Resolution, resolution, resolution…
There’s a lot of talk about resolution in lenses. And not only resolution but also corner to corner sharpness. And why not? When you decide to get a lens you read the specs on the web (well, I do anyway) and among other numbers, the resolution is always there. It’s good, since numbers can be compared, thus you always know which lens is the best?
No, not really! A whole bunch of other criteria and aspects also weigh in, but numbers are easy, concrete and precise and we often let us be lured by some hypothetical aspect just because it has a number lower than, or higher than, another number. It is measurable and we compare. Will the photos get better just because of a number? Of course not.
Another side of the coin is what people say about a lens. That is even trickier. Look around flickr or just about any other photo site or forum and opinions are plentiful. Who can you trust? Different photographers crave different gear because they shoot different things. A wedding photographer might want to be able to shoot an entire wedding in a dimly lit church, while others shoot macros of well lit machine parts for catalogues. It stands to reason their equipment will vary? What do you shoot mostly?
Just looking at pictures on the web can say more about the photographer than the lens. Skilled photographers can do wonders with a good lens, but the reverse is not necessarily true, i.e. good lens makes a skilled photographer.
In many instances a highly acclaimed lens will actually produce good shots (at least technically) but how much better than the one you already have? It is hard to say, inspiration, a good “look” for scenes also play in here. So, “I’d better get an expensive lens to be on the safe side”, isn’t that how we think? I’m a bit of a gear nerd so I’m biased here.
On the site lenswork I stumbled upon a lens test chart article with a printable test chart (or “chartelette” perhaps, as only the basic building patch is presented?). The site has other interesting discussions and articles too. Could be worth a read.
Anyway, I printed the 26x and 52x patches on a laser printer and nailed them to a wall. From a distance of 26x focal length of my Mamiya Sekor C 45/2.8 a number of shots were taken in the hope that resolution figures would pop out somewhere in the process.
Setup and test
Isn’t it surprising that every time you really try to do something properly it is much more difficult that initially anticipated? I get this everytime…
I rigged the camera on my trusty Berlebach Report 9033 tripod. Then measured the distance, 26 x 4.5 = 117 cm. Ok, measured from the wall, even I got that right, but where too? To the surface of the first lenst? To the plane of the shutter? Hmm, I decided to the surface of the first lens since I can actually measure that. Simplicity decided that. How do you properly measure a distance from the lens to the wall when the lens is on a tripod? I measured along the floor and used a plummet on a string to get the vertical from the lens to the measuring tape. It is sort of exact.
Next the plane of focus. How do make the plane of the camera exactly parallell to the plane of the wall? As tipped in the article’s comments I used a mirror on the wall and was as careful as I could. I did not find it all that easy as the comment suggests it is. Oh, well, perhaps it comes with experience? For now I only really trust the centre test patch
Then there was the actual test. On the 5D2 I used live view and 10x magnification, it was still not 100 percent accurate. Even the lightest touch of the focus ring transformed the image somewhat, whether better or worse was hard to tell. More magnification needed.
The actual test results come here. They are probably not comparable to anything else and for now only shows the centre patch. The image below shows the centre resolution of a Mamiya Sekor C 45/2.8 at f/2.8, f/5.6, f/16 and f/22 respectively (in that order from left to right).
We see that the lens easily resolves 40 line-pairs per millimetre (lpmm) but above that the lines are barely discernable in a moiré pattern. I can’t help thinking that more exact distance setting would have resolved the 56 lpmm better.
Another image, which show more optical deficiencies, is shown below. This time it shows the patches from the upper left corner of the view. Again it is a Mamiya Sekor C 45/2.8 at f/2.8, f/5.6, f/16 and f/22 respectively (in that order from left to right)
(I have no idea why the second image is smaller…)
First of all we see that the upper left part has the same resolution as the centre patch. Very good! It is also darker at f/2.8 (shows better on my screen than in the image above strangely enough), this is the expected vignetting. The vignetting disappears when stopping down and at f/5.6 is gone for all practical purposes.
But we also see a glow of red around the edges of the black bars. This is the aberration, an optical phenomenon due to all wavelengths not being treated the same in the lens. Normally the aberration also should be less when stopping down. I don’t know why it seems worse at f/16 than at f/2.8 and f/5.6.
When I had propped up the camera I also took some shots with the Ultron 40/2. I’m not yet ready to share them but it appears to have the same or slightly better centre resolution and less aberration, actually probably non-existant aberration as far as I can measure it.
Where does all this lead? One is definitely forced to think along the lines of “if it so difficult so set an accurate focus in this controlled environment will the results be useful at all when it comes to real life?” I should say no. Only by haphazard will I be able to set the focus in such a way the lens full potential is revealed. Too bad. You focus as good as you can and then hope it is enough.
Corner to corner sharpness then. When I think about it very few (read: none) of my daily shots are of objects perfectly parallel to the sensor’s plane. The corner to corner test is useful for testing if a lens is symmetrical in its field of view, which I think is something I expect it to be. But apart from that? In real life? Not much to bother about.
For a pure landscape artist then corner to corner sharpness would be interesting but they usually stop down a lot to get depth of field anyway, and with stopping down they are not all that picky about sharpness in a plane.
What might be more important is if the lens suffers from vignetting, aberration, distorsion, or what is the maximum f-stop, size, material, ease of handling etc etc. So perhaps resolution is in fact the least important consideration?
Either way the Mamiya Sekor C 45/2.8 is no slouch on a Canon 5D2. It is in fact very good!