15

What makes digital cameras more prone to suffer from chromatic aberration than film cameras?

I've read this at many websites, but the explanations differ, from what I think less credible explantions like "the digital cameras high resolution makes it more prominent" to more credible-sounding ones involving colour filters in front of the sensor creating another source of aberration in addition the what the lens has already produced.

Is there any truth in the statement in the first place, and if so why is that the case?

mattdm
  • 143,140
  • 52
  • 417
  • 741
Hugo
  • 8,288
  • 9
  • 46
  • 85
  • 3
    Provide a source, else this sounds like a silly myth. It's the lens that causes chromatic abberation, not the sensor. There are some issues due to the Bayer matrix, but I wouldn't call those "chromatic abberations". Chromatic abberations are caused when the lens focuses different wavelengths of light differently. – Olin Lathrop Feb 24 '14 at 22:43
  • 2
    I can't provide the exact sources where I originally read about it (It was a while ago), but a quick search for it gave loads of similar results: 1 2 3 4. It may very well be a myth as I wrote in the question, but if so it's widespread – Hugo Feb 24 '14 at 23:08
  • 2
    maybe because film camera photos were mostly viewed at 15x10cm light reflecting prints in cosy light, where the viewer had the big picture in mind, while virtually all digital photos are anally inspected closely for errors at 100% "crop" on 15-25inch light emitting monitors or 30-50 inch TVs? – Michael Nielsen Feb 25 '14 at 20:44
  • Digital is so much sharper, in general, that you see issues not apparent in film, also because in film you don't zoom-in to "1:1", while in digital it's common. Take a film, enlarge by 500x and let's see if, after the overal blurriness, you can't find any aberration. – FarO Jul 05 '17 at 15:55

5 Answers5

10

Obviously, chromatic aberration is created by the lens, and the amount of CA is the same.

However, film as a medium and the sensor respond a bit differently. True perpendicular light is handled in a similar way in both, but angled light meets a different surface when using film and when using a CMOS sensor.

CMOS sensors have tiny lenses over the color filter (see here), and it is pretty hard to provide a uniform group velocity inside a small lens for all kind of light wavelengths, so these create an angle-dependent and wavelength-dependent response to arriving light. (Consider white light going through a prism - same effect).

A film has much less sensitivity to incident angle. So you will just photograph the CA.

On the other hand, R, G and B coming from an angle will see different sensor sensitivities (each is different) than RGB coming perpendicular to the sensor. So that will show up as color shift or color change, making CA worse.

Well, this is the explanation I can think of for your question.

(And a good test would be to use directed white light on a CMOS sensor, and make photos starting from perpendicular and then tilting it more and more. I would expect a bit of color shift. But do not try this at home :-) ).

TFuto
  • 6,206
  • 21
  • 37
  • 2
    There is also the rather thick (~1mm or so) antialiasing/UV/IR filter glass that sits directly in front of the sensor. In the corners of the sensor, incoming light will hit at an acute angle (unless the lens is telecentric) and this can do nasty things to the colours. Especially on say some Leica M-mount wideangle lenses where a small-diameter rear element is deeply recessed into the, well, not the "mirror house" but the corresponding cavity in the camera, close to the film plane. This doesn't happen with film. Microlenses can only do so much to overcome this. – Staale S Feb 25 '14 at 18:57
  • @TFuto - please see my answer. There is no need for jargon like group velocity (a fancy word for what amounts to axial color) here. The microlenses are also not very relevant to CA, as even if there was lateral color for a decently designed microlens the blue light would either already be filtered on top of the red pixel, or be filtered out below the microlens. In general, if this did not happen the bayer matrix would become corrupted and you would get very strange images out of the camera. – Brandon Dube Nov 07 '17 at 00:22
1

A digital camera quantizes the light more coarsely than a piece of film. Consider if the lens has 3 microns of chromatic aberration. On a film image, you will get something a little bit bigger than 3 microns - maybe 3.1 microns - due to the film's silver halide crystals. On a digital camera the pixels are, say, 6 microns on a side. 3 microns is enough to significantly spill into the neighboring pixel, so the amount of chromatic aberration appears to have doubled compared to film.

They also see color differently. Consider this test someone put together. Consider example 6. The blue vehicle behind the overexposed one is almost black in the film image, and reasonably bright in the digital one. The red headlights are also exposed very differently, even relative to those things around them.

This implies that the film is less sensitive to red light, and also less sensitive to blue light. All of the fringing you see is magenta, which is not a color, but a combination of red and blue. If the film is less sensitive to these colors, compared to white or greenish elements of the scene the chromatic aberration will be reduced in intensity, and thus visibility.

