Many people are saying the 1DsII only has marginal extra quality over the 5D.
Indeed! Over at Rob Galbraith's forum there are some people even claiming the 5D produces better
quality images than the 1Ds2, but I have great difficulty in believing the objectivity of such reports.
What does seem clear, however, is that any resolution advantage of the 1Ds2 is so slight that professional photographers have great difficulty demonstrating the fact over the internet, which seems very odd to me because I have no difficulty demonstrating the superior resolving power of the 20D compared to the 5D, even using a medium quality lens like the Canon 100-400 IS at 400mm.
Using the same lens from the same position with the 20D, 5D and 1Ds2, then cropping the 5D and 1Ds2 images to the smae FOV as the 20D shot, we would be comparing 8mp (20D) to 4.8mp (5D) to 6.4mp (1Ds2).
Now there's no doubt in my
mind that in the above situation, the 8mp of the 20D produces better detail than the 4.8mp of the 5D. However, considering the very marginal difference between the 5D and 1Ds2, I would expect a similar very marginal difference between the 6.4mp of the 1Ds2 and the 8mp of the 20D.
I can't test it of course because I don't own a 1Ds2, but in principle 2 very marginal improvements add up to one bigger and more definite marginal improvement.
I therefore conclude that, if Canon offers us a 20mp upgrade to the 1Ds2, which will have the same pixel density as the 20D and 30D, those folks at Rob Galbraith's site will then be able to declare quite definitely that the new 20mp 1Ds3 produces better image quality than the 5D .