I read several claims on photography forums that the Canon 5D camera produced slightly more detailed files than the Nikon D700. Forum posters gave various explanations as to why this might occur such as the presence of a stronger AA filter on D700 or a slightly larger file size for the 5D. Since I had both cameras, I decided to test if in fact the 5D had noticeably better resolution.
I took a series of shots of two architectural subjects using the Nikon 105 micro AF lens on the 5D (mounted with a CameraQuest adaptor) and D700. In both cases, the camera was supported by an Acratech Ultimate Ballhead and a Velbon 640 carbon fiber tripod. I used a bungee cord attached to the center post and a remote release to fire the shutter.
I took 3-6 pictures of each subject with both cameras with the same framing and position. The pictures were shot at f/8, 1/500 sec exposure, and ISO 200. In between each shot, I reset the focus and refocussed the lens. This was to ensure that if I made errors in focussing they would not invalidate all the samples. On the Canon camera I focussed manually and on the Nikon I used AF for about half the shots and manual focus for the remainder.
I processed the Canon RAW files with ACR 4.5 with the default sharpening applied (amount = 25, radius = 1.0, detail = 25, masking =0). Nikon files were processed in Nikon Capture NX2 with the picture control sharpening set to 3.
Once the images were processed from the RAW files, I examined the series for each camera/subject combination and selected the sharpest one to evaluate. For this article, I converted the images to jpegs as the last step using Photoshop (save for web, quality = 60).
For the first test subject I selected the east entrance of Stanford University’s Quad (Figure 1). The east gate made an ideal subject since it was stationary with fine detail in the structure. There were slight color and exposure differences between the Canon and Nikon files, but since I was interested in resolution, I made no attempt to normalize these aspects of the photos.
Figure 2 shows pixel level detail at 100% magnification. Examining these crops, I find the photos to be essentially identical. This was somewhat of a surprise to me as I expected slightly better resolution on the 5D since that was reported in various internet forums.
The second test subject was the north entrance of the Quad (Figure 3). When examining the 100% crops in Figure 4, I find that the D700 appears slightly sharper than the 5D image. However, given the results from picture 1, where the images were of equal resolution, I believe any differences here are not significant and could be due to mis-focus or tripod vibration from wind, etc.
In the first crop of Figure 4, I can make out slight jaggies in the 5D image which are not present in the Nikon file. I highly doubt these would be visible in any normal sized print.
|Figure 1. The east entrance of the Quad at Stanford University. Red squares indicate 100% crops shown in Figure 2.|
|Figure 2. Pixel level detail from Figure 1 shown at 100% magnification.|
|Figure 3. The north entrance of the Quad at Stanford University. Red squares indicate 100% crops shown in Figure 4.|
|Figure 4. Pixel level detail from Figure 3 shown at 100% magnification.|
Comparing two different cameras is an exercise in controlling as many variables as possible. I was able to use the same lens on both cameras, which I believe eliminates a large source of variation, but there are several other aspects of this comparison that I was not able to completely control:
- RAW processors. I used different RAW processors for each camera and this may introduce quality differences unrelated to the camera’s native ability. However, using the same processor for both cameras is not necessarily a good solution either since there is no guarantee that the rendering algorithm is equally well suited for both. In this test, I do not believe that the RAW processor made a significant difference since both cameras produce the finest pixel level detail I have seen to date.
- Sharpening. As the camera makers use different strength AA filters, each file may require a different amount of sharpening to be comparable. In this test, I used the default sharpening (in ACR/NX2) which could potentially lead to differences in evaluation if one camera received more or less. However, looking at the 100% crops and based on my experience with Canon and Nikon RAW files, I believe that neither file received enough additional sharpening so as to change my conclusions.
Finally, as I processed the files and evaluated the results, I realized that when looking at 100% crops the variations caused by differences in manual focussing were bigger than the difference in the inherent resolution of the 5D and D700 image files.
The two cameras do not have any significant difference in resolution, even when the image files are examined at 100% magnification.