Brandon Dube
  • 2,445
  • 14
  • 24
  • The link which you have titled "not a color" does not support that claim. – mattdm Nov 10 '17 at 05:43
  • @mattdm _ it is physiologically and psychologically perceived as the mixture of red and violet/blue light, with the absence of green._ – Brandon Dube Nov 10 '17 at 19:04
  • 1
    "Magenta is an extra-spectral color". That is not at all the same as "not a color". I mean, sure, "color" has lots of different technical definitions, but I don't think such a limit one is particularly useful. (By this definition, pink and brown would also not be colors.) And anyway, the link you provide defines it as a color, definitely right there in the quote I give. – mattdm Nov 10 '17 at 19:13
  • @mattdm - if you object to the language of my answer, propose an answer. Make sure it does not make the message less clear in exchange for better fitting your own notion of what a color is. – Brandon Dube Nov 11 '17 at 01:28
  • 1
    I think your answer is fine, except, like I said, your link to "magenta is not a color" goes to a page which literally says says magenta is a color. If you want to argue that for some reason it is not a color, I think you should at least find a better reference. – mattdm Nov 11 '17 at 04:38
  • @mattdm if I said "pure color" instead of "color," would you be satisfied with that grammar? – Brandon Dube Nov 11 '17 at 05:00
  • I think "spectral color" is really the best thing to use. Yellow can either be a pure color (in the technical sense of generated from homogeneous spectral light) or a mix of red and green, for example. – mattdm Nov 11 '17 at 05:20
  • Please define "homogeneous spectral light." – Brandon Dube Nov 11 '17 at 05:36
  • 1
    Spectral color is color which we perceive in response to the stimulus of single (or in practice, very narrowly similar) wavelengths of light. "Pure color" can have this sense, but it has other senses as well. – mattdm Nov 11 '17 at 16:06
  • Purple fringing is typically accompanied by green fringing. If your theory that purple fringing is due to CA being more readily recorded by digital sensors is correct, shouldn't we expect film to see plenty of green fringing in film? – Caleb Nov 13 '17 at 18:47
  • The variety of chromatic aberration in the images I linked to on flickr is axial color, not lateral color. "purple fringing" axial color has green in focus, and red and blue out of focus. – Brandon Dube Nov 13 '17 at 20:26
0

this due to way color are extracted from sensor. Only few digital camera truely see in color (the old sigma foveon is one of them in large "public" SLR). The sensor only see intensity of light so "black and white" and a front grid with colored filter is used to, later in the process, try to define the original color. (see Bayer's grid and their evolution) (sample of bayer application) Due to this interpretation, some situation give the wrong color as deduction. This happened often at the edge of sharp surface.

JDługosz
  • 4,382
  • 1
  • 19
  • 38
NeronLeVelu
  • 267
  • 4
  • 10
  • 3
    Can you please elaborate and or provide some sources to why this results in colour fringing? If this is the reason it seems to be unrelated to chromatic aberration. – Hugo Feb 25 '14 at 09:05
  • 2
    Also note that the way the Bayer sensor works is closer to the way the human eye works than Foveon! By this standard, humans don't see in color either! – mattdm Feb 25 '14 at 09:18
  • I recommand this article that explain how it append http://phaseoneimageprofessor.wordpress.com/tag/bayer-pattern/ (not reinventing the wheel). You could also look at wiki pedia for foveon that partially explain also (advantage of foveon on this issue) http://en.wikipedia.org/wiki/Foveon_X3_sensor. @mattdm, right eye work more like bayer but dispersion of eye unit sensor are different than a "regular" bayer grid, also each base unit capture different information where bayer/sensor capture the same filtered so aberation is not handle the same way by the brain. – NeronLeVelu Feb 25 '14 at 10:43
  • 1
    This explains how a Bayer sensor can cause a certain class of color errors, but I wouldn't call those errors chromatic abberation as was asked in the question. – Olin Lathrop Feb 25 '14 at 14:49
0

I don't think any of the possible reasons you have read are wrong, as such. Certainly, the majority of the reasons in the links you provided seem plausable enough causes of a small amount of chromatic abberation.

Things like wonky lens elements and other manufacturing issues aside, the sheer complexity of modern lenses compared to those in the days of film and the addition of micro-lenses on the sensor will all contribut to the colour fringing you see. Increased resolution, as daft as it sounds, does highlight imperfections in many lenses, and frankly I don't think it's possible to study a print anywhere nearly as close as you can with a large screen at 100% zoom.

As nice as it would be to say that there is a specific reason why film is better than digital in this regard, it seems that it's actually a combination of many smaller factors.

Thomas Bisset
  • 85
  • 6
  • You are wrong. Expectations got nothing to deal with it. Chromatic aberrations ARE more pronounced on digital than film. It can be measured. One of such attempts was made by genotypewriter: https://secure.flickr.com/photos/genotypewriter/6147351879/ - result being quite obvious: Film wins when it comes to CA. – MarcinWolny Feb 26 '14 at 09:06
  • 1
    Altered my answer to remove my last comment. On my phone so I'll read the links later - in the mean time I'll take your word for it. :-P – Thomas Bisset Feb 26 '14 at 09:55
0

I think Hugo writes about Blooming, that occurs on sensors. Mainly small sized big resolution sensors are blooming prone. It caused by high intensity of light, much higher than photodiod may handle. So the electrical charge overflows into adjacent photodiodes. As a result it creates the fcoloured ringes on the edge of over-exposed areas.

Jeff_Alieffson
  • 101
  • 1