Is this CA problem on my 50/1.4 ASPH, it's normal or not?

[CheshireCat]

Advertisement (gone after registration)

All that has been beaten to death folks. Anything new since this thread ?

http://www.l-camera-forum.com/leica-forum/leica-m-lenses/119433-noctilux-0-95-purple-fringe.html

 

Thanks lct, I didn't notice that thread.

In that thread, 01af's explanation for the sensor-related causes (#3 and #4) has a problem: he says that sensel saturation renders a magenta color because there are twice the number of Green sensels than Red and Blue ones.

But demosaic algorithms obviously give G pixels half the weight of R and B. This means that when all sensels are saturated, the final color of the blooming will not be magenta but white, as most of us have experienced on CCD Leica cameras (CMOS sensors are far more resistant to blooming).

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[lct]

I have no clue about this sorry but in Thighslapper's pic above, we can see purple fringing caused by the sensor and CA caused by the lens so i guess you agree that the lens is not the culprit for both don't you?

[CheshireCat]

I have no clue about this sorry but in Thighslapper's pic above, we can see purple fringing caused by the sensor and CA caused by the lens so i guess you agree that the lens is not the culprit for both don't you?

 

No.

Why do you think that the out of focus cable purple fringing is caused by the sensor ?

[lct]

Because i have never seen that on film.

[FlashGordonPhotography]

Why do these things always turn into 15 page arguments over "facts"?

 

Purple fringing is normal, common and easily reproducable. If you give an image the right conditions it'll have it. If you don't it won't. You need:

 

* A digital sensor.

* Backlighting. The stronger the more likely.

* A high contrast edge between an in focus and out of focus area. Usually the dark edge is the one in focus. The more focused the edge is the more fringing you get.

* A fast lens. the faster the more fringing you get.

 

So if you don't want to see it avoid these conditions. Easy. And it's occasionally but not always, correctable in post processing. Sometimes you end up with an ugly grey band instead. My Noctilux and Summiluxes weren't the first n or will be the last that show this phenomenon. I first noticed it when I moved from using my 85L from film to digital.

 

Gordon

[jaapv]

Thanks lct, I didn't notice that thread.

In that thread, 01af's explanation for the sensor-related causes (#3 and #4) has a problem: he says that sensel saturation renders a magenta color because there are twice the number of Green sensels than Red and Blue ones.

But demosaic algorithms obviously give G pixels half the weight of R and B. This means that when all sensels are saturated, the final color of the blooming will not be magenta but white, as most of us have experienced on CCD Leica cameras (CMOS sensors are far more resistant to blooming).

And that is exactly the point: the sharp transition of contrast does cause a problem as the saturated sensel distribution at the edge is not what the software expects.

 

 

 

 

You are kidding, right ? Or do you really think a tripod could make the difference for magenta fringing ?

Of course. Magenta fringing can be diminished by making the contrast edge softer (as in defocusing). Thus blurring the edge by motion will give a reduction. Another proof it is a sensor phenomena.

If you doubt this, just give me one shot of a double or smeared magenta edge by motion blur.

 

I'm off with my Monochrom for the day, just returned from its heart transplant. No purple, magenta or whatever fringes there...

[Peter Branch]

Advertisement (gone after registration)

.............

An APO lens is doing exactly that. If you have 3 zeros on the LoCA graph then you have a better form of correction than achromatic lenses (2 zeros).

 

Actually no! Never mind the somewhat simplistic thought experiment let's stick with the observed and measured facts.

 

As the data published by E. Puts clearly demonstrates neither Leica or Zeiss are even attempting to correct three wavelengths in their latest Apo designs.

 

This is entirely consistent with his comments about the Blue/Violet part of the spectrum. By concentrating on minimising the secondary spectrum across all visible wavelengths they necessarily improve the Blue/Violet part of the spectrum.

 

A "simple" explanation of this is not easy. But if 3 wavelengths are corrected then the direction of the error changes from positive to negative at the extreme wavelengths, i.e. if blue/violet positive then red will be negative. The result is actually a greater average error across the spectrum than the shallow "U" shaped, i.e. both ends have the same direction, error that these modern Apo lenses have achieved.

