Thursday, January 10, 2008

DSLR lens testing - Canon EF 100-400mm f/4.5-5.6 L

All the bird photographers I know and those I've seen on line (DPReview, Fred Miranda, etc) seem to use Canon's 100-400mm f/4.5-5.6 L series lens. And there has been quite a bit of discussion about possible variability in build quality between samples. For this reason it became very important to me to test a new sample before I put it to use in the field. I wanted to use such a test to "spot check" the example I purchased to see if it needed to be returned for another.

As in previous tests, the numbers that follow the aperture are resolution in line pairs per millimeter (aka lines per millimeter). For this lens I tested both center (at some focal lengths) as well as the edge of the frame (in selected focal lengths). As in the 70-200mm f/4 L testing, as I got to the longer lengths I started to use the Canon 40D's Live View mode with 10x magnification to carefully focus on the USAF Resolution Test Chart.

One thing to note before we get to the numbers; I strapped this lens on the heaviest tripod I own, a Bogen 3033 set of legs on a Bogen #3057 head. This rig has firmly held a Folmer and Schwing 12x20 inch view camera in the field. It easily holds an 8x10 Deardorff quite solidly. Yet when I was attempting to focus in Live View mode it was very easy to tell that the 100-400L gets "shaky" when I barely touched the lens.

This illustrated very very clearly to me the need for careful technique when using the 100-400L in the field. IS may play an important role. High shutter speeds will also play an important role. And, finally, I may need to strap this lens down to the 3033/3057 Bogen and trigger the shutter with a remote release. All this in an attempt to get the highest possible resolution out of the Canon 40D/100-400L setup.

100-400L
(100mm) AF focused - center measurement only
f/4.5 58
f/5.6 65
f/8.0 65

(200mm) AF - center measurement only
f/5.0 58
f/5.6 58
f/8.0 58

(200mm) Live View, careful 10x focusing, center + edge measurement
f/5.0 65 82
f/5.6 65 82
f/8.0 65 82

(400mm) AF - center measurement only
f/5.6 58
f/8.0 52
f/11 46

(400mm) Live View, careful 10X focusing, center measurement only
f/5.6 58
f/8.0 56
f/11 46

At 100mm, the lens appears to test "OK". It's certainly not what the 70-200 f/4L measured at the same 114 inches from the target. I wasn't entirely thrilled. But I kept going, just to see what else this lens might have in store for me.

At 200mm I tested in both AF and Live View modes. I was also able to capture edge resolution performance readings. Here again, the lens tested "OK" against the 2D test chart. But then something caught my attention. When I read the edge performance of the Live View tests I realized that 82l/mm is outstanding! performance. I had to wonder what I'd done wrong. Perhaps the Live View focusing wasn't as accurate as I'd hoped. And, of course, there's the fact that at these higher magnifications, AF and manual focusing with the wee-Canon 40D is nothing like focusing a 120 format film camera. The brightness of the display as well as the distance to the subject must come into play.

It seems that the resolution drops off on my 100-400L at 400mm. But it's not far behind the 200mm center measurements on that lens when using either AF or Live View to focus this 100-400L. I wasn't sure what this meant, even after looking at DPReview's performance tests of this lens at 400mm and comparing it against the 400mm EF f/5.6 L. There was one last test I wanted to undertake before I headed into the field.

I took the nice and sharp 70-200L, set it at 200mm, and moved the tripod to 54 inches from the target. This made the subject size the same as the 100-400L at 400mm. I then Live View focused images. Testing f/5.6 on both lenses I put the results into an image viewer (Gimp or Canon's own software worked nicely in my tests) and observed the differences in apparent resolution at 200x.

The results are really quite interesting. The 100-400L was tested at 400mm and 114 inches from the target. The 70-200L was tested at 200mm and 54 inches from the target. In the following image, take a close look at the results. Resolution is really nice on the 100-400. Looks at the test bars for both the 70-200L and 100-400L at f/5.6. They look very nearly the same, don't they? Then look at how the performance degrades from f/5.6 thru f/11 on the 100-400L. This appears consistent with a few user observations I read on DPReview.


- Click on the following image to view it full size -


What's interesting to me is that when operating near the closest focus of either lens (6 feet for the 100-400, and slight farther but still near the close focusing limits of the 70-200L at roughly 4 feet) that performance between the supposedly less sharp 100-400L and the know to be brilliant 70-200 f/4L is identical! What this leads me to believe is that close focus performance of Canon's zooms might not be as good as their ability to resolve scenes at something greater than their close focus limits.


With this in mind, could it be that the 100-400L's performance is actually quite good at greater distances from the subject? To find out, I'm headed out into the field...


Wednesday, January 09, 2008

DSLR lens testing - Canon EF 70-200mm f/4 L

Shooting a Canon 40D, mounted on a heavy tripod, looking at the smallest resolvable block of information from a USAF Resolution Test Chart, at 114 inches from the target, I took a look at a 70-200mm f/4 L (non-IS) resolution. In general, I like this lens for it's size and weight. So I was looking forward to seeing just how sharp it could be.

