Yes, the Canon is not bad, better detail than the ZD in this test, but the 18MP P21 clearly wins.
I assume the P21 would have better DR than the Canon too, but we can't tell without the RAW files.
No, the P21 does not
clearly win. Do you guys have blinkers, or what?
This test is flawed as usual. When comparing equipment which you know beforehand is going to have a very similar performance, it's essential to get the methodology correct, which means being very careful about small factors which can tip the balance one way or another.
With these tests, no care has been taken to get the FoVs the same in all 3 shots.
I've done my best to crop the 1Ds3 and ZD images to the same FoV as the P21. I've then divided the file size by 3 to get the relative pixel counts for the 3 cameras.
They are, 17.23mp for the P21, 17.46mp for the ZD and only 14.47mp for the 1Ds3.
We're not here comparing an 18mp P21 with a 21mp 1Ds3, but an 18mp P21 with a 14mp 1Ds3. Taking that into consideration it's quite likely the 1Ds3 has the edge.
There are also other issues of methodology which should be mentioned because they have some impact on results at this degree of pixel-peeping.
(1) If RAW images cannot be provided then at least let's have maximum quality jpegs. Here we have 60mb files compressed to around 3mb.
(2) When comparing formats with different aspect ratios, it's useful to make two comparisons, one with equal FoV in respect of height and one with equal FoV in respect of width. That at least provides one with some useful information. For example, if the subject matter lends itself to a 2:3 aspect ratio, I might want to know if I should pull out of my bag the 1Ds3 or just continue using the P21.
If, in these tests, the FoV of the image heights had been matched, giving an advantage to the P21 and a disadvantage to the 1Ds3, we'd be comparing a 17.8mp 1Ds3 image with a 17.23mp P21 image; close enough to avoid quibbling.
(3) Choosing f11 for both lenses in this test sort of equalises lens quality. Photodo never tested lenses at f11 because they considered all lenses to be equally bad at that aperture. The fact that such a relatively small sensor that the cropped 1Ds3 image represents, can so closely match the quality of the P21 sensor which is about 3x the area (of the cropped 1Ds3 image), using a lens of very similar quality, seems quite remarkable to me.
Essentially, the smaller sensor with equal pixel count can only deliver resolution and detail equal to the larger sensor if it employs sharper lenses. We don't have MTF charts at f11 for the lenses used in this test, but it is possible the Canon 50/1.4 is marginally sharper at f11 than the Mamiya 80/2.8.
(4) A major problem in equalising lens quality in this way results from inequality of DoF. At f11, the 1Ds3 image must have greater DoF, yet in these tests, the most significant difference in resolution and detail between the 1Ds3 and the MF images is in the foreground where the 1Ds3 is clearly fuzzier than both the ZD and P21.
How can this be? Clearly focussing is not the same for all three cameras. The 1Ds3 shot must have been focussed at a more distant point. Yet another flaw in these tests.