Unsharp 28-70 f/2.8 L

Discussion in 'Canon EOS' started by stephen_delear, Sep 25, 2008.

  1. I'm having some problems with visible unsharpness (i.e. no point in tack focus) on a 28-70L on a Canon Rebel XSi
    when viewed at 100%. Photos were at f 2.8. Things seem a bit clearer at f 3.5. The lens has UV and CP filters
    I picked up at best buy on it (Tiffan I think). Shots were taken at either ISO 100 or ISO 200 at 1/125.

    I see a couple potential causes of the problem.
    1. CP filter (which was about $20) might have issues. These things use to go for $100, it's quite possible that
    what Best Buy is selling might be of a low enough quality to bring down the pictures.

    2) The lens may naturally not be sharp when you're talking about reviewing something at 72DPI, 50 inches tall.
    In this case the only thing to do is tripod mount and stop down.

    3) With the crop factor of 1.6 the lens at 70mm equaled about a 110mm lens. 1/125 may be to slow to handhold
    when large a file will be viewed at 100%. Some of the sharp images taken with this lens might have been set to

    4) Lens may need to go to Canon for repairs.

    I'll be doing some shooting after work to try to work out the issue. Are either #1 or #2 known issues?
  2. 1. Don't use cheap filters. Why put a $20 filter on a $1000 lens?

    2. If you feel the need to test a lens do it on a tripod to rule out a likely user error.

    3. Post samples.

    Are all your shots soft? are you shooting moving subjects? You could have a bad or broken lens but its more likely a
    mistake on your part.
  3. I suspect it's the filters. Why don't you test the lens without filter to eliminate or confirm that possibility.
  4. 5) you are shooting at f/2.8 and your DOF is so narrow.

    $20 for a CP filter is too cheap.

    100% viewing is another issue -- hwo do they look at natural size for posting to thwe Web, email, and printed?

    Are you shooting Raw and do you understand the weorkflow necessary? Often 2-3 sharpening steps are required for shooting digital. 1/125 should be fast enough provided you have good technique (and always use the lens hood, always).

    Have you used the much newer 24-70 2.8L in comparison?
  5. Are you using both filters at once? Probably not, but it kind of sounds that way from your description. If so,
    this could be a problem.

    Is the problem evident at all focal lengths? The 24-70 is prone to back-focusing at the wide end. I returned the
    first one I bought after confirming a back-focusing problem.

    Is the lens less sharp on this camera than others you are using?

    Most lenses are a bit less sharp at widest aperture, and the 24-70 is no exception, so it's not surprising it
    would look sharper at f3.5. Not really a defect, just the nature of the beast. BTW, I love my 24-70. It's a real
    workhorse of a lens and approaches prime lens IQ at moderate focal lengths and apertures.
  6. Is it a circular or linear polarizer? I believe you need a circular polarizer to focus properly. If it is a linear polarizer (usually cheaper) it may not autofocus accurately. It is most likely something simple. Good luck.
  7. I doubt it is the filter. I have never been able to see any difference in a shot when using different qualities of flter. (Exclide ND grads as they do vary a bit). This may sound daft but in a blind test if you can pick out the shot taken with a cheap filter of the same grade (eg multi-coated , polarising or whatever) then you have a better eye then me. 125th second is too long an exposure time for a handhelfd shot to give you a reliable indication of sharpness so I suggest you reshoot the test using a tripod and using delayed shutter release mode to minmise camera shake. Take off any flters to elimnate that question..When testing a lens I usually shoot at full aperture, f/5.6 and f11 to see how it varies with aperture. My 28-70 f/2.8 is pretty sharp throughpit its zoom range. It is also pretty good at full apreture. Just to give an indication of what you should be getting here is a recernt shot witj my 28-70 f2.8 and a 500-or-so pixel crop.According to the EXIF file this was taken at 70mm, at f5.0 and at 1/3200 at ISO 400..
  8. 1/125 handheld? I can't get really sharp handhelds at anything slower than 1/500, except an occasional lucky shot.
    Also, get rid of the UV filter. Digital sensor is not sensitive to UV, so is unnecessary. My 24-70 2.8L is my favorite,
    but like all lenses, wide open is least sharp and lowest DOF.

    You may also need to test the lens for AF with a tripod, lights, and a good pattern. Also, if your camera has live
    view, try manual focus while zoomed. Don't evaluate using camera LCD, but take picture, process, then view.. I
    usually have to view at 100% to get a good evaluation.
  9. There are a lot of good suggestions here.

    The only thing I will add is I did once "get what I paid for" in a cheap polarizer that was so bad I couldn't focus at all. That was an extreme case.

    I notice a huge difference in sharpness between hand-holding even a normal to wide lens at good shutter speeds vs a tripod.

    Remove all filters when you decide to test.
  10. I had a tiffen filter once, just a plain UV protector. It was on a different lens but I saw the results and never used it again.

    I keep a Hoya Pro 1 in my bag in case I'm going to shoot somewhere where I want to protect the front element from the enviroment (sand, rain, pints of lager etc) but even with that there is a notable degradation if you look close enough. The tiffen was ridiculous though, very obvious degradation even in 6x4 film prints, it was like the lens was in soft focus all the time.

