Purple Reflection

[denoir]

Advertisement (gone after registration)

Ok, I know the difference between serial and multi, although the terms get confused for killers;) - but why can I see it on my 135 Apo Telyt then? A lens that is apochromatically corrected to the third spectrum? That would be a vase breaking without falling.

 

I suggest you take a look at the Puts article as well as this one: Chromatic aberrations (Secondary spectrum part). You can get CA from a superachromat and a hyperachromat without any problems which have even higher orders of correction. That it has corrections does not mean that it has been corrected. In fact it can never be entirely corrected for the full spectrum at all focusing distances - you would need an infinite number of corrections.

 

In addition Leica tends to play fast and loose with their APO definitions. I showed this comparison before - the non-APO 90 Cron vs 90 APO Cron:

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!

 

The APO shows basically no improvements in LoCA at close focus. Lateral CA on the other hand is handled much better:

 

So don't put too much trust in the APO designation. Usually it just means that some form of corrections have been applied but nothing more.

[denoir]

I would like to draw attention to this post by MJH, where he states that purple fringing can be induced on a sensor by laser light, which means without any CA present:

 

 

http://www.l-camera-forum.com/leica-forum/leica-m9-forum/107052-m9-50-lux-asph-too-much.html#post1132055

 

It would be interesting to see an example of that - something the person that put forward the claim did not provide. And no, I'm not going to point a laser at my M9 sensor, I'll leave that to somebody else I don't think there is any risk of actually damaging the sensor, but I don't feel like experimenting as the only laser that I have is a high energy green one.

[Lindolfi]

Have already experimented with digital sensors and green lasers of the 5 mW type and used a 1000x ND filter to keep the light levels low, so no problem there: at short exposure times the laser dot is not even overexposed.

[denoir]

Have already experimented with digital sensors and green lasers of the 5 mW type and used a 1000x ND filter to keep the light levels low, so no problem there: at short exposure times the laser dot is not even overexposed.

 

That might be a problem though - an ND filter is a piece of glass with certain optical properties. It may not be aberration free - even though I seriously doubt that the aberrations would have the same nature as the ones you get from a focused lens.

[Lindolfi]

No problem there either: the filter is positioned at right angles to the laser beam, while the rays from the laser beam are almost perfectly parallel. Even if this B+W filter has curvatures, the laser light will not be dispersed, because it is monochromatic.

[denoir]

Yes, in any case, monochromatic light should per definition not be able to result in chromatic aberrations. It's a good experiment - looking forward to seeing your results.

[denoir]

Advertisement (gone after registration)

Actually, I remembered now that I have done laser experiments with a camera years ago. Not for any scientific purposes, I just experimented with the visual effects. The camera in question was a Minolta Dimage Z3 - a bridge camera, so it's with the lens on. It had a CMOS sensor.

 

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!

 

 

 

In this case at least I can't find a trace of any PF.

[Lindolfi]

Here is the result from the green laser experiment on the M9 sensor, shot through a 35/2.0 asph summicron at f/8 with a 1000x ND filter (10 stops) and a 64x ND filter (6 stops) stacked on top of each other. The laser was a 5 mW laser positioned at 0.5 meter from the camera (to reduce the spread) and the lens was focussed at infinity. The laser was placed exactly in the optical axis. The exposure was 7 stops longer than needed (to get the heart of the beam well exposed) in order to get as much fringing as possible by a huge overexposure of the light source. This is from the center of the image at 100%:

 

 

And this is at the side of the image (the camera was rotated around the vertical axis, to make the laser shine close to the edge of the sensor:

 

 

As a reference here the image from the center, but now properly exposed to the right of the histogram, also at 100%

 

 

 

The only shades visible in the first two highly overexposed images is that of green (apart from white due to the overexposure, which is confirmed with spectral analysis) and so we may conclude that a monochromatic green light does not produce purple fringing.

 

So for now we can stick to the serial causality model, in which the primal cause of purple fringing is the chromatic aberration of the optics: the sensor can not do it on its own.

[jaapv]

thanks. An experiment can convince me Are you going to rewrite that Wikipedia article now?

[mjh]

Here’s the result from an experiment with a red laser, captured with a Fuji FinePix F10. Obviously the purple in the image cannot have be caused by chromatic aberration of any kind (or indeed any purely optical phenomenon that would be conceivable in this setting). I will try again with a green laser one of these days.

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!

[denoir]

Michael, there's purple in the image, but it's definitely not fringing - which should appear around high contrast edges. In your shot it's basically everywhere. It could be something as trivial as in-camera color correction. A white balance correction would probably eliminate the purple altogether.

 

I think both our laser samples are flawed as 1) preferably there should be no optics in the way 2) P&S cameras do all sorts of postprocessing on the images 3) a green laser should be used as red is too close to purple and in-camera PP can mess up colors.

 

Lindolfi's experiment was better as he shot directly at the sensor with a green laser and with a camera that does only a small amount of internal image processing (compared to a P&S anyway).

 

Speaking of wikipedia, there's actually an article that covers shooting lasers on digital camera sensors There are a couple of sample images:

Speckle pattern - Wikipedia, the free encyclopedia

[mjh]

Michael, there's purple in the image, but it's definitely not fringing - which should appear around high contrast edges. In your shot it's basically everywhere. It could be something as trivial as in-camera color correction. A white balance correction would probably eliminate the purple altogether.

