biased against the A900 (and to some extent the Nikon); I would like to say that I’m surprised to find 3 such errors in an article this short, but I’m not. It’s widely known that there is a great deal of prejudice in the media against Sony in the DSLR field (often due to brand loyalty/bigotry and/or volatile advertising revenue streams from the big 2: in the broadest sense corruption). I would have thought that a site named PhotoclubAlpha would be above this and would be fair, I can only assume/hope in this case it’s because of sloppy reporting.

Will you show us some examples to high ISO??

Good review…