My T1i + Canon 50 1.4 - TONS of axial CAs @ f/2

[Daniel Browning]

The purple fringing is not always axial CA. One particular sensor designer, RKM, shoots Canon and is pretty confident that cases such as this are a reflection between the microlenses and the filter stack, and should be solved by multi-coating the microlenses (rather than single AR coated, as they are now), but that's probably expensive.


[Edit] Oh yeah, and another camera designer said that purple fringing can also be caused by overload, and that disabling electronic first curtain can help. (In Canon DSLRs, electronic first curtain is controlled is the "silent mode" liveview.)

[Jon Ruyle]

The purple fringing is not always axial CA.

Thanks Daniel. I didn't know. But... are these other causes more apparent with fast lenses?


Anyhow, in a controlled test, it should be easy to tell axial CA from these others just by adjusting focus a bit, right?

[Daniel Browning]

But... are these other causes more apparent with fast lenses?

As I understand it, yes. Reflections from the microlens are a function of the AOI (angle of incidence), and fast lenses tend to have wider AOI, so worse PF (purple fringing). If pixel overload was the cause, f-number wouldn't affect it, just absolute exposure.


Another reason I forgot to mention is infrared contamination. From what I remember of the explanation given to me by a brilliant lens designer (J. Brian Caldwell), the Bayer color filters tend to have different IR rejection, so any IR that gets through is going to increase brightness in varying amounts, typically resulting in purple fringe. I would imagine that since lenses also don't tend to correct IR to the same plane of focus as visible light, that would also contribute to the issue.

Anyhow, in a controlled test, it should be easy to tell axial CA from these others just by adjusting focus a bit, right?

Yes. Another test is to see if it improves by stopping down -- that's definitely axial CA and not PF. (There is a rumor going around that axial CA does not improve by stopping down, but it definitely does.)

[mmodica]

As I understand it, axial CA improves from stopping down, but lateral CA doesn

[mmodica]

Oh, more about the testing:


It was definitely axial CA because it gradually got better as I stopped down from f/1.4 to f/5.6. It also appeared all over the frame, not just in the corners. However, I was disappointed by the sharpness delivered by the fifty. I would have expected it to be better than my zoom, especially at (slow) overlapping apertures).

[Daniel Browning]

I can

[Jon Ruyle]

Yes. Another test is to see if it improves by stopping down -- that's definitely axial CA and not PF.

But if PF from microlens reflections are a function of AOI, wouldn't they, too, improve when stopping down? (Probably not and I'm just being stupid)

[Daniel Browning]

But if PF from microlens reflections are a function of AOI, wouldn't they, too, improve when stopping down? (Probably not and I'm just being stupid)

Good point. [Edit] The only way that stopping down would be a good test is if the f-number has no relationship to the AOI, but I don't think that is ever the case. There may be some lens designs (reverse tele, I think), where it's at least non-linear, between AOI And f-number, but even then it will at least change by some amount, partially invalidating the test.