Cause of Fringing

canonloader wrote in post #5408354
Well, whatever the cause, it's fairly common or there wouldn't be tutorials on how to fix it.

Mitch:

Do you have any links to the tutorials?

Glenn NK wrote in post #5408617
Mitch:

Do you have any links to the tutorials?

Sure, here's the one I used and it did an excellent job of it in seconds.

I think that the "fix" is probably to throw out the shot with bad purple fringing or crop out the bad part, if possible. Also, learning how to minimize having shots with blown out areas might be another "fix".

Since Mitch mentioned that he has not seen purple fringing when using his 1D or 30D, I speculated that the sensor site center-to-center spacing along rows and columns might be a factor in the amount of bloom around blown out areas. For the 1D, the spacing is 0.0115 mm and for the 30D, it is 0.0064 mm (that is also the value for the 1Ds Mark III). For the XTi and 40D, the spacing is 0.0057 mm. The difference between 0.0064 and 0.0057 might be just enough to make the difference with respect to purple fringing showing up.

Yeah, that's generally my "fix", but someone posted a bird in the bird forum and asked once, how to fix it. I had just found that tutorial and gave it a try. Boy was I surprised. As the tute says, the purple is a narrow band and easy for PS to separate, then you can just change it. It worked very well.

The tute says it also happens with film, but I don't recall it from my film days. That has to be a malfunction of the lens coating or glass, no sensor to screw up.

canonloader wrote in post #5408704
Sure, here's the one I used and it did an excellent job of it in seconds.

Yikes, that example image has serious purple fringing! That is interesting in that the author says that a UV filter makes a difference. Sounds more and more like a lens thing.

I think that I know of a possible reason that I never saw purple fringing with film -- all of my prints were only 4 X 6 inches and when slides were projected across a room that was not very dark onto a cheap screen, little details like that would have gone unnoticed anyway.

It does, for sure, but I am using the same lens I used on my other models and never got any fringing. I will say though, it is a 14 year old EF 300/4L Non-IS version, so the coating is different since there were no digitals back then, as far as I know. I think I would have seen it crop up on the 30D and the 1D if it were happening.

Just remembered, I also have an old D30 and it's never cropped up on that either.

And it's not a huge problem. I have gotten a few series of shots that it appeared in and in all honesty, I would have deleted them anyway.

Do 4/3 cameras have purple fringing since they are near telecentric?

canonloader wrote in post #5408704
Sure, here's the one I used and it did an excellent job of it in seconds.

Thanks, I've bookmarked it.

It appears that this problem won't easily be solved, and like other difficult problems, I suspect the answer doesn't lie in one simple solution. More and more it seems to me to be a combination of lighting, lens, and sensor - which is alluded to in some of the links posted to date.

I have some strongly backlit images that don't seem to exhibit any fringing - taken with a 30D + 100 macro (which according to photozone, "CAs are very low and nothing to worry about").

canonloader wrote in post #5408704
Sure, here's the one I used and it did an excellent job of it in seconds.

I tried out the tip on the GBH with the purple striped stockings and it worked very nicely. Here are the reworked images:


The 48% crop:

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

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

Wow, that's quite dramatic. I only used it a few times but was pretty impressed with the results also.

bill boehme wrote in post #5409362
I tried out the tip on the GBH with the purple striped stockings and it worked very nicely. Here are the reworked images:

What does this technique do to shots that have objects with a lot of blue in them?

I don't have Photoshop so I can't test it myself...

kcbrown wrote in post #5410508
What does this technique do to shots that have objects with a lot of blue in them?

I don't have Photoshop so I can't test it myself...

Nothing really. There are a couple bracket sliders at the bottom of the Hue/Saturation menu that allow you to tweak the limits of the color selection range. If you do not do anything, the color range is very narrow. I found that after I made my selection, it changed the selection from blue to magenta since the actual color was closer to being magenta. I also found another tutorial that is essentially the same as what Mitch referenced except that it says to select Magentas rather than Blues. I edited one image where I made two different selections -- one for blue and the other for red. You sample the color with an eyedropper tool and then it selects a narrow range around that color -- it does not affect all of the blues or magentas, or reds in the image. I ran another test this evening about an hour before sunset. I shot straight at the sun through the canopy of a tree using my EF 70-200mm f/2.8L IS USM lens and guess what -- there actually was a fair amount of purple fringing in the area surrounding the sun. I guess that is no surprise. If you try hard enough, it is possible to take a bad picture. I just did it for the challenge of playing with the Hue/Saturation tool.

Was that with the XTi? Pretty much the same sensor as the 40D isn't it? I got purple fringe when shooting towards the sun, but not right at it. Lot's of contrasty stuff, but the sun wasn't actually in the picture.

canonloader wrote in post #5412480
Was that with the XTi? Pretty much the same sensor as the 40D isn't it? I got purple fringe when shooting towards the sun, but not right at it. Lot's of contrasty stuff, but the sun wasn't actually in the picture.

Yes, I used the XTi. However, the amount of purple fringing using the L lens was really quite small in comparison to the amount of fringing that I get with the 75-300 mm lens.

I did some more searching about purple fringing on my favorite astronomy web site, Cloudy Nights and came up with several articles, mostly related to reviews of refractor telescopes. After reading various articles, my uneducated opinion about purple fringing is that it could be the result of a combination of flare and spherical aberrations in the glass. Some refracting telescopes go to great pains to reduce flare by employing such things as aperture rings at various locations between the optics along with more conventional things like optical coatings and anti-reflective material lining the interior of the telescope tube. Of course, spherical aberrations affect the ability to get good focus.

Classical chromatic aberration is the result of the index of refraction in glass being a function of the wavelength of the light. However, the focal length is not greatly affected so the colored edges around contrasting edges is still close to being in-focus. On the other hand, purple fringing (which isn't necessarily "purple", but since guys can only name seven colors, that is what it is called) is quite blurry ... something that spherical aberration can influence. Anyway, the astronomy guys seem to have a much better handle on understanding optical performance than photographers do.

Since purple fringing can be seen by looking directly through optics, I think that the whole idea of it being related to digital sensors is a red herring.

I think your right. I at first thought it was an artifact of digital sensors, but then found out that it shows up in film too? No, that's got nothing to do with digital. I blame it mostly on cheap coatings, which follows, cheap coatings, cheap glass.