Feeds:
Posts
Comments

Posts Tagged ‘lens sharpness’

Few months ago, we compared three Nikon 50mm lenses and examined center and corner sharpness. To complete the exercise, this time, we’ll compare three Canon lenses – Canon 50mm/1.4, Canon 50mm/2.5 macro, and a 17-50mm Tamron zoom XR DiII SP. The Tamron zoom is better and faster than the Canon 18-55mm kit zoom that was omitted from the comparison. For my type of shooting, I find also, that the 17-50mm range is more useful than 18-55mm.

All images were obtained on a bright, sunny day, at F8, and 1/1600s, and a APS-C format Canon T2i camera was mounted on a solid tripod. The images were not manipulated nor sharpened.

The first set of images shows the overall uncropped images (reduced to 720×480 size).

1. Canon 50mm/1.4 – ISO 200, F8.0, 1/1600s

2. Canon 50mm/2.5 – ISO 200, F8.0, 1/1600s

3. Tamron 17-50mm – ISO 200, F8.0, 1/1600s

The primes show better contrast and more vibrant colours than the zoom lens (notice the red chimneys and the reflection in the water). On the other hand, Tamron lets more light through at the same aperture. If you are going to print it in small size or use it just for Web, you won’t notice any difference in image quality, regardless of the lens used.

 

Next set of images shows a crop of the middle section (you can click on the images and see it in full 100% size).

1. Canon 50mm/1.4 – ISO 200, F8.0, 1/1600s

2. Canon 50mm/2.5 – ISO 200, F8.0, 1/1600s

3. Tamron 17-50mm – ISO 200, F8.0, 1/1600s

Both Canon primes are very sharp in the middle section (remember, this is output from a RAW, unsharpened image). As expected, the zoom lens is not quite as sharp, but again, printed at small to medium size or reduced to a web size, it is quite acceptable (you’ll need to view the samples at 100% to see the difference).

 

The last set of images shows the leftmost section in 100% crop

1. Canon 50mm/1.4 – ISO 200, F8.0, 1/1600s

2. Canon 50mm/2.5 – ISO 200, F8.0, 1/1600s

3. Tamron 17-50mm – ISO 200, F8.0, 1/1600s

Both Canon primes show excellent corner sharpness and some moire on the metal netting of the deck balcony. The 50mm/2.5 macro shows slight amount of green fringing along the vertical edges (noticeable on the two white beams under the roof), but that can be easily corrected in Lightroom). The Tamron zoom, as expected, is not quite as sharp, but due to the loss of knife-edge sharpness, it indirectly gets rid of the fringing, and moire on the balcony is also less pronounced.

Conclusion:
If you need the best quality, the 50mm/2.5 macro is hard to beat (it is also the least expensive of three tested lenses). In image quality almost indistinguishable from 50mm/1.4, and you get also the macro capability. Of course, if you need a fast lens in low-light situation or better isolation of the main subject, the 1.4 aperture will be the better choice. Tamron zoom is the longest and heaviest of the three, and not quite as sharp as the 50mm primes, but in good light and exposed at its sweet spot, it delivers good image quality and a useful range of focal lengths.

One more difference:

If you look carefully at the rightmost window in the cropped version in the last set of pictures, you’ll notice a startling difference. In the first two images, photographed with Canon prime lenses, the test subject sits straight and appears interested in his surroundings, whereas in the last photo he stopped cooperating and fell asleep.

Read Full Post »

Recently, as I was culling my latest batch of images, I noticed a focal asymmetry in one of the images photographed by Nikon D300 and Nikkor AF-S 70-300mm VR-G lens. The building in the image with its bricks and windows makes for a very good test subject. I was standing parallel to the building, across a river about 100m from it, and although its corners are technically slightly further from the camera than its middle portion, the DOF at that aperture should still portray everything in the focus.

The EXIF data shows the following info: ISO 200, 110mm, F5.6, 1/1000s. Shot under almost ideal conditions, one would expect a perfect picture. Captured at the minimum/base ISO, the zoom lens was in the middle part of the barrel, aperture close to the lens sweet spot, and the shutter speed more than adequate for this focal length. To my surprise, the left third of the image was badly blurred relative to the middle and right portion of the image (see below – and click on the images to viwe them in full 100% crop size).


Left Image Segment:
notice the blurred bricks and windows, with the blur getting stronger toward the left edge. The building behind, although not that much further, is totally out of focus.


Middle Image Segment – everything sharp


Right Image Segment:
As sharp, if not sharper than the centre portion.
Notice the sharp bricks and windows, and even the buildings behind are still in focus.

Obviously, there is a problem with the actual lens or alignment of the camera sensor. Very likely, both. Focal microadjustment by LensAlign won’t help in this case, since the center of the image is already sharp. But that is a topic for another article.

As disturbing as this situation is, it is not that uncommon. Many lenses will show a dropoff of the sharpness towards the edges. As a matter of fact, one reason why the rotational panoramic cameras take very sharp pictures even with less than stellar lenses, is the fact that they utilize just a narrow slit in the middle of the lens that is always sharp (unless you use a really bad copy of a lens).

What does all that have to do with panorama stitching?
Let’s assume, you just shot three consecutive segments in horizontal orientation that you plan to stitch. Well, most stitching programs will attempt to use the leftmost image as the left portion and the righmost image as the right portion of the panorama. The middle image will be inserted somewhere in the centre, not necessarily along straight vertical lines. Makes sense, n’est ce pas?

Unfortunately in a situation, as described above (the lens is less sharp in the left third), if you use an automated stitching program, you’ll be dealt an unsharp half of the left image and possibly also an unsharp left part of the middle image. If the lens is sharp on the right, the right side of the panorama should turn out OK. Now, if you take five or seven segments instead of three, you may get slightly better results, but not that much, since that first (leftmost) image will be used almost in its full size anyway.

So what to do if you want a sharp panorama? F8 or F7.1 aperture may yield some improvement when it comes to the optimal lens aperture, slightly greater DOF, and lower diffraction (depending on the format of the camera), and of course, factory calibration of the lens and camera is in order. But nevertheless, most lenses will exhibit less than ideal sharpness on their extreme edges, so the basic problem remains.

One practical solution is to shoot more segments (i.e. 5-10 or even more, rather than just 3-5), overlap them generously at 30-50 percent, and crop all these images on both sides, so that you retain only the sharpest middle portion of each image. Now YOU are in control! Feeding the sharpest portion of the images to the stitching software will result in a much sharper pano. Just be careful not to lose the overlap when cropping.

Another trick is to compose the first (leftmost) segment in such a way, that its left third will be to the left of the planned edge of the panorama and consequently trimmed off (planned throwaway). Coincidentally, we had to do it also in the old days of rotational cameras to allow the motor to ramp up to its proper speed.

Read Full Post »