Sony A700 Dynamic Range test: 9.5 f-stops

[Ray]

I get a sense here, Jonathan, that you are overcomplicating a situation which is really only useful to the practical photographer as a guideline. I'm sure there are far more precise scientific methods of determining pixel quality without eyeballing the legibility of small text.

Anyone who employs your method is giving herself unnecessary work, because it's the performance of the whole sensor that's important.

If we take Marc's second set of P30/5D shots where he's got the focussing better (on another thread), we can see that the P30 delivers more detail in the shadows in same size images, by virtue of the fact it has a great deal more pixels, yet there seems to be some doubt as to whether the P30 at the pixel level has any more DR than the 5D.

Your method of shooting the chart would be useful in this situation, to shed light on the performance of these two cameras at the pixel level. However, in the two equal size crops below of the dark cupboards and drawers, it's clear that the 5D image has unacceptable noise but the P30 image would be acceptable as a print.

If one were to substitue the image of the drawers for your chart, it is easy to imagine that the P30 image would reveal any text more legibly.

[attachment=4488:attachment]

[Jonathan Wienke]

I get a sense here, Jonathan, that you are overcomplicating a situation which is really only useful to the practical photographer as a guideline. I'm sure there are far more precise scientific methods of determining pixel quality without eyeballing the legibility of small text.

Granted, but those "more precise" statistical measurements have not shown themselves to be particularly useful in actual practice. S/N ratio is very precise, but tells little about real-world DR performance.

Anyone who employs your method is giving herself unnecessary work, because it's the performance of the whole sensor that's important.

Not as important as you imagine. The P30 has 31MP and the 5D just under 13MP. The image-level DR advantage from the P30's extra pixels is less than 1 stop. Having more pixels to throw at the noise problem is nice, but trading resolution for DR is very expensive. You give up 3/4 of your pixels for every additional stop of image-level DR. Unless you have gross pixel quantity overkill, pixel quality is far more important than pixel quantity because trading pixels for DR is a very ineffective solution to the noise problem.

If we take Marc's second set of P30/5D shots where he's got the focussing better (on another thread), we can see that the P30 delivers more detail in the shadows in same size images, by virtue of the fact it has a great deal more pixels, yet there seems to be some doubt as to whether the P30 at the pixel level has any more DR than the 5D.

Your method of shooting the chart would be useful in this situation, to shed light on the performance of these two cameras at the pixel level. However, in the two equal size crops below of the dark cupboards and drawers, it's clear that the 5D image has unacceptable noise but the P30 image would be acceptable as a print.

My DR test method will tell you exactly how much of the P30's DR advantage comes from increased pixel quality and how much comes from increased pixel quantity. If you upgraded from the 5D to the P30 because you needed the extra resolution, knowing how much of that extra resolution you get to keep in DR-challenging situations is of considerable practical value. Can the P30 beat the 5D DR-wise only when you downsize its files to 12.8MP, or can it still beat the 5D at 31MP?