I am not under the impression that any of my lenses are less sharp on my 1Ds3 than they were on my 5D. Actually, all of my L primes give excellent performance because of the MA. They have never been sharper. To take this a bit further... A while ago we did some comparisons between the Nikon D3 + 85 1.4 + 24-70mm and the 1Ds3 + 85L II + 24-70L and there was no real world difference between the files regarding sharpness. Of course when equalizing the files the 1Ds3 showed the impact of having more resolution... it's files were sharper and more detailed. Unfortunately we didn't do a noise comparsison
IMHO this whole "the 1Ds3 needs the best lenses argument" is a bit of a myth. YMMV
Daan B wrote:
IMHO this whole "the 1Ds3 needs the best lenses argument" is a bit of a myth. YMMV
I'd say yes and no. In order to achieve the absolute maximum resolution posible from the sensor it's obvious you need a very precisely focused, quality optic, the same way you would on any format.
However, because of the number of pixels you can get away with a 'poorer' lens because you don't need to enlarge the 'defects' so much.
I posted here some time ago a shot taken at 3200 ISO with my 1Ds3 and my prosumer 70-300IS which I thought was astonishing, considering. You can actually read the label on the apron on his hip, at 200%:
I don't buy into the assumption the 1Ds3 is more demanding on the lenses than the 5D or any other lower pixel count camera. Of course if the purpose of taking pictures is to examine them at 100%, then yes, the above assumption would be true. But most people take photos to display them as prints and in this case an 8x10 print from a 1Ds3 and a 5D should look similar. A good sharp lens will be good on both cameras and a soft mediocre lens will look bad on both. It all depends on the final output.
edwardkaraa wrote:
I don't buy into the assumption the 1Ds3 is more demanding on the lenses than the 5D...
I use both at every job. If you want the 21 megapixels to earn their space on your hard drive, you have to choose lenses and apertures more carefully than with the 5D. For example, my 35 f2 and 35 f1.4 each show more pixels of CA on the 1Ds3. My Zeiss 24-85 often requires extra sharpening on the 1Ds3, whereas the 200 f1.8 doesn't. Some lenses don't seem to resolve as crisply, PER PIXEL, on the 1Ds3 as they do on the 5D. The lens isn't worse, but the extra pixels aren't showing good data and earning their keep. To use the extra resolution of the 1Ds3 to good effect, you need to tighten up lens choice, technique, and focus accuracy. It's a small effect, but it's there.
Yes of course, I agree completely with you. But I'm making a slightly different point, in that when we used to shoot Velvia 50 for example instead of Provia 100, we knew the Velvia had an extra 20 pl/mm resolution over the Provia, but I personally was careful with my lens selection only when I knew the photos were going to be enlarged substantially. Otherwise, my choice of Velvia over Provia was purely of esthetic reasons not resolution. The 1Ds3 files, if they are destined to be reproduced at sizes within the reach of the 5D, I believe that the final output would be similar with both cameras. Only when you intend to use the 1Ds3 to the maximum of its resolving capabilities, to produce huge enlargements, then of course you need a lens that walks the walk. I never doubted that, per pixel, the 1Ds3 requires much more from lenses than the 5D, this is absolutely true.
edwardkaraa wrote:
I don't buy into the assumption the 1Ds3 is more demanding on the lenses than the 5D or any other lower pixel count camera. Of course if the purpose of taking pictures is to examine them at 100%, then yes, the above assumption would be true. But most people take photos to display them as prints and in this case an 8x10 print from a 1Ds3 and a 5D should look similar.
Who on earth buys either a 13 or 21 megapixel camera for an 8X10 print? In my experience 100% view on screen pretty much shows exactly what the print will look like at native resolution in print unless you buy into the whole 'viewing distance' bull**** (like viewers ever stay behind an imaginary line!). I've found that 50% on screen is accurate for 'official' viewing distance detail wise.
This very indepth and technical article http://luminous-landscape.com/tutorials/resolution.shtml explains why testing has to be at the limits of the mediums capability but also I think answers the question this thread is dealing with.
A lot of the issues seen with the 1Ds mkIII may well be diffraction based, I decided not to buy one for a specific project as I was pretty certain that I would be negating any resolution benefits due to the high level of DOF needed and the resulting small apertures (f16/22).
There seems to be a physics brick wall if you are attempting to resolve a certain level of detail with sufficient DOF and without using wide angle lenses. Bigger sensors demand longer lenses for an equivelent FOV which in turn requires a smaller f-stop and the resulting diffraction. At present I decided to fall back on stitching for when I need a lot of resolution for huge prints, but don't want to use wide angle lenses. I shot this picture today, 30 megapixels from a 5D (7 frames), shot at f22 (due to 100mm lens) and sufficient resolution to easily outweigh the diffraction limitations of using f22 which on the 5D, though noticeable, are far from as bad as on higher pixel density sensors.