The numbers from left to right after the test aperture are the resolution figures noted in Lines Per Millimeter (aka line pairs per mm - they're the same thing). My comments follow.

70-200mm f/4 L (non-IS)
(70mm) AF enabled
f/4.0 65 65 58
f/5.6 74 74 65
f/8.0 93 83 74

(100mm) AF enabled

f/4.0 73 73 65
f/5.6 73 73 73
f/8.0 73 73 73

(200mm) AF enabled, center measurement only
f/4.0 40
f/5.6 58
f/8.0 58

(200mm) Live View - careful focusing at 10x, center measurement only
f/4.0 82
f/5.6 72
f/8.0 72

For the first pass I let the camera autofocus on the USAF Resolution Test Chart. The resolution numbers you see here reflect this.

Looking at resolution at 70mm the lens performs very nicely. As some people feel, the resolution improves from f/4 through to f/8 as the lens is stopped down. At f/4 the numbers look good compared with the two previous tests of fixed focus lenses. By f/8 the 70mm resolution is the best I've seen so far. Simply brilliant.

Looking at resolution at 100mm the lens performs very consistently at all apertures all across the field. Very nice and even, but no improvement in resolution by stopping down. I measured this length specifically to compare with a soon to be acquired 100-400mm f/4.5-5.6 Canon EF-L.

Looking at the top end of the zoom range I performed two test passes. The first pass let the autofocus select the focus point. The resolution of the 2D subject (the USAF chart is flat, after all) is down from 100mm resolution. For a second pass I used the 40D's Live View and 10x magnification to hand focus the lens. I did this to see if there was any improvement to be had in the lens measured resolution performance, and there is. As you can see, the lens is just as sharp at 200mm as it is at 70mm, depending upon the aperture.

The second test pass taught me that the 40D's autofocus, while good, might not be the absolute best in all cases when shooting a 2D subject. I'm sure there are many factors which come into play with regards to the 40D's AF performance. I just happened to stumble upon on sample/observation. This is a good learning and will be applied to the 100-400L tests when I get there.

Please Note: Resolution tests should not be confused with Modulation Transfer Function testing. MTF looks at a lens' ability to preserve contrast from the original scene. Resolution tests look for the finest possible sharpness, independent of contrast. The two tests are somewhat related in that the human eye perceives resolution as contrast. But the nuance is important as many folks feel that MTF is the only valid optical test. It's not. There are many ways of looking at lens performance. These happen to be just two, and I have chosen the simpler to implement resolution test process.

Tuesday, January 08, 2008

DSLR lens testing - Canon EF 50mm f/1.8

A second lens that I tested during a big weather event. It was raining like crazy around here and the wind was of Hellenic proportions. I didn't feel bad about testing instead of going outside and shooting live subjects.

To get an idea of what my optics and camera would do in my situation, and ignoring the fact that MTF measures contrast reproduction and USAF resolution test charts measure resolution - leading to vast differences in how results are to be interpreted, understood, applied... on a Canon 40D mounted to a sturdy tripod and letting the AF do it's thing, measuring line pairs per mm (same as saying lines per mm), at a target 114inches from the CMOS sensor plane...

...three numbers: Center ~ Middle ~ Edge of the frame...


EF 50mm f/1.8 II

f/1.8 46 52 52
f/2.8 52 52 58
f/4.0 73 73 65
f/5.6 73 73 65

I see that the resolution is fairly flat across the field. By f/4.0 this seems like a sharp little lens. The performance at wider apertures might actually be better than whats recorded here, but for that kind of test I would use Live View and 10X magnification to carefully focus.

For $70USD new, this is really a nice lens!

Monday, January 07, 2008

DSLR lens testing - Canon EF 28mm f/2.8

Just got out of the basement from running tests of my various optics against a USAF resolution test chart. It's raining like crazy here and the wind is of Wizard of Oz proportions, so I didn't feel bad about testing instead of going outside and shooting live subjects.

If I have time in the future, I may write up something more complete. Until then, my words and calculations will have to do.

To get an idea of what my optics and camera would do in my situation, and ignoring the fact that MTF measures contrast reproduction and USAF resolution test charts measure resolution - leading to vast differences in how results are to be interpreted, understood, applied... on a Canon 40D mounted to a sturdy tripod and letting the AF do it's thing, measuring line pairs per mm (same as saying lines per mm), at a target 114inches from the CMOS sensor plane...

...three numbers: Center ~ Middle ~ Edge of the frame...

Canon EF 28mm f/2.8

f/2.8 82 65 52
f/4.0 82 65 58
f/5.6 82 65 65

Compared with other DSLR lenses I tested, this is a really nice result . The edges come up cleaner and sharper as the aperture is stopped down. This is a seemingly sharp little optic.