I'm going to use lenses that we optically adjusted back into proper alignment as examples. I'm doing this because it's a great opportunity to show you the difference between an optically misaligned and a properly aligned lens.
I'm not giving a tutorial on optically adjusting your lenses. Optical adjustment is very different for every lens, very time consuming and requires at least partial lens disassembly. Most people who try optical adjustment at home convert a below average lens into a totally useless lens.
For a couple of the examples, I'll show some Imatest MTF charts of the lenses so you can see some correlation with how our home test looks compared to the MTF numbers of the lens.
One last point. I'm using mostly Canon lenses for these demonstrations. The main reason is very simple; we stock more Canon lenses than all other lenses combined. So we see more Canon lenses with optical problems, and we're more practiced at fixing those. I can show you decentered pictures of some other lenses, but not 'after' pictures of them all better. Because we can't make them better.
Finally, if you haven't read the last article, there's not too much reason to read this. You probably won't understand what you're seeing.
High Quality Zoom: Canon 70-200 f/2.8 IS II
This lens was caught on routine screening after a rental. The tech who tested it wide open saw that the four corners looked like this and sent it straight back to the repair department. You don't need a good working knowledge of how sharp the corners should be to tell something is wrong.
First, let me make a point about why we only test with the lens at widest aperture. The image above was at f/2.8. I stopped down to f/5.6 and repeated the test to make the point that even bad lenses look reasonably good with the aperture closed down 2 stops.
The center looked fine and reasonably sharp when in focus, but just out of focus it exhibited a flare or halo pointing up toward 2 o'clock. (If you don't make it out at first, look at the center black circle and see how the white ring has smeared over most of it, leaving a little black dot at 2 o'clock.)
We examined the lens and immediately discovered one of the concentric adjustment collars under the focus ring was very loose.
In fact, when we removed it for a closer look, we found it was broken in two pieces.
We replaced the adjustment collar and repeated our test. The corners wide open now look like this and there was no center flare.
It's not quite perfect. You can see the left upper corner is still a bit softer than the others, especially the vertical lines. With this particular lens we do expect near perfection in all 4 corners, but for most zoom lenses, getting the 4 corners like this is the best you can get. If you took pictures with this lens right now I think you'd find it just fine, but we'll see if we can't get it a little sharper in the corners.
Wide Aperture Prime: Canon 50mm f/1.2
This particular copy came back from rental and the customer told us he'd dropped it but it seemed OK and he had continued to use it for the rest of the rental period. It did seem OK when we first examined it and in all the images the center was fine. For those of you who have never shot this lens, trying to assess anything other than the center at f/1.2 is challenging -- you never know whether the field curvature, the very narrow depth of field, or an actual problem with the lens is causing off-center softness.
When we tested it, however, the corners looked like this (focus was on a corner for this view, not the center). The upper two corners, especially the left upper, aren't quite right. This is a great example of why careful alignment is so important. If someone took this shot hand-held, I would just assume they weren't quite lined up with the chart properly. We also watched on the video monitor while we manually focused from near, to in-focus, to far. The top never sharpened up.
One of the nice things about the 50mm f/1.2 is the front element is a centering element; just pop off the front ring, loosen three screws and you can slide the front element around to recenter it. In this case, we recentered while watching the image in live view on the monitor. That got us pretty close, then we made a few final tweaks with photos of each.
Let me emphasize, though, that an optically bad lens doesn't necessarily mean a simple optical adjustment is all that is needed. Decentering occurs with all sorts of damage to the lens. It just happened that this lens simply needed an optical adjustment. We ended up moving the front element about 0.5mm upwards and a smidge to the left, after which the test chart looked like this.
Two things to note with this lens. First, I didn't show centers because the 50mm f/1.2 is one of the rare lenses that always has just a bit of flare in the center. Not a large flare like 70-200mm lens above had, but always just a bit. So center flare is only helpful when the lens is way off.
Another point is you can notice the left upper corner is still just a bit softer, particularly in the verticals, than the other corners. This is the best adjustment we could obtain (more movement could sharpen up the these verticals, but only at the expense of another area). With really wide aperture prime lenses, and many zooms, this is not unusual. If we test a bunch of brand new copies, most will have this degree of difference in one corner or another. Could you notice it in a photograph? I seriously doubt it, even knowing it's there.
Finally, you might notice the color difference in the top and bottom images. These were taken with only our ambient, industrial frequency fluorescent lights. The color difference is from the fluorescent lights, not the lens.
Wide Angle Zoom: 16-35mm f/2.8 Mk II
Wide angle zooms, being retrofocus designs, seem to have a lot of alignment problems. They're also very difficult to test with pictures because there's just so much stuff in the frame of a 3-D picture and the wide angle makes lining the image up square to the sensor very critical.
