| Comparison At 28 |
 |
| Comparison At 50 |
 |
| Comparison At 70 |
 |
In previous tests I have done, I just scanned the images at 2400dpi and compared the scans to see if I could find a difference. This basically corresponds to looking at nice 8x10 prints. The differences in resolution were often subtle at this scan resolution, so I thought I would try something a bit different. I used a microscope to take pictures of matching regions on different slides, then I scanned those pictures. The links below contain the scans of the microscopic pictures.
The tests results I have posted so far really only deal with resolution. I am not convinced that a typical CCD scanner has the ability to pickup very subtle differences in tone and therefore have not tried to post comparisons other than resolution. However, I did try a couple of comparisons looking for a difference in the lenses abilities to render subtle tones differently, evaluating the results on a light table. In general I found that the exposure between lenses often varied by 1/5-1/3 a stop or so and this had a bigger effect on the tone than any other difference. I am also guessing that shallower depth of field, and less light falloff of the "L" lenses accounts for some claims that the "L" images have a certain quality to them that you don't get with the consumer versions. When looking at the slides on a light table, I did not consistently see any difference between the photos other than, exposure, light falloff, and resolution. This does not mean it is not there, just that I have not found it yet.
I also tried a quick flare comparison. I was not looking for a comparison where there was a bright light source directly in the view finder, because I can usually work around that situation. I was looking for cases where light sources, the sun, were far out of the frame, but were shining on the front lens element. Again I failed to detect a difference because the exposure variation was too great. I am guessing for cases where a bright light source is not close to being in the image frame, the differences may be subtle.
A few notes about the comparison shots to give a sense of scale. Each of the microscope pictures is aproximately 1.4mm across. I have reduced them to a width of 600pixels, resulting in aproximately .00233mm on the slide per pixel. If you are looking at the scans on a typical monitor with .27mm dots, you are looking at an aproximate 115.8 times enlargement of the slide. This is equivalent to a 4.2m x 2.8m (14.6ft x 9.7ft) print.
Also .00233mm should be enough resolution to, in theory, detect 213 lpmm. Given all of the optical systems that were involved in producing the final image, I would not guarantee that, and I have not tried to test it empirically. I am pretty confident the results do show any difference in resolution one is likely to encounter using the lenses for normal photography.
Technical information:
EOS 3
Bogen 3001 tripod with 3025 head.
28 2.8, 50 1.8 Mk I, 28-105 3.5-4.5, 28-70 2.8L
All shots taken with mirror lockup on a day without wind.
Provia 100 (I heard someone report that you could not see the subtle differences in tone when super saturated films like Velvia were used, so I tried Provia)
All pictures were taken on two roles of film and sent to Kodak for processing.
EOS 3 with T-Mount adapter used for microscope pictures.
Fuji G-200 negative film used for microscope pictures.
All scans done on Minolta Dimage Dual
While you are here you might want to check out some of my more interesting photographs.
The green rectangles indicates the region compared for each image. Click on an image to go to the corresponding index of comparison pictures. The images are fairly large so if you have a slow connection it will take a while for them to download.
| Comparison At 28 |
|
| Comparison At 50 |
|
| Comparison At 70 |
|
