I got ahold of a Nikon D800 and tested it for image resolution compared to the Nikon D3s using a high quality Nikon 85mm f2.8 PC-E lens. I am very happy to report that the D800 does indeed capture an unprecedented range of resolutions available from any full frame 35mm sensor camera. However, as I expected, the D800 falls far short of the D3s in terms of signal to noise ratio, especially in the high gain ISO ranges and in particular long exposures coupled with high ISO gain settings.
First lets tackle the improved image resolution available with the D800. Here is a side-by-side comparison of resolving power using an 85mm lens taken at its peak aperture of f/8.0. The images were cropped and enlarged using Photoshop while using the Nearest Neighbor setting to preserve the hard edges of the individual pixels. The original D800 crop was 140 by 240 pixels.
Note that the D800 (left) has a much finer resolution and can resolve the largest type on the paint container while the D3s (right) has troubles resolving the same lettering. Also note the leaves and grass which are much more detailed in the D800 image. Below is the full image from which the above image was cropped from, the cropped area is shown highlighted:
Here are the actual tests comparing the MTF function resolution tests us ing the Koren methods and target chart. The D800 has about three times the pixel density and thus, as expected, an increase in resolving power of about the square root of three, or 1.7 times. Note that the best resolving aperture is f/8.0, the worst apertures are both f/32 and f/22 due to diffraction interference.
These resolution tests results were just about exactly as expected and the D800 is indeed superior to the D3s, strictly from a resolving power perspective. This is expected since the D800 sensors have pixel dimensions that are about 1.7 times smaller dimensions than the D3s. The cost of this higher resolution comes at a very dear cost, and that cost is the signal to noise ratios, particularly the signal to noise ratios achieved during long exposures. In this aspect, I’m sad to say, the D800 falls very short of the capabilities of the much larger pixel sized D3s.
These are cropped side-by-side comparisons of dark frame exposures taken at ISO 6400 with the D800 on the left. The first image is with an exposure of one second and the second image are sixty second exposures. The images were captured in a dark room to eliminate the possibility of light leakage. The noise that is visible is a combination of hot pixels (pure red, green, or blue dots) and thermal noise found in ALL digital sensors.
In order to achieve these results I exaggerated the noise by using the following setting in Adobe Camera Raw; exposure +4, Brightness +150, Contrast -50, Sharpening 0, all other settings were the initial program default values. I also set the camera menu settings to Set Picture Control = NEUTRAL, High ISO Noise Reduction = OFF, Long Exposure Noise Reduction = OFF, and Active D-Lighting = OFF. I did test the noise reduction capabilities and found about the same differentials of noise levels between the two cameras. However, I have found that taking at least one dark frame at the same time as a the long-high ISO exposure, and then using these dark images as subtraction frames during the post image processing is by far superior to any in-camera noise reduction capabilities.
In short the D800 is a great tool for well lit stationary subjects taken at relatively short exposures, but is completely unsuited for long exposure or high ISO photography. in other words the D800 is completely unsuitable for astrophotography. The D800 is very well suited for portrait, product, and landscape photography, but it is much less appropriate for fast action sports or wildlife imagery. These latter areas are best coupled with the lower noise faster frame rate Nikon D3 or the more current Nikon D4 models.