I hate CA. I know you can remove it in Lightroom but sometimes, its not all removable without it affecting overall image quality in certain situations.
It's an eye sore for me and can ruin a photo to my eyes.
I know that you can avoid it by not shooting in harsh light with strong contrast areas and by stepping down the lens.
However, sometimes I want to shoot in harsh contrast and sometimes I want the DoF effect of a wide open aperture.
I've been reading however that UV filters such as the haze 2A and perhaps even circular polarizers might mitigate this effect. Is this true, bs, nobody really knows?
I suppose if the polarizer was reducing any sort of glare or perhaps darkening a sky it could in theory reduce the resulting CA from those bright sources simply because it darkened those sources to begin with.
I suspect the UV/Haze filter theory is from film days, many films did have a fair bit of UV sensitivity and of course lenses are not corrected well at those wavelengths so cutting the UV off would help with CA. Most all modern digital cameras have very little if any UV sensitivity so I suspect there wouldn't be any improvement with a UV filter.
Anyway, no experience on my part - just educated guessing. It would be easy for you to give it a try if you have either kind of filter handy.
Ed Sawyer wrote:
Purple fringing is an artifact of digital, rather than Chromatic abberation within the lens.
I know digital sensors are less forgiving with CA, but I don't believe that PF is purely related to digital. See the last section and reference #17: http://toothwalker.org/optics/chromatic.html
Also, if PF were purely an article of the digital sensor, then one wouldn't expect lenses with largely similar specifications (focal length, aperture) to show different amounts of PF. However, some lenses produce a lot of PF and others very little on the same scene, indicating it is a lens property (though perhaps made more visible on digital). To cut it off with a filter, one would probably need a filter with a significant visible color cast (e.g. a yellow filter); the purple is probably mostly from actual visible purple light, so a UV filter designed not to block visible wavelengths won't do much good.
AhamB wrote:
I know digital sensors are less forgiving with CA, but I don't believe that PF is purely related to digital. See the last section and reference #17: http://toothwalker.org/optics/chromatic.html
1+.
A lens that I own that demonstrates this phenomenon amply is the Nikkor 180/2.8 ED AIS. On film, the CA remained an issue despite the ED glass but appears more prominent on digital. Axial and lateral CA both compete for prominence on images and in certain light, contaminate the image throughout. Its an optical phenomenon; NX2 claims to be able to undo both now though I haven't used the software and haven't seen the output to comment on how the image the is affected once axial CA is defeated.
Well, since PF chromatic aberration is worst for the shortest wavelengths that a sensor is sensitive to (violet and/or ultra-violet light) I would expect a UV cut filter to help somewhat, yes. Additionally since a polarizer decreases the amount of "stray" light rays which contributes to bloom and haloing, I would expect it to help as well. The reason it's considered mostly to be a digital phenomenon is because most films are not nearly as sensitive to the V and UV portion of the spectrum (which is where PF is from) as digital sensors are. I also hear it's much worse on CCD sensors than N-MOS or C-MOS ones.
But true PF can all be removed in Adobe's Camera Raw or many other processors. Here's some links about PF and the removal thereof - either with filters or with software:
The trouble is that most of what we see as PF is really a combination of things.
I guess how much a UV filter will help will depend on the camera's own internal filter blocks and the sensor's sensitivity to UV. And how much a polarizer will help will depend on how susceptible the lens in question is to flare and ghosting.
FlyPenFly wrote:
I've been reading however that UV filters such as the haze 2A and perhaps even circular polarizers might mitigate this effect. Is this true, bs, nobody really knows?
Not implausible if the PF is due to one of the R/G/B-channels clipping. A polarizer (not only circular - any polarizer) will reduce the amount of light by filtering out some polarizations.
CA is because the focusplane differs for various wavelengths. It will be independent of exposure level, while PF due to clipping will disappear if you reduce exposure.
p.s. A 'circular' polarizer is just a linear polarizer followed by a 'quaterwave' plate to de-linearize the light for the benefit of the AF-sensor.
Hmmm, yeah, could be. I just went with what Wiki said. They've been found (by me) to be very wrong on others issues in the past so I guess this could be another one.
I wonder if it's correct to say that the PF on digital shows up as purple because it's less sensitive - and would look less saturated on film because of the higher sensitivity then? You know, because it would be like, overexposed and white-ish looking instead of that deep dark purple... <shrug>
EDIT:
What.. Where did your post go mpmendenhall? You deleted it?
Bifurcator wrote:
EDIT:
What.. Where did your post go mpmendenhall? You deleted it?
