I recently sold a couple of large prints to a friend and went through a couple of days of testing how to get the best appearance from big enlargements. Along with that I’ve been reading about sharpness differences in lenses and wondering about the impact on enlargements. Being a nut for sharpness in my landscapes I decided to do a little testing on my gear and see what differences I could find along with techniques to maximize sharpness.
You may never need to worry about this issue but I’m the photographer who walks up and puts my nose right at a big photograph just to see how sharp it is. For normal people, the “correct” viewing distance is around 2 times the diagonal dimension of the picture (where infinite sharpness is less of an issue). Your brain is pretty good at dealing with this – watch people at an exhibition moving back and forth in front of a painting or large photo and you’ll discover most of them land in the same spot, which is around 2 times the diagonal of the image they are examining. Has to do with angle of view of our eyes and how much we jump around in an image to take it all in.
Anyway, to compare lens sharpness I have two lenses with a 50mm focal length – one is a fixed macro at 50mm and the other is a zoom whose range includes 50mm. In order to learn what differences might show up in processing I have two software tools for enlarging – Photoshop CS5 (bicubic) and Perfect Resize 7 (formerly known as Genuine Fractals). Software packages make enlargements by literally creating new pixels in your image using sophisticated algorithms that try to made the new pixels blend in with the old ones. I can’t begin to explain how this works – it’s just magic to me.
My test is a simple one: photograph a straight edge with the two lenses at a fixed distance and then enlarge the image with the two software tools and compare the results.
As a summary I found the two lenses have little difference in sharpness in this test – the differences were greater between the software tools than the lenses.
I set up my camera at 64 inches from my target (just a random choice), a razor blade in front of a piece of white foamboard. Lighting was from above, a 65 watt flood light (hence the really warm color in the images). Camera was on a tripod, manual focus using LiveView on the back of the camera and a 5 second delay between the mirror lock-up and the shutter release.
I took the two image files (one for each lens) and processed each through the software tools, enlarging the image to 30″ in the longest dimension, which is about 2.5 times larger than the original image at 300 dpi. After setting my view in Photoshop to 100% I cropped each image to the same dimensions. No further processing was performed and the final, cropped images were saved as JPEG’s at maximum quality.
What I’m seeing is little difference between the lenses for sharpness; i.e., rendering the edge of the razor blade as a crisp, distinct line. Viewing the original files at 200% showed little difference. As far as enlargement, though, there are definite differences between the software tools.
Photoshop’s bicubic enlarger results in a smoother look to the final image, which for me results in a less sharp edge. Perfect Resize did a better job of retaining the sharp edge but the work of the algorithm to add extra pixels around each existing pixel gives a type of pointillist look to the image (granted, when viewed at 100%) that Photoshop’s smoother result doesn’t have. I’m concerned the image from Perfect Resize will be more sensitive to halos during final sharpening than Photoshop’s version, these halos being what makes some digital images look artificial to the eye. In final sharpening the software doesn’t know which pixels are original and which were created during the enlargement so it works on all of them, which can result in artifacts around edges that distract from the image.
Two conclusions. First, using what I already know about maximizing sharpness in the camera minimizes differences between the two lenses I use – tripod, manual focus, mirror lockup, shutter release delay. Second, at significant enlargements (greater than twice original size) I need to be careful with the software tools, perhaps choosing one over the other depending on how important fine details are to the final image. Additionally, subsequent sharpening will have to be applied carefully so as not to hurt all the prior work getting to a pleasing, larger image.
From all this I’m less terrified about offering larger prints on my website as I now know what to watch out for as I make adjustments to the images I upload to my site. It would be nice if one tool did everything perfectly but photography, like other crafts, has learned more specific tools applied correctly usually return the best results. That and the craftsman knowing how to use them properly and for the right application.