2nd copy of 17-40mm F4L, edge to edge inconsistency.

I tried two 17-40's , both were softer on the right side of the image than on the left on 2 different bodies, decided to go with the 16-35 instead, which seems to have better edge to edge sharpness on FF.

Have a look here:
http://www.pbase.com/j​ackcnd/super_wide_on_5​d

Jack_C wrote:
I tried two 17-40's , both were softer on the right side of the image than on the left, decided to go with the 16-35 instead, which seems to have better edge to edge sharpness on FF.

Have a look here:
http://www.pbase.com/j​ackcnd/super_wide_on_5​d

I checked out your samples. I liked the 24-70 ones the best. Not just because I'm waiting for that one to arrive either.

Put the 24-70's in for a baseline, its a classic. While 24 is wide on FF, not quite the perspective possible with 16 or 17mm.

Ok, I have to say I'm really starting to become despondent with Canon's QC now. This is the third 17-40mm F4L lens, at 17/F4 , top left and top right 100% crops:

Top left:

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

From a quality control standpoint, you may want to review your test procedure. You are using a kit lens as the control. Using $50 optics to evaluate $300 optics is like measuring the accuracy of calipers with a pull tape.

If you want to verify the focus of the 17-40, try placing a straight edge across the lens body (without lens cap or hood) and verify you are parallel to the subject surface. Even then, Canon allows themselves tolerances for perpedicularity and cylindricity of the body during assembly. I'm not entirely sure of their quality control process, but I AM sure that its highly unlikely Canon ships anything that hasn't passed their EOL (End of line) control plan.

Also consider that Canon produces products that meet the specifications for 90% of the potential market. There will always be individuals who have expectations and applications that exceed Canon's design specifications. Does that mean Canon has poor quality control? No. Does it mean Canon doesn't produce a quality product? Definitely not. It means that Canon, like every other manufacturer, isn't perfect.

I guess my soapbox here is that people need to realize that when a product doesn't meet their expectation, it doesn't automatically mean the manufacturer's quality control sucks.

The next question I have is why the left and right crops all appear to have noticeable differences in light and color. The background surface and paper even has a different color. When I have reproduced similar tests, my two crops are much closer in appearance.

CanonAndy

From my experience with my EF17-40L and Canon's actions regarding this lens, and the many similar stories I've heard from others, I'd have to conclude that Canon doesn't give a damn about individual users, they only care about selling product and making a profit.

gcams, you should really send their crappy product back and get something else. For me, I'm moving over to Nikon and giving up on Canon entirely.

I would have to challange that analogy. I appreciate what you're getting at, but I don't think it's invalid to compare a cheap and expensive lens, and expect the expensive lens to outperform the cheap lens (just as I would expect my calipers to be more accurate than my pull tape). All I'm saying is that in the same test scenarios, my kit lens shows consistent "softness" on both outer edges. In other words, it's symmetrical. The L lens is not. I don't think that is an unfair comparison to draw.

Regarding my test scenarios.. the above is one sample of several tests I carried out, under different lighting and different setups. I only ever post a sample of my tests to avoid cluttering the thread with hundreds of test images. I would never claim a problem existed on the basis of one test scenario. I shot over 50 test images with the new lens under different scenarios and all had the consistent issue of RHS softness..

Even deliberately setting the camera off-square to the subject in either direction, I cannot get rid of the RHS softness and again, I would expect the kit lens to exhibit the same softness if the camera was off-square. The advantage of shooting so many test scenarios is that a pattern has emerged, that pattern being it is *always* the RHS that is soft, and only on the L lens.

Going by the numbers of other users that have had issues, I think this raises questions about the standards of Canon's QC process itself, rather than whether it has or hasn't passed their QC. I believe all the lenses I've had HAVE passed their QC, and there in lies the problem!

Even if only 20% of the reported issues are genuine, it's still not a super record, particularly when it comes to these 17-40mm F4L's.

Of course if you're really confident in Canon's QA, I'd be happy to swap my 17-40mm F4L that shot the above images with your copy.

..snip..

Yes I admit, the lighting is not even over the entire test image in this example, however I have another examples where the lighting is much better on the RHS and still the softness problem is there on the RHS. I refer to my previous comments about trying multiple test scenarios with different lighting conditions of which this is only one example. This is a 100% reproducable problem with all the shots I've taken (although mainly noticeable at 17mm/F4 obviously).

You may have inflated expectations about QC processes. Typically, any tests applied to 100% of the production run are only the most basic (does the lens respond correctly to electrical signals?), and are highly automated. A more rigorous test of a product is more labor-intensive and is applied to a small fraction of the run, with the size of the sample inversely proportional to the time required for the tests. Think of new cars - the final QC check consists of running the eyeball over it for obvious flaws, and driving it off the line onto the truck. Very few are actually taken out to the test track for a full performance evaluation, and even fewer would be put through tests of, for instance, heater/A/C performance. This is in addition to component testing during the process, but the intensity of that testing is, again, inversely proportional to the sampling interval. And component testing only assures that the single component tested is within spec.

It's quite possible you do have a batch issue, if the sample interval for an extensive test is large enough.

I would be interested to see the results on a different DSLR body to confirm that the issue is lens related and not DSLR related. I'm really stumped, I would assume if L's were out and about like this pro's that use them daily would be hopping mad.

Yes it would be nice to test on another body just to be sure. I've emailed the supplier to see if they've had a chance to test the 2nd copy I returned earlier in the week, to see if they can reproduce the fault at their end but as yet, I haven't heard back from them.

This is VERY scary .
Can it be that you've stumbled into a batch that is Asymmetric around its center lens Axis ?

My 45K car pulls to the right when driving down the road. Is it the car or the road?

Testing every lens you have on a 3478 megapixel camera will just drive you insane. I had a focus issue on my old 300D and fixed it myself through some very thourough testing.

You've gone through 3 17-40s. At what point did you think Canon's QC was perfect on the body? At what point did you deduce your test was perfect?

The scientific process must be adhered to. It must be taken to its' end result before spouting off results.

Pick another test, use another body. Or just take pictures with it. Obsessing over lens sharpness will drive you to financial ruin as you bounce between lens systems. Psychologically, once you've spent enough money you will see placebo like results.

My best friend is like this with his motorbike. We have the same make, model, year and colour. He puts different oil in it and feels a giant difference. He changes one little thing and the bike is transformed! I feel nothing.

My take on this is get to the bottom of it fast. Posting here and worrying others doesn't help.

BTW - My 17-40 is fine. I hope you eventually get real world results that please you.

Excellent! Wanna swap??

Hmmmm.. thought not!

Try checking the tire pressures?

First of all you do need a more accurate test
However if you are still sure with a better test subject that the problems there, + that the Focal plane is parallell, mirror locked up, self timer etc and you still get clear evidence that the lens is dramatically inconsistant then don't be afraid to keep going until you get a good un'

(there is a BIg difference between those that niggle at every tiny minutae on their bike and those that have one that pulls to the left on each one they get.)

The reason that I sympathise is that this happened to me with exactly the same lens- and there is no excuse for it from Canon as far as I'm concerned

There may very well have been a dud batch - I certainly found two of them. Like yours they were VERY OOF down one edge. My third was a good un ( VERY obviously not got the same problem) but if it wasn't I'd have kept going or got a different lens altogether

Maybe he expects a little more quality out of an expensive lens, I say he/we all deserve it. Been down this street, it's not fun! cut the guy some slack!