The same should actually be done when comparing a 1ds2 with a D2x for instance.
[a href=\"index.php?act=findpost&pid=74149\"][{POST_SNAPBACK}][/a]
Bernard,
Indeed it should. I sometimes forget about such issues myself when I don't have BJL to remind me (what happened to BJL? Hope he's all right.)
I'm beginning to think that the thorough tests at dpreview deliberately avoid confornting these issues in order to generate controversy, although to be fair, they do sometimes mention the 'actual' ISO values as in the following chart.
[attachment=915:attachment]
I think you'd agree, the D2X with its 3 options of noise reduction, complicates the comparison. If you compare the D2X noise with full noise reduction on at ISO 800 with the 5D at ISO 1600 (which is actually ISO 2000), then the D2X is definitely less noisy, but at the expense of resolution loss. If you compare the D2X at ISO 3200 in standard mode, which uses a moderate amount of detail-destroying noise reduction, with the 20D at ISO 3200, we see that noise levels are about the same. The questions then arise, 'How does total image quality on equal size prints compare?' 'To what extent does the greater number of pixels (of the D2X)compensate for, perhaps, a more aggressive, resolution-reducing noise reduction?'
It seems to me that these important practical considerations are often skipped over far too readily.
[attachment=916:attachment]