[IanCD]

Some reviews refer to chromatic aberration being "easy to correct in post-processing"

1. Just to be sure: is this chromatic aberration? It's 100% crop, AF-Nikkor 85mm 1.8D, f1.8, 1/3200sec



2. This was a raw (.nef) file. I've used Lightroom (LR 3.4.1), with the Adobe Profile for that lens, default settings.



I don't see much (if any) difference when I change the CA settings... the purple fringe stays there.



Q1: Is the auto lens profile adjustment what's being referred to as "correcting in post-processing" in the lens tests, or is there something else (straightforward & simple) I need to do to get rid of that purple fringe, without detailed editing in Photoshop (CS5)?



Q2: The lens test suggests CA should be very low with this lens, even wide open. This seems surprising to me - is it?

[popo]

That looks like longitudinal CA to me.



There are two types of CA, lateral and longitudinal.



Lateral CA is caused by different magnification of different colours. It gets worse as you go away from the centre of the image. That is the "easy" one to correct as it is a matter of applying different magnification to the colour channels of the image.



Longitudinal CA is caused by the focus planes of each colour being at a different distance and not a single plane. This commonly results in purple and blue-green fringing which can happen anywhere in frame, depending on if it is in front or behind the focal plane. This is the hard one to process out. It does reduce in strength as the aperture is closed. You might be able to get away with a global selective colour desaturation as a quick fix, assuming there are no other strong "purple" areas in the image.

[Brightcolours]

1. : No, this is not CA. What you are seeing is purple fringing (PF). Closing down the aperture, or shooting less contrasty scenes will "cure" it.



Q1: No lens profile will "correct" the PF. Either avoid it in manners stated above, or minimize its impact by desaturating the purple colour or selective replacing it with other colours.



Q2: PF is not CA. The PF you see only occurs when a LOT of light hist the sensor. When stopping down the lens, less light will hit it, and less PF will occur on edges around bright areas. When there are no such bright areas, PF will not occur.



CA is something optical, nothing to do with the brightness of the light. It will occur all over the scene toward the edges, but be most visible in contrasty edges. The CA of that lens is quite low, but PF wide open can be high.

[PuxaVida]

I believe this is CA, but not caused by the lens you used, it is caused by the microlenses on the sensor. Therefore it is called "purple fringe" not to be mixed by LoCA or LaCA.



In such scenes with high contrast and with the light coming from behind, it is visable if the shot is over-exposed. Allowing less light will cure the problem. Or during post processing, you can try to adjust the Hue/Saturation/Lightness slides (by selecting the correct color channel).



PS: some say that using an UV filter might avoid PF (by not allowing the shorter wavelengths), but I didn't try it.



Serkan

[Dick England]

popo is right. Brightcolours and PuxaVida are not. Note the green fringe behind the plane of focus.

[mst]

[quote name='Dick England' timestamp='1308132470' post='9246']

popo is right. Brightcolours and PuxaVida are not. Note the green fringe behind the plane of focus.

[/quote]



The green fringing is bokeh fringing (loCA), the purple fringing is not.



-- Markus

[mst]

[quote name='PuxaVida' timestamp='1308129486' post='9245']

some say that using an UV filter might avoid PF (by not allowing the shorter wavelengths)

[/quote]



Honestly: that sounds like utter nonsense to me.



-- Markus

[PuxaVida]

[quote name='Dick England' timestamp='1308132470' post='9246']

popo is right. Brightcolours and PuxaVida are not. Note the green fringe behind the plane of focus.

[/quote]



Hmmmm... I' d rather focus on the "in focus" part of the image when I make comments for CAs...

[PuxaVida]

[quote name='mst' timestamp='1308133240' post='9248']

Honestly: that sounds like utter nonsense to me.



-- Markus

[/quote]



Really? It would be good to know if you could give a bit more detail Markus...



Serkan

[popo]

I really don't get why LoCA is ruled out as a source of this type of purple fringing.



Referencing figure 5 at http://toothwalker.org/optics/chromatic.html I suggest the following thought experiment which could apply in achromatic, apochromatic, or superachromatic cases:



At the nominal focal point we supposedly optimise the luminance (green) for best focus. The red and blue will then be slightly defocused. Red and blue combines to give purple.



At a point in front of or behind the nominal focal point, then the red/blue channels will be either more or less defocused than green. This gives purple or green-ish fringes depending on the direction. There may be further colour skew depending on the degree of colour correction and the spectral composition of the light source itself.



So why don't we see them everywhere all the time? I believe this is because the correction is fairly decent, such that the fringe would be subtle enough not to be detected under most normal circumstances. Only in case of very bright areas, such as blown highlights, does it get strong enough to become visible in neighbouring darker areas. The key point here is that this does provide a mechanism for purple fringing to occur at the nominal focal plane as well as off it.



Note in the above, I am not ruling out the possibility for there to be other mechanisms for producing purple fringing, but this one seems to fit best and I haven't heard of anything else that explains it as well.



The UV/IR one is interesting but needs some more thought. I think UV is unlikely to be a cause as modern digital cameras and lenses are very poor at passing it. Do look up UV photography and see how hard it is to get an image even if you want to. IR is more possible, after all that is why they put the IR cut filter on the sensor. This mostly ends up in the red channel of the sensor though. Due to the longer wavelength, it would likely have a different focal point than for visible light but I'm not sure that would result in purple necessarily, my gut feeling is it would be more red than purple.