Vignetting problem on EF 24-105/f4L IS

I have used this new lens for over two weeks on an EOS 20D with

satisfactory results. Yesterday I mount it on my new 5D and took a

test shot wide open (24mm at f/4). When I examined this picture on my

computer, not only were all four corners vignetted, the left and right

sides were darkened as well.

 

I am puzzled over such poor performance of this new US$1250 "L" lens

at the edges. I am wondering whether the old EF 24-70/f2.8L would fair

better.

 

I took a second shot at my floor tiles. This time not only were there

vignetting around the corners, but straight lines were converging like

my EF 15/f2.8 fish eye lens. Is this suppose to happen at 24mm? I am

considering returning this lens to Canon in exchange for the EF

24-70/f2.8L. Such performance on a full frame sensor is simply not

acceptable.<div></div>

I posted test results for this lens on here about two weeks ago from the French photomag Chasseur d'Images. They said while the lens is exceptionally sharp at all apetures and focal lenghts, vignettage and distortion were majot issues, especially at the short end wide open. You got exactly what they found out on their high-tech test bench. Distortion is almost 1.5% barrel, what you'd expect from a kit lens and it shows up in your photo just like the dark corners show up. In this kind of situation, bump up the ISO and close down the lens by two or three stops each. Of course that won't help the distortion but at least the corners will be clean.

Hi Julius

 

Did you have the lens hood and any filters mounted? If so, is the vigenetting still as severe with them removed?

Thanks Joe for the advice. No I have not read that French report. And Cliff, I did not use any filters nor any lens hoods. I double checked my result by shooting at floor tiles but the result was just as disappointing.

 

However, when I mounted my EF16-35/f2.8L to shoot at 16mm, the vignetting and distortion are similar. So I guess this is a typical Canon problem at the wide end which has escaped my attention due to the fact that I have been shooting primarily on the 10D/20D APS-C sized reduced frame. The edges of my "L" lenses were rarely used. Hence I had never noticed this problem.

 

I have heard in the past that Canon's wide angle lenses leave much to be desired. Now I know why. I should perhaps heed my friend's advice of buying the Contax 21mm lens instead.<div></div>

Hmm, that's bad on such an expensive lens...

I am attaching another shot of the same floor tiles for comparison.<div></div>

Maybe Canon wishes to encourage people to use 1.6X and 1.3X bodies?

 

:-(

 

Happy shooting,

Yakim.

I can hardly agree, Yakim. With the arrival of the EOS 5D, FF is Canon's advantage over all competition. As such I just cannot understand the vignetting problem I discovered by chance.

 

Since my couriosity has been aroused, I tried a couple of more EF lenses on my 5D. Sure enough the same vignetting of corners appeared on the EF50/f1.4 standard lens and the legendary EF135/f2L portrait lens.

 

My observation has been confirmed by a friend who owns the 1Ds2. Funny none of us noticed the vignetting problem on "L" lenses back in the dark ages of film. His explanation is that perhaps it has something to do with how digital sensors collect light. Thus the problem has become more pronounced after we switched to digital camera bodies.

 

Perhaps the great Bob Atkins has a more thorough scientific explanation.<div></div>

and is this only the 5D sensor, or is it the same story with 1Ds and even tele lenses vignetting?

Most lenses vignette wide open. Perhaps it is just visible on 5D more beause people use more these new contrasty and saturated picture styles that amplify vignetting. Take 2 shots, one in "neutral" mode and another in "landscape" mode, and I'm sure "landscape" shows more vignetting. It is the same on film cameras with different films.

Also, your sample shots are underexposed, that also makes vignetting more visible.

You are bound to get light fall-off at 16mm, but at 135mm - I would be very concerned. There's something wrong here. I find it hard to believe that this normal for the 5D! I think I would have Canon ckeck out that 5D if it were mine.

I think I've read a review somewhere that says that this lens is an exceptionally good performing 35-70 f/4 - go outside those parameters, and you face compromises, especially at the wide end as you have found. I do think it would be interesting if someone managed to test the vignetting issue on FILM, because this would give some indication as to the extent that the problem is lens or sensor optics induced. I would not be at all surprised if some noticeable differences occurred.