Indeed I believe that the purple in purple fringing is due to an in-camera correction. It’s probably for the same reason that smearing is purple, even when it affects all kinds of pixels equally and one would rather expect white. But since red- and blue-sensitive pixels are generally less sensitive than the green-sensitive ones (mostly due to the different transmission characteristics of the red, green, and blue filters), the signals from the red and blue pixels get an extra boost to compensate for that disadvantage. For this reason, any effect (be it electrically (smearing) or optical (purple fringing)) that is affecting all pixels equally will create purple artefacts in the resulting image, due to the extra boost for red and blue.

[Lindolfi]

Now, what about images like those shown in this posting by ArtZ (click). Here we see dramatic blue fringing next to some very limited (in terms of pixel spreading) purple fringing.

 

Here also the claim is that it is the sensor. The images that are shown have many stops over exposed backgrounds and what is happening here is the following, I think:

 

Again, the CCD sensor is not the primal cause, this time the blue fringing comes from the shortest wavelengths that can be captured by the sensor: blue/violet. The stray blue/violet light is produced in the optics. The light behind the head of the picture of ArtZ may be 7 stops overexposed and so you need less than 1% of stray light to produce this effect. Good coatings have a reflectivity of around 0.2%. But there are many air/glass transitions in a multi-element optical design, so that adds up. The lens elements are all coated and if you look at the colour reflected from a coated lens element, it is often blueish or even purple.

 

So here the blue/violet fringing is not caused by the chromatic aberration, but by the reflections in the optics, eating their way from the overexposed parts into the normal exposed parts. The sensor is not the cause.

 

Now that is just a theory. We need another experiment and this time it is very simple: we need a lens without coating and I think I may have some in my collection...

[denoir]

Indeed I believe that the purple in purple fringing is due to an in-camera correction. It’s probably for the same reason that smearing is purple, even when it affects all kinds of pixels equally and one would rather expect white. But since red- and blue-sensitive pixels are generally less sensitive than the green-sensitive ones (mostly due to the different transmission characteristics of the red, green, and blue filters), the signals from the red and blue pixels get an extra boost to compensate for that disadvantage. For this reason, any effect (be it electrically (smearing) or optical (purple fringing)) that is affecting all pixels equally will create purple artefacts in the resulting image, due to the extra boost for red and blue.

 

Ah, but as purple fringing appears on film as well as on digital, it can't be a question of in-camera correction. Scroll to the bottom of the page for film examples: Chromatic aberrations

[denoir]

Now, what about images like those shown in this posting by ArtZ (click). Here we see dramatic blue fringing next to some very limited (in terms of pixel spreading) purple fringing.

 

Honestly, I have no idea what is going on there - I've never seen that effect with any camera or lens - and definitely not with the 50 Lux ASPH.

 

It looks like a software correction for CA has been incorrectly applied, but I've never seen something so extreme. Faulty Bayer array perhaps?

[Dannybuoy]

Just convert any photos with excessive CA to black and white

[01af]

Indeed I believe that the purple in purple fringing is due to an in-camera correction. [...] any effect [...] that is affecting all pixels equally will create purple artefacts in the resulting image, due to the extra boost for red and blue.

I believe this is indeed, fully or partly, the core of the explanation what purple fringing is (as opposed to longitudinal chromatic aberration). However for this hypothesis to hold water, it requires a way for either the photons or the resulting electrons to cross the boundaries between photosites behind (!) the Bayer colour filter array. Undue purple will occur only if a photon has travelled through a green filter but the corresponding electron ends up in a red or blue photosite.

 

If this happens then there will be two effects that make the shifted electrons appear purple. First, due to the higher transmission of the green colour filters, green photosites will be prone to spill over to adjacent (i. e. red and blue) photosites more than vice versa; second, the asymmetrical colour-correction coefficients supposed to compensate for the different transmission values will increase the purple hue even more.

 

So the question is: Can photons that have travelled through a green Bayer filter induce an electron that ends up in a red or blue photosite? It doesn't matter if the photons induce an electron in the wrong place or the electron gets created in the right place but jumps elsewhere afterwards; both ways the result will be effectively the same. Modern imaging sensors have provisions to drain surplus electrons so they won't pollute adjacent photosites ... but maybe if there's only enough photons then possibly some electrons will make it to an adjacent photosite nonetheless?

 

This might also explain why a wide aperture and a fast shutter speed can lead to a different (higher) level of purple fringing than a smaller aperture and a slower shutter speed, even when the resulting exposure value remains constant. In the former case the drainage system briefly gets overloaded; in the latter, not.

 

Finally, in real life longitudinal chromatic aberration and purple fringing will occur at the same time, adding up their effects. That's why in some cases, purple fringing seems to be affected by different lenses when actually only one part of the sum is affected.

[jaapv]

Honestly, I have no idea what is going on there - I've never seen that effect with any camera or lens - and definitely not with the 50 Lux ASPH.

 

It looks like a software correction for CA has been incorrectly applied, but I've never seen something so extreme. Faulty Bayer array perhaps?

I cannot find them on the run - there are at least two other threads showing this phenomen. I believe it is basically the same effect as the blue haloes around specular highlights in night shots, that most of us have encountered at one time or another.

[01af]

Ah, but as purple fringing appears on film as well as on digital, it can't be a question of in-camera correction. Scroll to the bottom of the page for film examples: Chromatic aberrations.

Uh oh ... you are confusing longitudinal chromatic aberration and purple fringing again. The former occurs on film and on digital both. The latter on digital only.

[Advertisement]

Hallo,

Look here: Leica M

[ldhrads]

Halo reduction quickbrush in aperture should help.