Oops, sorry, I realized I had mis-interpreted the figures in the film data sheet (I was looking at dye transmission and not sensitvity), so I deleted the post (I had hoped before anyone would see/respond to it) to have time to get better info.
Anyway, the main point of my previous deleted post: I thought film had higher (not lower) sensitivity to near-UV than digital, hence the actual need for UV filters to prevent UV-induced haze on film cameras (while UV filters are nothing more than mechanical protection on digital).
Bifurcator wrote:
I wonder if it's correct to say that the PF on digital shows up as purple because it's less sensitive - and would look less saturated on film because of the higher sensitivity then? You know, because it would be like, overexposed and white-ish looking instead of that deep dark purple... <shrug>
I guess some forms of purple fringing could be due to different sensitivity or clipping issues; this is especially suspect when the purple forms a very bright edge around a blown highlight. However, I have also seen PF in the form of larger-radius diffuse "glow" around highlights, where the purple (showing up against a dark background) is far below the clipping threshold of the sensor, in a range where digital can't have very different color response (or else the whole image would be purple). There are also lenses that produce white glow around highlights where others would produce purple; presumably, these are better corrected for the chromatic aberrations that cause the purple glow (and demonstrate that the digital sensor doesn't magically turn any glow around a highlight purple; only those that are already colored as a result of chromatic aberrations).
Ed Sawyer wrote:
Doubtful a filter would help. Purple fringing is an artifact of digital, rather than Chromatic abberation within the lens.
Shooting film would help, though. ;-)
it might help, but i've seen lots of purple fringing in shots on film with the cv 35/1.4. i also saw some in scans from film shot with the leica 350/4.8.
You'll need a very steep, and "early knee" filter to get the PF lowering effect you seek. Steep enough to seriously impediment the rendition of cold blue and purple.
UVK-02, or maybe a Schott GG420 would do the job (with some lenses).
What most people don't consider is that in a normal longitudinal CA correction you have an equal amount of misfocus as you approach 700nm (long-wave red). This kills contrast in many lenses, and it also makes most landscape (greenery) detail-rich pictures very unbalanced in the color transitions when you go from green-in-sun and green-in-shadow. The shadow green is contaminated by an awful lot of red, and that makes the shadows very bleak and low in detail contrast.
If you look at the APO-testing made by most serious telescope manufacturers, you can see why the effect exists, and how i looks when you go into detail. Typical secondary spectrum looks like a "U" shaped curve.
An "APO"-corrected lens has three zero-crossings in the correction, an "S"-curve, but can still show a lot of secondary colour:
A "Super-APO" corrected lens has an even higher order of correction, the equation this correction ends up in gives four zero-crossings. This makes a "W" shaped curve
(Images from Wikipedia)
Zeiss has a few words on the subject in their first CLN_MTF_Kurven paper, at the page second to last. The PF effect from the sensor proper (optics taken out of the equation is VERY low in CMOS sensors. So low that the diffraction from an F11 setting is basically covering it completely.
The reason the PF gets so much stronger close to a blown edge is that you lift the amplitude of the aberration. White (the blown part) can't get brighter, but the aberration can. You clip the cause, but not the effect. The effect is equally strong in all parts of the image, but most of the time image contrast is so low that it gets invisible. A really strong contrast edge, like a rooftop edging a blown sky can have a real-world contrast of 12-14Ev, or a brightness ratio of 1:4000 to 1:16000. When the signal is 16,000 times stronger than the dark part just next to it, even a very small aberration effect (in percent or -Ev) will show up as "visible" on the darker parts.
The diffusion effects of film combined with the Lorentian diffusion (normal types of reversal or negative film types that is, not archival density microfilm) can keep an MTF-50 up to 30lp/mm in red and maybe 70lp/mm in blue.
Normal MTF50 for the green layer in reversal film is 45-50lp/mm. This gives a 3600x2400 resolution frame on a 36x24mm negative - about the same resolution capability as a 9MP digital camera. Kodak and Fuji have film datasheets at their sites, even today.
But the important part is that film both diffuses resolution and compresses DR a lot, and that it diffuses differently in red and blue. Film behaves like if you took the red channel in a digital capture and ran a few pixel radiuses of Gaussian blur on it (without touching the others).
Add this to the fact that the three (sometimes more) layers of emulsion add up to a thickness of 60-90µm (less in some T-grain films, more in quite a few other types), and that the output DR is seriously constricted, and you have contained/minimized the problem.
Where did your post go mpmendenhall? You deleted it? 
