How much purple fringing is normal for this lens?

Manfr3dManfr3d Registered Users Posts: 2,008 Major grins
edited May 23, 2009 in Cameras
Hi, I was testing my new Canon 200mm/2.8 L II USM
and noticed the tendency for purple and sometimes
green lines on high contrast edges. The 85mm/1.8
I owned was much worse in this respect but none of
my other non L primes (24mm, 50mm, 100mm macro)
show this behavior ... considering that the 200mm
is an L lens, shouldn't it be visibly better than that?

Some sample images taken with 5D II at f2.8:

Full scene:
bike.jpg

100% crop:
bike-crop.jpg

Here is another one:

Full scene:
moto.jpg

100% crop:
moto-crop.jpg

Another one at 100%:
car-crop.jpg

The purple is sometimes even visible in the thumbnail images, I don't know
how to judge this. What do you think?
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston

Comments

  • swintonphotoswintonphoto Registered Users Posts: 1,664 Major grins
    edited May 22, 2009
    Well, purple fringing can occur more on FF sensor cameras. Lenses on FF have to be excellent to avoid some pitfalls such as this. While the lenses you mentioned are great "L" lenses, when these lenses were designed needs to be taken into account. The 85mm 1.8, and the 200mm 2.8 lenses are holdovers from the film days. Digital sensors are much more sensitive to the direction light hits each photosite. As a result things like purple fringing can occur, particularly closer to the edges of the frame. Most lenses designed now supposedly take this into account. Your examples are a little confusing because some of the fringing appears somewhat close to the center.
    This issue is part of why Olympus decided not to make a system that relied on their old OM lenses. According to Olympus:
    ""With digital SLR cameras, the image sensor takes
    the place of film, so it is no longer necessary to
    design camera bodies around existing film sizes.
    You can simply use the sensor size that best achieves
    the goals of digital SLR photography.
    "But you need to address one of the
    most fundamental differences between film
    and digital imaging, which is that silver-halide
    photographic film responds to light regardless of the angle
    at which the light strikes it,
    but a digital image sensor responds accurately only
    when the light strikes the image sensor head-on.
    With lenses designed for film cameras,
    light rays at the periphery are often angled too sharply
    for the image sensor to capture them properly.
    "Even our acclaimed OM system lenses could not
    deliver light rays straight onto every part of
    the image sensor, and we quickly realized that to do it right,
    we would have to develop new lenses specifically
    for digital SLR use. "The mount and lens need to be
    much larger than the image sensor to ensure that every area
    of the sensor receives light rays head-on.
    If you make the image sensor bigger,
    light rays at the periphery inevitably wind up striking
    the sensor at an angle."


    Check this link for more on this possibility: http://www.four-thirds.org/en/about/benefit.html

    I suspect the fact that these lenses were designed with film in mind, light direction might be some of what you are dealing with.
  • LiquidAirLiquidAir Registered Users Posts: 1,751 Major grins
    edited May 22, 2009
    I suspect the fact that these lenses were designed with film in mind, light direction might be some of what you are dealing with.

    What you describe is an issue with wide angle lenses, but rare in telephotos. I have a 200/2.8 and the rear objective is a long distance from film plane which puts a clear limit on the extremes of the angle of incidence.

    Personally, I have not noticed any issues with purple fringing when using my 200/2.8 on a 5D (mark 1), but I haven't looked for it either. When I get a chance, I'll go through my catalog and see if I can find any.
  • swintonphotoswintonphoto Registered Users Posts: 1,664 Major grins
    edited May 22, 2009
    LiquidAir wrote:
    What you describe is an issue with wide angle lenses, but rare in telephotos. I have a 200/2.8 and the rear objective is a long distance from film plane which puts a clear limit on the extremes of the angle of incidence.
    That is a good point. Well, I guess I am stumped then...
  • ziggy53ziggy53 Super Moderators Posts: 24,133 moderator
    edited May 22, 2009
    Purple Fringe may be caused by both a lens chroma aberration and by several sensor related phenomenon including oversaturation of the photo sites. I believe that is what is happening in these images.

    A circular polarizer could have helped reduce the extent of the specular highlights and the severity of the purple fringing. Of course, a PL filter has its own share of problems which might complicate the decision to use it.

    I use the following PhotoShop action which does seem to help reduce the visibility of the PF, but it cannot reverse the image degradation caused by the specular highlights themselves.

    http://www.whodunit.com/photos/Another_CA_Fix.atn

    (I have successfully tested the above action on PS7, PSCS2 and PSCS4, Windows versions.)

    Note that this action creates a layer which you may use to selectively target the problem after the action finishes.

    As far as why this only happens on the one lens it may be that the high contrast of this lens causes the problem to occur or become more visible.
    ziggy53
    Moderator of the Cameras and Accessories forums
  • Manfr3dManfr3d Registered Users Posts: 2,008 Major grins
    edited May 23, 2009
    That was in interesting read Jonathan. I will have a look at
    the PS action over the weekend. If I can reduce the CAs
    good enough it might be a solution.

    Here is an example of green lines on a highlight edge:

    Full scene:
    car2.jpg

    100% crop top of frame:
    car2-crop1.jpg

    100% crop top right:
    car2-crop2.jpg

    Again this was shot at f2.8.

    I investigated the matter further and noticed that another shot of the car above
    which I took with identical camera settings except for a slightly darker exposure
    shows much less green lines. In fact the histogram of the first shot (above)
    was slightly clipped to the right and the 2nd (not shown) was not. It looks like it is
    indeed the sensor that somehow amplifies the CA's in such a situation. It also
    looks to me as if he green lines show up only in highlight edges that lie behind
    the focal point and the purple ones show up only in front or on the focal point.
    (I used center AF in all images)
    “To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
    ― Edward Weston
  • PindyPindy Registered Users Posts: 1,089 Major grins
    edited May 23, 2009
    I'm not disturbed by your purple fringes. I own the 135L which, despite being one of the top 3 primes in my opinion, behaves similarly to your 200 f/2.8. A little lightroom and it's gone. Stopping down has helped as well.
Sign In or Register to comment.