Pixel Peeping

I recently had the opportunity to learn firsthand how difficult meaningful pixel peeping can be. A friend let me use his Canon 5D MKII and I was interested in finding out how the native file sharpness compared to my 1Ds MKII. So I decided to get out my tripod and take the same shots with the two cameras and see how the RAW files compared. I knew there would be some difficulty in comparing the two images based purely on the difference in the number of pixels (16 vs 21 MP), but what I hadn't considered is what other issues might effect the comparison.

I learned a lesson as to just how difficult it can be to make such comparisons in a meaningful way, as each time I looked at a pair of images on the computer screen I thought of another confounding variable that might be affecting the results. The exercise pointed out to me how careful one needs to be when reading comments and evaluations on various internet forums, where perhaps everyone does not take all these factors into consideration when making such comparisons.

What follows are some of the confounding variables that I was able to recognize:

1) Metering: As it turns out, the two cameras gave different exposures despite having no exposure compensation dialed in for either. I hadn't considered the fact that different exposures meant different appearances to the highlights and shadows. Thus, the photo need to be taken in manual mode with the exact same aperture and shutter speed and not simply taken on aperture priority.

2)White Balance: Yes, the color balance of the two cameras was somewhat different and this can lead to different amounts of apparent contrast which appears to make sharpness differ. To make a fair comparison I had to do a custom white balance for each.

3)Focus: By far the biggest issue of all is the factor of how well the camera is focused. I don't have the greatest eyes in the world. I was very surprised at how different (at 100% viewing on the computer) several different manual focus shots were in terms of sharpness (I was shooting at a distant object at 200mm focal length to try to 'stress' the camera's resolving power). Many images were similar but some demonstrated quite a difference in sharpness despite coming from the same camera with repeat manual focusing. So one would have to take several shots and pick the sharpest one for comparison.

What about auto-focus? Despite using the center focusing point I was surprised to see that on the 5D MKII the sharpest area was not what was directly focused on. There was a bit of front-focus. We are not talking about a lot here...perhaps a few inches and I was using a Canon 70-200 F2.8 IS at 200mm focal length and f2.8. So one could not compare sharpness across two images if the exact same point is not optimally in focus in each image.

Not ever having used Live View before I really hadn't thought it would be that useful to me. However, it turns out that it is an amazing manual focusing aid. With live view, one can magnify the image 10x on the LCD and manually focus, watching the sharpness change right on the screen....amazing. It really is the ultimate focusing aid.

So I used Live View manual focusing on the 5D MKII and auto-focus (which seemed more accurate than my manual focusing without Live View) on the 1Ds MKII. Is this 'fair'? What if the 1Ds MKII had Live View? Would manual focusing using Live View provide better results than auto-focus? I don't know, and it is a theoretic question that was unable to be answered.

4)RAW Processing: I soon learned that to be fair about the comparison one should not process the RAW file in any way in Camera RAW or Lightroom as this will alter apparent contrast and sharpness. Likewise, sharpening in the RAW processor needs to be disabled.

5)Different File Sizes: As I mentioned, I expected differences based on the 16 vs 21 MP count. I found myself wondering if the two images might be more directly compared by upsizing the 1Ds MKII or downsizing the 5D MKII images? But wait....that would require interpolation, which would affect apparent sharpness, potentially making the downsized image appear sharper or the upsized image appear less sharp.

So what was my conclusion? The two cameras seemed pretty comparable, but I was amazed that the 5D MKII images looked at least as sharp out of the camera as the 1Ds MKII images despite the larger number of pixels. In fact, as one might have expected, the amount of detail seen in the 5D MKII images was greater with no apparent loss of edge definition (definition might have even been a tad better) given the higher megapixel count.

However, that was not the major conclusion that I arrived at from this exercise. What struck home the most is how difficult it can be to make these types of comparisons in an accurate and meaningful way. So beware the next time someone you don't know says that camera x produces images that are more (or less) anything than camera y. At least ask how the comparison was made and whether these issues were taken into consideration. For all I know, I might well have forgotten something obvious in the way I made my comparison.....believe me, I didn't figure out these variables that could affect results all at once!