    CP's though useful would normally have even more of an effect on IQ as they are two pieces of glass with a third, the polarising material sandwiched in the middle. I'd suggest not using filters at all until you get better ones and even then only when there is a good reason.
  11. If you suspect the stack of filters is compromising your lens sharpness: just take them off and do some more test shots.
  12. I don't want to hijack the thread and it's a bit beside the point. In any case: Has anyone seen a serious test of filters and filter effects as far as image quality degradation is concerned?
  13. "I don't want to hijack the thread and it's a bit beside the point. In any case: Has anyone seen a serious test of filters and
    filter effects as far as image quality degradation is concerned?" I havent seen a test like that, but I did take a shot a while
    back at the differences in light quality on UV filters. (so ill repost a sample). From top left to right, 1.Canon UV Filter, 2.
    Hoya UV Filter, 3. Hoya S-HMC UV Filter. (and the bottom pic the S-HMC is at the top).
  14. My guess is that the CP is the problem. As others have suggested, it's a simple matter to test the lens from a tripod without both the filters. A few years ago, I purchased quality CPs (about $100 each, same brand) for my 28-135 and my 100-400 L. The CP on the 28-135 function flawlessly. But the CP caused noticably soft images on the 100-400, even when using a tripod. Even though the lens would focus normally, all images with the CP were very soft. Looking through the VF, the image never "popped" when focus was achived. I never did discover what the problem was. Oddly, I recently obtained the 17-40 L, which accepts the same CP as the 100-400. The CP seems to work fine on the 17-40. Strange.
  15. I think for starters you should never stack filters. Especially on a wide angle zoom. All that extra low quality glass in
    front of the lens is asking to make matters worse. I only use a CP when needed. And its rare.

    I did notice a difference when I removed the UV from my 24-70. I saw a slight sharpness increase and more contrast.
    (Apparent sharpness)

    Wide angles are inherently softer than telephotos. Then take into account you are using an older version zoom wide
    angle. Adding 2 filters is asking for trouble. No need for UV. I leave my lens hood on and it seems to protect all
    needed in conjunction with cap. Besides, If my lens gets damaged with the hood on and cap, then a UV wouldnt
    help anyway. That means I dropped my lens from 15 feet straight down onto a nail pointing straight up to hit the
    front element. The UV wouldnt stand a chance. I'd just be out an additional $65. LOL

    One more thing, dont expect much detail and sharpness at all looking at 100% crop of a subject taken 15-20 feet
    away with a wide angle FOV. It will always have less fine detail than a close up. Its just the nature of it. The subject
    is just too far away for the sensor to register any fine detail. If it looks sharp viewed normal, let it be. Think about it, if
    youre looking at someone 15-20 feet away, you dont notice the skin texture,skin pours, peach fuzz that you'd notice
    if they were 5 feet away. So taking that picture at the same FOV wont record anything more than that really. Yes
    with a
    300 f2.8 for Head and Shoulders, but you are showing the sensor more. And with world class optics, not a wide
    angle zoom. If it does record more, be happy. But know
    one expects to see it.

    Look at Colins
    example. The focused subject is pretty far way as the picture was shot, now consider he was at 70mm on a wide
    angle zoom. Subject must have been 20+ feet away from him. Cant really ask for more than the 100% crop he
    showed,unless your shooting with a Phase One P65.
    Then you can ask all you want. I'd demand perfection too at $41,000. :)

    Moral of the story, view it at normal size, if it looks good, go with it. If it recorded what you saw with your eyes,
    great. Dont expect to see a flea tack sharp on someones shirt with a 20+ feet away FOV. Wont happen
  16. Actually a quality multi-coated circular polarizer for that lens is closer to $200.

    Almost certainly, the problem is the cheap filters you are using. Use them as coasters under your beer to protect the coffee table, not on your lens.
  17. Amen Alan. Now let me get a beer.
  18. Sigh. I remember back in the days of film where you NEVER took the UV filter off, under any conditions. I have stacked a UV and CP before but that was with a stardard UV and Cokin drop in CP. Anyway I'm linking in a full size shot. (Didn't get to run a lens test after work).
  19. Opps sorry. Full size image can be found at:
  20. I'll disagree with the issue being a cheap filter. Even a crappy UV filter doesn't do much to image sharpness. On
    flare, yes. Resolution, no.

    Look at the test images I posted in this thread (about 2/3 way down). Compares a $175 Heliopan filter, a $70 B+W
    filter, a totally beat up ugly Tiffen filter I got for free, and no filter.

  21. Spearhead

    Spearhead Moderator

    I'm having trouble understanding this. You are testing a lens hand-held and with a filter? What's the point of posting about that? If you think your lens has a sharpness problem, you eliminate all variables that could also cause sharpness. Then, if you still have a problem, you post. You don't post if you haven't done any kind of reasonably scientific test.
  22. 1. No lens is perfectly sharp at 100% crop on the screen.
    2. Some sharpening is necessary, either in-camera with jpg (not as good) or in post if you shoot RAW (better).
    3. Even excellent lenses generally will not be at their sharpest when you shoot them wide open. Not only is the lens
    generally less able to resolve fine detail (and there are other issues) but you have a much narrower depth of field, making
    perfect focus more critical.
    4. Leave the UV filter off. If you _must_ put on a "protective filter" make it the best one you can buy, and make it clear
    glass instead of UV. There is no functional value in UV filtering on your DSLR.
    5. Eliminate camera shake before you jump to conclusion about your lens.
  23. Can we maybe get some shots to better address the issues?
  24. My Canon 450D has gone back to the manufacturer for the third time. Focus is unreliable, something picked up in a recent magazine with the 1000D. Each time it goes back, Canon say they've found the problem and rectified it. Comes back, no improvement.
    Filters won't bring the quality down as stated, and 1/125 should be fine for sharp pictures. The worst lens on the market wil still not be so bad it looks unsharp.
    The proof of the pudding is a test chart. Use a tripod. Set a wide aperture, preferably full aperture. Tape the chart to the wall - this can even be a flattened sheet of newspaper. Move the camera randomly nearer or further from the chart between each shot. You will find randomness in the focusing if it is the same problem as mine.
    The real clincher is having a control. I used my three year old entry level Nikon D50, which had bang on focus every time.
    Canon have reproduced the problem using the above.

Share This Page