The 16-35 is a lens we commonly end up doing in-house because it's one the service center seems to struggle with. This one had two trips to the service center and came back just fine at 16mm but unacceptable to us at 35mm. The center had a large amount of flare just off focus at 35mm. As with primes, wide-angle zooms often have a bit of flare when they're perfect, but this is way off base.
And the 4 corners looked like this at 35mm. The most obvious thing is the right side corners are softer than the left. Also the left verticals are a bit blurrier than the horizontals, while on the right side the horizontals are a bit worse than the verticals.
Still, this doesn't look as awful as the 70-200 f/2.8 we showed first, so I ran the lens run Imatest. Imatest makes the issue more obvious, but the simple test we did above certainly was enough to show the lens was out of optical alignment.
I'll also use this lens as an example of why you don't want to do optical adjustments at home. In order to reach the compensating elements for this lens you have to disassemble it, then partially reassemble it so you can remount it, but with pieces of the barrel and the main PCB left off.
You leave the PCB off so you can reach the rear centering group. (The three screws hold down a plate. Loosening the plate lets you slide the rear element around to center.)
You leave the front barrel and focusing ring off so you can reach the adjustment collars for the three groups of compensating elements inside the lens (two are shown, the third only shows at certain zoom ranges). For those of you with enquiring minds, you focus by using a small screwdriver to move the zoom key, the piece of metal in the slot between the two adjustment elements.
I won't bore you to tears with a blow-by-blow account of adjusting the lens at 35mm, which then threw off the optics at the wide end. Then adjusting the wide end and going back to readjust at 35mm because the wide end adjustments affected 35mm, finally making a minor compromise at the wide end to get things right in the middle of the zoom range, etc.
My point is simply that we did most of that adjusting using this chart and testing screen, with a very occasional Imatest image for confirmation of what we were doing. At the end of all that adjusting (about 3 hours total), this is what the corners looked like at 35mm (and throughout the rest of the zoom range, but those were fine when we started).
And here is the Imatest graph repeated after the final adjustment.
As before, Imatest certainly clarifies things and gives us a nice set of numbers that our database tells us is what we expect from a good copy of this lens, but the simple chart test clearly shows the difference between a good and bad copy.
This Home Test Isn't Perfect
I mentioned earlier that this test is very accurate in detecting bad lenses: when a lens looks bad on this test, it almost always has an optical problem. Very occasionally a lens looks OK on this test but does have a minor problem. It's rare, and the lenses are not awful, but it can happen.
I thought I'd show you an example of the kind of lens it might miss. This is a Canon 35mm f/1.4. In retrospect we probably should have caught it, but it's pretty subtle and I'm not surprised we missed it.
The four corners look acceptable, fairly evenly matched and all resolving well. The horizontal bars on the upper corners are a bit softer than the verticals, but I would consider them acceptable on a wide aperture prime.
However, there is a fair amount of flare in the center when the lens is just off focus. We probably should have caught this, but it's certainly not as bad as the examples above, and you do sometimes see a bit of flare on an optically good lens. I have the advantage of looking at this retrospectively, I certainly wouldn't give a tech a hard time for missing this one.
The strongest hint the lens wasn't OK was that the best focus on the center of the lens wasn't very sharp. I constantly say the center of the lens is the least sensitive indicator of an optical problem, but here is the exception to the rule. (We see 50-60 decentered lenses a month, and most months we don't see any lenses like this.)
Imatest shows the problem, but mostly because I know what MTF50 numbers are acceptable for this lens. The center should resolve better than this, and well, the highest resolution should be in the center, not at the bottom of the image. Even with Imatest my first thought was we'd had a technical problem or missed focus despite focus bracketing. But repeated attempts gave the same result.
Here are the Imatest results of this lens post adjustment.
I thought adding this last one was worthwhile, particularly because it should give you an idea of what a lens looks like right on the borderline of acceptable. This one did need some adjustment, but most lenses that look about like this one on our optical charts are perfectly fine.
Hopefully, for those of you interested in this kind of thing, these examples will give you a good starting place to do a little optical testing yourself. I strongly encourage you to set this up with some friends or a camera club. Looking at a few copies of a given lens this way will really help you see what is normal and what isn't.
If any of you are interested and will be at WPPI next week, we'll have a similar set up in our booth and would be happy to look at your lens with you.
Roger Cicala and Aaron Closz
If you read this post and thought, "Yeah, I know that," Midwest Camera Repair is looking for an experienced Lens and D-SLR tech, mainly to service Nikon Lenses; knowledge of Canon lens would be helpful but not required. Someone who has experience servicing high-end lens would be best. Most of our repairs are middle to high-end equipment to Professional customers. We will train on use of Nikon & Canon test and adjustment equipment. Our Nikon lens volume has more then doubled in the last year. Knowledge of D-SLR repair would also be helpful.
Midwest is a Nikon Authorized Service Center and one of only four authorized, trained and equipped to service the VR-Series lens. They are also authorized on Canon IS-Series lenses.
Send confidential resume to firstname.lastname@example.org