EF 70-300 L: CA Bokeh and Focus area

I took these 2 pictures with similar composition but with different focus area (one is focused on the wall, the other on the weed/thorn)
Interestingly I found the following:
Out of focus area with high contrast creates a purple CA in combination with blurring of the small details – i.e. A bokeh with heavy CA

Look at the 1st picture weed on top of the wall weed looks ok (just a bit of CA) & see CA on the thorn to the right – the opposite is happening on the second picture

See next post for crop details
Note: the 70-300 L has generally small CA (see photo zone quote “.. The average CA pixel width remains well below 1 pixel throughout the range…”)
I wonder if it was discussed before (I also assume it is not unique to a specific lens)

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

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

see crop details (~100%)

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

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

Coloured bokeh is quite common on fast lenses - particularly those that don't use ED glass.
Since it's relatively slow and does use ED glass, I'm a little surprised to see it on the 70-300L.

noiseJammer - you have wonderful work on your flickr

did you see same CA on any of your lenses
why is it more commom on fast glass (you are a physicist after all)
- i was surprised by the effect of the focus on the CA

noisejammer wrote in post #13220718
Coloured bokeh is quite common on fast lenses - particularly those that don't use ED glass.
Since it's relatively slow and does use ED glass, I'm a little surprised to see it on the 70-300L.

That's not really that bad. The Tamron 70-300 VC is worse and believe it or not the 300 f/4 non-IS is much worse in that regard still.

Anyway what he sees makes perfect sense since he is looking at font/back longitudinal CA (as opposed to lateral side to side CA which gets worse near the edges and to which his 1 pixel quote refers). I actually find the lateral CA near 70mm to be where the 70-300L is a bit on the weak side moreso than with the longitudinal CA where I have seen worse.

omer wrote in post #13220962
noiseJammer - you have wonderful work on your flickr

did you see same CA on any of your lenses
why is it more commom on fast glass (you are a physicist after all)
- i was surprised by the effect of the focus on the CA

Longitudinal CA is caused by the different wavelengths not all being focused to the same plane of focus. Most lenses only make one correction where they bring blue up to meet red so you get blue/red focused on one plane and then green slightly behind and thus if you go to live view and focuses on a subject that tends to make longitudinal CA easy to see and slowly roll focus from front focused to back you see a purple (red+blue) haze shift to a green haze. Expensive full APO lenses also bring green to match plane of red and blue.

Fast lenses have very narrow exact planes of focus when used at or near wide open and it makes this effect easier to see (plus they are trickier to design in general).

Stopping down always makes longitudinal CA become less but in certain cases can make regular lateral CA get worse, it depends.

Lateral CA is the spreading of different wavelengths of light out across the frame in the plane parallel to the sensor, lateral to the line of focus while longitudinal is the spread perpendicular to the sensor plane.

Lateral CA gets worse to the edges since the lens has to bend the rays more there so whatever issues in bending the different wavelengths it has get magnified toward the edges. Shoot say a fine black line on white background near the edge: First order lateral CA makes redish fringes on one side of the line and greenish on the other. Second order lateral CA makes blueish fringes on one side and yellowish on the other.
They use special glass or fluorite crystals mixed with regular glass to try to bring them all back to the same point in fancier lenses.

Lenses that tend to show longitudinal CA:

70-200 2.8 IS original (definitely has a lot more than the 2.8 non-IS or f/4 IS)

300 f/4 non-IS (really has quite a lot of it and can get major areas of purple fringing and green bokeh hazing)

most fast primes, it seems the 50 1.4s and 85 1.8s tend to have a particular lot and 85 1.2 and the 135 2 certainly has it too


Lenses that seem to control it a bit better than average for type:
70-200 2.8 non-IS from canon
canon 60 macro
300 2.8 IS (almost entirely free of it, basically a real APO design)
tamron 17-50/28-75 (intense highlights of glint will show it for sure so hardly free of it but they are relatively free of it for non-glaring highlights while some other 2.8 zooms show it a lot for that too)


i dont think the 70-200 f/4 IS or even 70-300L are really all that bad when it comes to it, there are other 70-200 and 70/100/120-300 from canon and sigma that have more, although your samples demonstrate that it does have a bit more than I thought

24 1.4 II certainly has it but considering 1.4 which is super fast it doesn't seem quite as bad as it could be imo

Wombathorror - thanks for the clear explanation
fyi - i really like this lens but i happen to notice it (the out of focus CA) with the 2 shots
one thing you did not mention which is evident (from your explanation) is that out of focus areas will suffer more from color (wave length) variation and therefor we see a CA bokeh

Thanks for bringing this up Omer.
I get it too in very high contrast situations. But in most cases it is hardly noticeable, especially at regular sizes.

to me what was really new was the differences between the CA effect on in focus and out of focus areas (as evident in the two sample shots below)