[wlaidlaw]

Here are the edge crops as promised of the 18mm SEM with both the M240 and M9. Taken with lens wide open in the first two images, so you get a range of in and out of focus. There is surprisingly little fringing in both cases and I would not say that either the CMOS or CCD sensor is noticeably better or worse. My guess is that this lens is not producing the high edge contrast necessary to make this problem an issue. The third crop image is with the M9 stopped down to f8 to see if the increased sharpness/contrast (guessing that f8 would be the lens at its best) make much difference. There is still very little fringing. I think all this really shows is that fringing is not an issue with the 18mm SEM. All taken in DNG and processed with minimal alterations in ACR 8.7.1.

 

Wilson

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[lct]

I think all this really shows is that fringing is not an issue with the 18mm SEM

Or that your pics are not over-exposed enough for that. I would get the same results with any of my asph lenses including 50/1.4 asph and 35/1.4 FLE.

[wlaidlaw]

Funnily enough it was taking a picture of that same tree later in the year, when it was covered in blossom, which convinced me to get rid of my particular 50 ASPH Summilux. I cannot find the original M8 DNG but only the JPEG, which I had corrected to get rid of most of the fringing. The JPEG does not look any more over-exposed than the SEM 18mm pictures.

 

Now we all know that "Friday" lenses come out of Leica from time to time and this may have been one of those. It was sent back to Solms for the fringing and very stiff focus, where I suspect it spent three weeks sitting in a drawer, because it came back exactly the same on both problems. I swapped it for a MATE and later bought a Noctilux 0.95 for a "fast 50".

 

Wilson

[thighslapper]

Or that your pics are not over-exposed enough for that. I would get the same results with any of my asph lenses including 50/1.4 asph and 35/1.4 FLE.

 

yes ...... and I have spent ages hunting for a a suitable example of GENUINE chromatic aberration ...... and all credit to Leica ...... I really cannot find any convincingly obvious examples with any of the dozen or more Leica lenses I own and use.

 

I can find a few examples of purple fringing similar to the one I posted, but all are just the same and in almost identical conditions...... blown highlights adjacent to sharply delineated edges of much darker image....

 

Mr Cat can call it Chromatic Aberration if he likes ...... I will continue to refer to it as Purple Fringing, acknowledging it as a phenomenon with a different causation ...

 

I have hunted on the web ,..... and most of what is shown as examples of chromatic aberration from digital cameras IS NOT chromatic aberration in the optical sense it was originally described .....

[CheshireCat]

Why do these things always turn into 15 page arguments over "facts"?

 

Because we are trying to understand why these facts happen. If we understand why something happens, then we know how to avoid it whenever possible.

[CheshireCat]

And that is exactly the point: the sharp transition of contrast does cause a problem as the saturated sensel distribution at the edge is not what the software expects.

 

If you care to check my shots, you will see that the purple one has slightly more blur than the other. And, by the way, these were shot at 1/2500s with a 100mm lens on FF. How do you explain this ?

 

Of course. Magenta fringing can be diminished by making the contrast edge softer (as in defocusing). Thus blurring the edge by motion will give a reduction. Another proof it is a sensor phenomena.

If you doubt this, just give me one shot of a double or smeared magenta edge by motion blur.

 

It is not difficult to understand that it is hard to preserve edges on a blurred image. However, the blurred image will have a magenta cast that the trained eye can see.

 

I'm off with my Monochrom for the day, just returned from its heart transplant. No purple, magenta or whatever fringes there...

 

Watch out, they may have transplanted a color sensor by mistake

[CheshireCat]

As the data published by E. Puts clearly demonstrates neither Leica or Zeiss are even attempting to correct three wavelengths in their latest Apo designs.

 

You do not realize that the third zero is in IR, not shown in the graphs published by E.Puts.

Nevertheless, this gives a lower error in the visible spectrum (cubic vs quadratic approximation), and to some degree also in UV even if a lens is optimized only for the visible spectrum.

[CheshireCat]

Here are the edge crops as promised of the 18mm SEM with both the M240 and M9. Taken with lens wide open in the first two images, so you get a range of in and out of focus. There is surprisingly little fringing in both cases and I would not say that either the CMOS or CCD sensor is noticeably better or worse.

 

Of course. This is not a sensor issue.

 

My guess is that this lens is not producing the high edge contrast necessary to make this problem an issue.

 

Edges are much more contrasty than the ones in the palm tree shot I posted.

