Fringe: Is it the lens, the camera, or both?

Question

There is clear evidence of chromatic aberation in the form of the dreaded purple fringe. I have placed a red dot at each end of the fringe where it appears in this photo, which is identical to "Osprey in Tower," immediately adjacent to this shot my photostream.

The sun was somewhat low in the sky. You can tell the direction of the light by the highlights on the bird's chest and by seeing which sides of the tower steel are lit.

Fringe appears at high contrast interfaces, so I'd like to know why it is on the lit edge of the steel in some spots and on the shaded side in others, including along the Osprey's back.

Would a better lens have the same problem under similar conditions or would the fringe go away?

f/5.6 is the maximum aperture for this 70-300 VR lens at the shooting length of 300 mm.

Thanks for you help.

Gosh, I'm sorry!!! I forgot to include the link. Thanks, Ben.

http://www.flickr.com/photos/samfeinstein/1198936287/

There is a pair of pictures of a boat in tests for the D200 and 30D on dpreview.com. The D200 shot is perfect and there is a lot of purple around the sunlit end of the boat in the 30D shot, so this problem exists with CMOS sensors as well.
30D: http://dpreview-img.fotki.com/gallery/canoneos30d_samples/originals/img_2074.jpg
D200: http://dpreview-img.fotki.com/gallery/nikond200_samples/originals/dsd_0250.jpg

I had thought it was the better lens (17-55) on the Nikon compared to the non-L glass on the Canon (17-85 EF-S - half the price) or maybe the sensor strategy, but now I realize that it may have been the time of day or the angle at which the photo was taken more than any other factors. (The light on the columns outside the window shows the difference in time.)

This requires more research and study on my part. If altering the angle of view relative to the light source will improve this, I'd certainly try to do it in the future.

I looked at 15-20 shots of that bird on the tower and I have a non-scientific observation. I say that, because I only looked at one set of images of one subject under nearly identical lighting conditions. I have observed that the fringing also occurs relative to the center of the frame and radial to it. The farther from the center you get, the worse it is. It is not on the lit side as opposed to the unlit side, either. It seems like it occurs at the edges that are "away from center." Maybe this makes sense, as a lense is spherical and the farther from the center you go, the greater the angle between the face of each piece of glass - usually. I'm sure that's part of the reason for using concave and convex glass for various elements within any given lens.

Answer 1

Hi Dr. Sam, lens 'quality' can have an effect, but your 70-300mm VR does have ED glass, which is supposed to minimize those aberrations. I have also had similar problems with this phenomenon, I shoot with a D80 and 70-300mm VR, and I have seen this in that same lens as well as my 50mm f/1.8, and the kit 18-135mm. I think this is a problem inherant to the nature of digital cameras. Something about the light that strikes the image sensor in differing wavelengths. Film is a lot more forgiving. Believe it or not there are fixes in photoshop, I read an interesting article about it in one of my photog mags and it had a tutorial on the fix. I'll have to find it and share it with you.

Answer 2

Dr. Sam,
It's possible that you may be dealing with two different phenomena.

Color fringing, or more properly later chromatic aberration, is the result of the inability of a lens to focus all wavelengths of light in the same plane. I'm sure you already knew this.

Unlike film, digital sensors don't have a "shoulder" in the highlight region of their sensitivity curve, so that when the intensity of light exceeds a certain value, the highlights are gone immediately without the sort of graceful overload that happens with film.

The problem with digital sensors, though, is that the three channels don't necessarily overload at the same value. Typically, yellow will overload first, followed by cyan and then magenta.

What this can mean is that sometimes, on the edges of spectacular highlights, the yellow channel will completely blow out slightly before the cyan and magenta channels. This can lead to colored fringes on the edges of spectacular highlights that actually look a lot like chromatic aberration.

So, I think that what you're probably seeing is a combination of sensor overload on the light side, and perhaps a small amount of secondary later chromatic aberration on the dark side that just happens to be about the same color. A better lens would probably fix the latter, while the only thing that could probably fix the former would be a Fuji DSLR with a Super-CCD.

In any case, though, this is all just a guess on my part without having seen the image in question.

Answer 3

What Ben H says makes a lot of sense and there are others that I guess have this same problem.

Dr. Sam, since you shot this on your Nikon D200 and I have also had a few shots like that on a model's skin but mine was more prominent and was more of a "purple" and I thought I was getting a cast coming from the red wine that she was holding in her hand. But when I looked I was wondering why it was on the other side also. I did this shot again with a different model but at a different time of day and hers came out without the "purple" fringe.

I would say it deals with the CCD that Nikon is using and the way the light hits our subject at a certain angle.

Tech info: Both photo sessions shot on my balcony and the sun coming from a SE direction Nikon D2Xs on a Nikon 80-200 2,8. Model standing outside and I was shooting from my living room so the sun was not hitting my lens.

Hope this also gives you answers.
Kevin

Answer 4

I can't add anything to what Ben said.

If you shot RAW, you can try recovering the information from blown out channels in Photoshop CS3 using recovery in ACR. That can help reduce it, but not usually eliminate it. It can be 'fixed', but it is a lot of work.

Vance

Answer 5

Thank you Ben and DR.

cant do better than Ben so wont try, as he says better lens would sort one of the issues, shame i cant comment further with out seeing the shot.

nice question thanks

EDIT: just saw the link, yeah thats digital, who says films dead eh? better lens yes is the answer, or film

a