Long CA effect on apparent color

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

ray_parkhurst
Posts: 2750
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Long CA effect on apparent color

Post by ray_parkhurst »

I've noticed some lenses have a strong Long CA characteristic that affects the apparent color of the image. I took some pics of a 1921 Morgan Dollar to show the effect. The phenomenon is most visible on highlights at 100% so I cropped the cornflowers (near the center of the coin) out of 3 images: Focused Just Right; Focused High; and Focused Low. This is a problem because it is not correctable in post-processing, unlike Lateral CA, which can be corrected if you have the lens data. Note there is VERY little difference in focus between these, basically just a nudge high and low from the critical focus point.

Several lenses I own show this effect. The color shift on some is significant enough that it changes the effective color temp / white balance.

Am I interpreting these results correctly?

Focused Just Right
Image

Focused High
Image

Focused Low
Image

SONYNUT
Posts: 635
Joined: Sat Jan 22, 2011 2:27 pm
Location: Minnesota USA

Post by SONYNUT »

TRY LESS LIGHT...
..............................................................................
Just shoot it......

g4lab
Posts: 1434
Joined: Fri May 23, 2008 11:07 am

Post by g4lab »

Less light is probably a good starting suggestion. I am not certain what you are seeing is longitudinal CA. You can see these color effects through lots of different microscopes, especially when looking at coins. And other metallinc surfaces. I think it may be some kind of surface diffraction effect.

I don't know for sure.

mgoodm3
Posts: 273
Joined: Mon Sep 08, 2008 8:50 am
Location: Southern OR

Post by mgoodm3 »

I can certainly agree that long CA can affect the color seen on highlights and edges. In more complex areas you will probably see a little too much of the color in the fringe.

In homogeneous areas it should all even out.

I will also depend upon how much depth you have in the image. A stacked image will likely not have the same problems (or at least considerably less).

ray_parkhurst
Posts: 2750
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

I am assuming this is due to Long CA since the Apochromatic lenses I own don't do this at all. I'm also not sure why reducing light level will help. The entire coin is shifted in color, not just the highlights. It's just most obvious on the highlights...Ray

rjlittlefield
Site Admin
Posts: 21028
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Post by rjlittlefield »

I haven't studied this particular effect, but here is a possible explanation.

What longitudinal CA does is to spread light from highlights into surrounding darker areas. The color of the light that gets spread depends on whether you're focused in front or behind. In a linear world, this would not affect overall color balance because all the light ends up somewhere. But the transfer function from photons to pixel values is not linear -- brights get compressed and in the extreme case they get clipped. So the light that gets spread "counts more" than the light that stays focused. Spreading green doesn't just produce green edges on highlights; it produces an overall shift toward green as well.

Run the experiment with a smooth gray card, and I'll bet the effect is a lot less. Adding more diffusion to minimize the brightness of highlights should also help. Reducing illumination or exposure could help if it keeps highlights in more linear parts of the transfer curve.

Just some speculation...

--Rik

ray_parkhurst
Posts: 2750
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

I ran the same test using same lens on an index card (did not WB). Here are the results:

Focused High
Image

Focused Low
Image

As you guys predicted, the differences are far less. Looking at the histograms you can see differences going in the same direction as the coin photo (red shift when focused below critical, and blue shift when focused above) but with the more homogeneous surface it all "evens out" as Mark put it.

Very interesting. Thanks for the explanations...Ray

ChrisLilley
Posts: 674
Joined: Sat May 01, 2010 6:12 am
Location: Nice, France (I'm British)

Post by ChrisLilley »

SONYNUT wrote:TRY LESS LIGHT...
That isn't going to change the axial chromatic aberration of the lens in the slightest.

ChrisLilley
Posts: 674
Joined: Sat May 01, 2010 6:12 am
Location: Nice, France (I'm British)

Re: Long CA effect on apparent color

Post by ChrisLilley »

ray_parkhurst wrote:I've noticed some lenses have a strong Long CA characteristic that affects the apparent color of the image.
Yes, it does.
ray_parkhurst wrote:Several lenses I own show this effect. The color shift on some is significant enough that it changes the effective color temp / white balance.

Am I interpreting these results correctly?
Yes. The focus position is wavelength dependent (in all lenses) and so different wavelengths focus in different places. Clearly, this will affect scenes where the colour varies significantly from point to point. At each pixel, the image will be in perfect focus at one (single lens), two (achromat) or three (apochromat) wavelength, and form a circle which is larger than the airy disk of a perfectly focussed point at other wavelengths.

If that disk is larger than the size of a photosite it will spill onto adjoining photosites with two effects:

- the central photosite gets less of that wavelength, affecting the total spectrum and possibly influencing the perceived colour
- the adjoining photosites, in a Bayer array, have very different color filters so the colour is different. Even on a foveon sensor, light from one place is contributing to adjoining photosites and thus muddying the colours.

The question is clearly, to what degree does this affect the perceived image. This depends not primarily on how many zero crossings there are but rather on how large the deviations are between the crossings. It also depends on the size of the photosites relative to the sizes of blur circles at different wavelengths. Lastly, it depends on the contrast diffferences in the image and thus, the relative proportions of 'correct' light and 'spillover' light.

i is for this reason that false colours are most often noticed when ther e are bright specular hilights, because a small amount of a very bright light added to a large amount of a dim light yeilds a large colour difference.

Its is a mistake though to assume that axial CA affects only hilights or only high contrast transitions. It affects all pixels, to varying extents. Ths is why very well corrected lenses give richer, more vibranbt and jewel-like colours (on suitable source material) while less well corrected lenses give a blurred, somewhat desaturated result.

Also, having a different spectrum than the correct one does not necessarily result in a different perceived colour. For radiant sources, and for reflective sources if the illuminant is held constant, a multiplicity of spectra will result in the same CIE Lab value and thus the same percieved colour.

ChrisLilley
Posts: 674
Joined: Sat May 01, 2010 6:12 am
Location: Nice, France (I'm British)

Post by ChrisLilley »

ray_parkhurst wrote:I am assuming this is due to Long CA since the Apochromatic lenses I own don't do this at all.
They do do it, in fact. Don't get me wrong, I own apochromatic lenses and love using them but they are merely better, not perfect. They exhibit the same effect but to a lesser degree (and affecting different portions of the spectrum, so the characteristic false colours are different also.)

Post Reply Previous topicNext topic