You will also notice that the little purple fringing you have is in the most blurry branches in front of the focus plane, and this is enough to disprove the nonsense theory of the "evil demosaic algorithm".

 

I think all this really shows is that fringing is not an issue with the 18mm SEM.

 

No, this simply shows that fringing is not an issue with ultrawides and slow apertures, because the purple part of the spectrum will be less magnified by the lens. Fringing will be much more visible with a longer FL (my palm tree was shot with a 100mm).

 

Funnily enough it was taking a picture of that same tree later in the year, when it was covered in blossom, which convinced me to get rid of my particular 50 ASPH Summilux.

 

Exactly. The 50 Lux has a longer focal length, and wider aperture. It is not an APO lens by Leica standards, hence the fringing.

[CheshireCat]

I can find a few examples of purple fringing similar to the one I posted, but all are just the same and in almost identical conditions...... blown highlights adjacent to sharply delineated edges of much darker image....

 

You probably need an eye doctor if you call "sharply delineated edges" that horribly defocused example you posted

 

Mr Cat can call it Chromatic Aberration if he likes ...... I will continue to refer to it as Purple Fringing, acknowledging it as a phenomenon with a different causation ...

 

Do whatever makes you feel good.

 

I have clearly demonstrated how Chromatic Aberration produces Purple Fringing. You and many others continue to call it a sensor issue without the slightest proof, because you stubbornly refuse to accept the fact your Leica lenses are great but not perfect.

 

I just found another thread from 2011:

http://www.l-camera-forum.com/leica-forum/leica-m9-forum/187196-purple-reflection.html

In which Luka (denoir) explained and proved as well how purple fringing is not caused by the sensor. But people could not believe him and continued to talk nonsense with an authoritarian attitude, so much that most of you are still convinced the sensor is the cause. No wonder Luka is not around this forum anymore

Do yourself a favor and read the entire thread.

P.S. You will also find there nice examples of how the "APO" Summicron 90 performs.

[thighslapper]

I surrender.

 

As you have absolute faith in your interpretation you have bludgeoned me into silence....

 

All I would point out is that I cannot find ANY authoritative information on this subject from an optics or digital sensor expert ..... everything is derivative, from self-styled photo 'experts', and the examples shown are ASSUMED to be showing what they state.

[CheshireCat]

I surrender.

 

As you have absolute faith in your interpretation you have bludgeoned me into silence....

 

All I would point out is that I cannot find ANY authoritative information on this subject from an optics or digital sensor expert ..... everything is derivative, from self-styled photo 'experts', and the examples shown are ASSUMED to be showing what they state.

 

How sad is your attitude.

You don't even know the definition of "faith".

 

If you do not agree with the proof I (and not only) have provided, do your own experiments and prove us wrong. I will be glad to learn something new from you (no sarcasm).

 

But doing your own homework requires some effort, and 95% people will abide with Internet myths and accept anything that is more fascinating (UFO) or makes them feel ok (your expensive Leica lenses are perfect).

 

Please react

[Advertisement]

Hallo,

Look here: Leica M

[FlashGordonPhotography]

Because we are trying to understand why these facts happen. If we understand why something happens, then we know how to avoid it whenever possible.

 

I've understood the conditions under which happens for 15 years. More importantly I can identify a shot that's going to suffer from purple fringing before I take it, so I can make a decision to change something or live with it.

 

I don't need to know what type of aberrations it is to do that. I just need to know the conditions that will make it happen.

 

Gordon

[Peter Branch]

You do not realize that the third zero is in IR, not shown in the graphs published by E.Puts.

Nevertheless, this gives a lower error in the visible spectrum (cubic vs quadratic approximation), and to some degree also in UV even if a lens is optimized only for the visible spectrum.

 

What evidence do you have for this assertion?

 

You seem fixated on the idea that three colours "in focus" is axiomatically better that two colours.

 

If you look at the E. Puts graphs they can be interpreted as saying that the current Leica and Zeiss lenses have two colours in a focal plane and an almost infinite number of colours in between these colours that are very nearly in the same focus. Also the colours towards the Blue/Violet and the near IR away from the two colours are very much better corrected than non "Apo" lenses.

 

This is a considerable achievement. If you doubt this get out your old school text books and design for yourself an achromatic lens. Then using you PC, which is the easiest way, calculate the errors for that design across the visible spectrum. You will be surprised how bad such lenses actually are.