Roison, all 4 shots that I have uploaded on this thread was shot using my new 5D. I don't have an 1Ds2. Only my friend has it. He is the one who actually alerted me that this vignetting problem is prevalent on FF DSLRs. He even sent me a shot of sky using his EF135/f2L on his 1Ds2. I became more alarmed and managed to duplicate the same problem shot using my 5D on my EF135/f2L.

 

Ahto, I will try your suggestions tomorrow when the sun comes out. My shots were taken using evaluative metering. I added +1 exposure compensation for the floor tiles and adjusted the levels on PhotoShop, taking away only that part of the histogram where there is no data. I did not want to blow out the highlights. My tests were not meant to make Canon gear look bad. Heck I ONLY own Canon SLR gear. I just wanted to share my discovery that even "L" lenses show excessive vignetting when shot using a 5D. Perhaps this problem existed previously in the film era but I have not noticed it. I am not about to retest them using film.

 

Cliff, assuming my 5D is not functioning properly (which I highly doubt), how do you explain my friend KY's vignetting problem shot using his 1Ds2 and EF135/f2L? You may view his test shot at http://www.ec-photo.com/341421-1/135+2L.jpg

 

Mark, I concur your statement. However, the issue has evolved to be more than just vignetting on the EF24-105/f4L IS lens. It is now on all EF lenses when shot through FF DSLR bodies. I agree someone should repeat these tests using film and compare the results thus far with digital.

wow. that is just totally unacceptable from a so called L lens. thanks for the heads up.

"I think I've read a review somewhere that says that this lens is an exceptionally good performing 35-70 f/4"

 

I think I saw a pile of comments like this on the Fred Miranda forum.

 

My personal opinion is that the 4.3x zoom was a stretch. . . .and it shows. For the cost of this lens. . .I would expect premium quality throughout the range. If you can't do that with a 4.3x zoom -> then make it a 2.9x zoom like the 24-70/2.8L. If I want prosumer performance. . .or a lens only usable in bright conditions at F8 . . .then I would have already bought a 28-135/IS.

 

I think most of us who were waiting for a 24-XX/4L expected top quality across the range, but were expecting a compromise in the form of a reduced zoom factor. (ie, a 24-70/4L).

I'm not sure the 24-105mm is to blame here as these effects were observed with a range of lenses

These shots are all underexposed, in ACR they would need at least a +1 stop of exposure, maybe +1.75. In that the vignetting is heavily exaggarated by the exposure. The floor tiles shot is probably close the the lens's closest ability to focus, at such close distances the distortion is also heavily accentuated.

 

I'm not making excuses for the lenses, the distortion isn't great on the top photo, but although I haven't got mine yet to test against the 24-70L, I wouldn't be surprised if the distortion is similar. The vignetting is annoying even if it is very easily fixed in ACR. I hope PTlens get some profiles for the 5D/24-105L soon and then I'll quit worrying about the lens and start hating how slow my computer runs while batching 300 files through PTLens!

 

This lens was never to be as good as the 24-70L in every way, I think that for it's main clientele (not landscape/portrait/architecture) it is a pretty good compromise through those focal lengths. For me it will be a stellar wedding lens and useful street/PJ lens.

Lest everyone starts to believe that all pictures made with this lens and camera combination

at f/4 are unusable, have a look at different situations first - like the attached image.

<p>Know the limitations of any machine you buy, <u>before</u> you buy. <p>I hope

that anyone who's unhappy with the 24-105 actually takes it back to the store, not just think

about it. That'll help drive the price down for fans like me. The word limitions starts with "L"

too.<div></div>

Another example...<div></div>

here...<div></div>

ALL - no exceptions - wide zooms and other primes too - have some degree of vignetting. Because your shots are greatly UNDEREXPOSED such problem is HUGELY (falsely) amplified. Get a CORRECT exposure and watch the vignetting almost disappear.

Also, on the floor samples, it seems the lens was *not* perpendicular to the floor as the amount of distortion is different on the top. Any wide lens would do that. If the lens had been perpendicular your feet or the tripod's would show (lest you reversed your column).

How would you then explain 135 mm lens vignetting?

It is interesting to note that even one of the sample images posted by Canon Japan shows significant vignetting issues (lens EF17-40mm f/4 L, at f/8)

 

http://web.canon.jp/Imaging/eos5d/html/eos5d_sample_4e.html