What can lens flare tell us?

[yessammassey]

Lens flare is something all photographers encounter from time to time, but what can we determine about a lens from looking at an especially well-defined lens flare?

I noticed this fine example in a recent image. Bronica PG 50mm.



I can count at least 10 different pentagonal flares in 7 groups beneath the central light. I believe this lens is specified as having 11 elements in 8 groups, and my guess would be that in 'perfect' flare-inducing conditions, one interpret the correct number of elements in then lens, based on the flare.

But what about the alignment of the elements? The third pentagon from the top in the series below the central light looks a little more skewed and off-axis compared to the others. It appears skewed and off-axis in a couple of other shots with overhead streetlights taken on the same night. Does this mean anything?

 

[markbarendt]

So, I think what you are seeing as skew in #3 is that it's shape is inverted, rotated 180. Maybe there is a difference between a flare on the back of a group v front?

Is you aperture 5 bladed?

 

[AgX]

I can count at least 10 different pentagonal flares in 7 groups beneath the central light. I believe this lens is specified as having 11 elements in 8 groups, and my guess would be that in 'perfect' flare-inducing conditions, one interpret the correct number of elements in then lens, based on the flare.

You mix up aperture-image with reflections of outer lights.
(And your way of counting is wrong too.)

The latter is used to establish the number of lens-groups.
Some say they can for sure establish by this the number of lens-elements too. I did not succeed so far.

The formula is n/2 = L
(with n=reflections, L=lens-groups)



When using the number of aperture-images for establishing a lens count the formula would be [(n/2)2+(n/2)]/2 .
But as above the problem of distinguishing between lens-group and -element would remain too.
And the sheer number of images would make the aperture-image method often impractical.

 

[LAG]

But what about the alignment of the elements?

It depends on the focal point and the focal distance of each lens.

Does this mean anything?

Multiple lens system (concave & convex related with object & image, positive and negative focal lenght) and the distance between lenses.

Also, you can appreciate different figures tonality ("pentagons" in your case: 5-blades), that tells you how each lens treat light at different distances.

Best

 

[yessammassey]

It depends on the focal point and the focal distance of each lens.

So a reflection being oddly-shaped and somewhat of the axis of the other reflections doesn't really tell us anything?

Would a technician be able to diagnose any issues with a lens, based on these reflections?

 

[Dan Fromm]

So a reflection being oddly-shaped and somewhat of the axis of the other reflections doesn't really tell us anything?

Depends. I have a lens with a tiny conchoidal chip at the very edge of the front element. The lens' flare pattern points to that very strongly.

Would a technician be able to diagnose any issues with a lens, based on these reflections?

[twit twit]The lens is uncoated and has a badly decentered element. You have been had. Send it back.[/twit twit]

 

[Sirius Glass]

What can lens flare tell us?
That you should use a lens hood.

 

[flavio81]

What can lens flare tell us?
That you should use a lens hood.

 

[Sirius Glass]

What can lens flare tell us?
That you should use a lens hood.

Lens caps have also been know to block lens flares. Just FYI.

 

[Nodda Duma]

As a lens designer I can say the info you can tell is quite extensive, but it's not so simple as allowing you to count the number of lenses.

Each "aperture flare artifact" is a separate ghost reflection path which traces from near-focus at the aperture to near-focus at the image plane. It's loosely but not directly an indicator of the number of lenses in the system. The potential number of single-bounce and double-bounce paths that could creat these artifacts is something like 2n + 2^2n, where n is the number of elements in the system. The actual number of artifacts depend on how well these ghost paths were controlled (historically not well). Veiling glare..which you can't see in the example image...is a result of the ghost image paths which are far out of focus. The visibility of the artifacts is an indicator of coating performance. The intensity of the "corona" around bright objects is an indicator of coating and surface quality (resulting from forward scatter at each surface).

All of these effects can be modeled and mitigated with modern design tools. For the applications I design for, this stuff has to be minimized to the greatest extent possible. I would not be a happy camper if I put together a prototype and saw these artifacts, but of course photography is a different application.


Oh, and I wouldn't say the lens used to take the photo would have a decentered element at all. The object is off-axis and the flare artifacts are going in the right direction. The elements are probably centered very well.

Decentered elements induce coma, astigmatism, and chromatic aberration to varying degrees depending on the specific orientations. I see no evidence of misalignment of optics in the presented photo. Trust me on this one... you put together enough prototype optics with really tight tolerances and you pick up on those clues pretty quick.

You'd have to precisely center a small bright object in the FOV to identify decenter from looking at ghost reflections, and you can't line up a lens like that on a camera. You need a centering station or similar optical test station to do that. Easier to deduce misalignment from evaluation of off-axis aberrations.

Regards,
Jason

 

[flavio81]

Jason

Aha! I finally know your name!

So, this is very interesting, you mean you can check for lens decentering by looking at the flare?

 

[Nodda Duma]

No, not in any reliable way...particularly outside of an optical shop in the hands of a consumer. A more effective evaluation is to observe the symmetry of aberrations across the fields. The image quality is where it really counts...not how the lens flare looks.

Best way I know of for consumers like us to evaluate alignment is via a star test (Harold Suiter's book written for testing telescopes...a great reference for any optical aficionado).

-Jason