30

I am wondering if there is a physical filter that would allow a camera to make black and white pictures without using any software effect/filtering?

  • 4
    You've got actual answers to this, but I'm wondering what problem you're trying to solve here? – Philip Kendall Aug 15 '16 at 9:55
  • 4
    No, IR filters do not "replace" colors. They pass different wavelengths. – Carl Witthoft Aug 15 '16 at 11:10
  • 1
    If you're talking digital cameras, you can't make any picture without using software. It'd be difficult to draw a clear line between interpreting data from a sensor to create an image and applying an "effect," which in many cases is just a different interpretation of the same data. So you're kinda talking semantics here. – Caleb Aug 15 '16 at 16:52
  • 3
    You can find a monochromatic filter that only passes one color of light, so your image would be "black and red" or "black and green", but that's the closest you can get. – JPhi1618 Aug 15 '16 at 18:08
  • 3
    There's a really simple way to accomplish this actually: Use black-and-white film. – thanby Aug 16 '16 at 19:50

10 Answers 10

19

No.

It is not possible to create a physical filter that can completely "De-saturate" incoming light.

The only way to achieve this without post-processing is at the film / sensor level.

  • 2
    I think it's theoretically possible, using beam-splitters and monochromatic filters tuned to the colors of the camera sensor's pixel bins... – Hao Ye Aug 15 '16 at 19:12
  • 2
    @HaoYe you can never remove the frequency component of the light, so you can never make it black and white. – Brandon Dube Aug 15 '16 at 19:35
  • 2
    Exactly, you can't create an optical filter that will only pass luminosity regardless of frequency. – Digital Lightcraft Aug 15 '16 at 19:37
  • 5
    @HaoYe: Consider an example: bright green monochromatic light. Your filter has to transform that into white light, so the sensors detect equal levels of red, green, and blue (equal after taking their sensitivity into account). Introducing new frequencies is not possible with traditional optics, AFAIK. It may be theoretically possible with quantum effects, like absorbing an re-emitting light, but probably not while preserving the direction of photons. (A photon has energy and momentum that depend on wavelength...) – Peter Cordes Aug 15 '16 at 20:48
  • 3
    What you have described there is almost exactly how a night-vision light amplifier works. But its not a "filter" ie a pass-through single element. – Digital Lightcraft Aug 16 '16 at 7:11
30

Pardon me while I get a little metaphysical for a bit. "Color" as we understand it isn't a real property of anything in the universe. It's something created by our vision system — a complicated interaction in our eyes and brains. It's useful for things like "don't eat the poison berries", "look at that tiger over in the grass", and, more recently, "stopping our vehicles at intersections".

This sense is based on something which is a real property of objects in the universe: different materials scatter, reflect, and absorb different wavelengths of light in different ways. Our eyes have receptors which are differently-sensitive to different wavelengths of light, and the vision system translates that to what we call color.

Color itself can be thought of in a lot of different ways. One way which is helpful in this circumstance is to break it down into chromaticity and luminance — luminance is basically "brightness", and chromaticity is... the other color stuff — hue (red, orange, yellow, green, blue...) and saturation or colorfulness. Dividing the concept of color in this way works nicely with our mental model — but isn't actually immediately translatable back to the physical universe.

A filter which resulted in black and white would need to filter out the chromaticity and pass through only the luminance, because that's what a "black and white" photo basically is — just a record of the brightnesses, without all the other "color stuff".

But, I don't know of any way to do that. It certainly isn't possible with something similar to the kind of filters we normally use. Those just block either certain wavelengths (in the case of color filters or UV or infrared filters) or generally all wavelengths to a small degree (in the case of neutral-density filters). A "filter" which converted to black and white would have to actually transform the wavelength in some way (since light with no wavelength is ... darkness), rather than filtering it. This would probably involve some kind of nonlinear metamaterial and nothing I can explain with my high-school-level knowledge of physics. And it would have to convert all different wavelengths to the same wavelength, or else scatter them randomly so the result would be white light; this seems like it's probably preposterous. I feel safe in saying that even if it were possible, the result wouldn't be something you can attach to a camera and carry around.

On the other hand, one can certainly record just the brightness. That's what black and white film does, and it's actually what digital photosites do, too. They're inherently just measures of brightness, but today's digital cameras use filters to record brightness only in certain wavelengths, measuring blue, green, and red separately. (This roughly matches how human vision works, so we can combine that back to make a full-color image.) If you have one of the few cameras made without these filters (like the Leica M Monochrom), you just get a black and white image.

Of course, another approach is to filter out everything except one specific wavelength. You can see this in Jerry Coffin's answer here, or in this other question involving an almost-monochromatic sodium-vapor light. That's black-and-some-single-color rather than black-and-white, but may be close to you want. Of course, that cuts out quite a lot of light, and the other downside is that it also cuts out brightness levels from other colors — so you'll just see variance in green (or whatever selected color) and nuances of shade in the other colors won't register at all.

  • Mattdm... the philosopher :o) – Rafael Aug 16 '16 at 8:13
  • Color filters do not block everything except the color of the filter. Some of the entire visible spectrum gets through each of all three color filters. It's just that more, often much more, of the colors closest to the color of each filter gets through. A good bit of red gets through the green filter and vice-versa. Some green gets through the blue filter and vice-versa. Even a small amount of blue and red get through the other colored filters. That is the way human vision works, that is the way color film works, and that is the way digital cameras work. – Michael C Jun 22 '17 at 13:40
  • Anyone who has used color filters in front of B&W film understands this intuitively. A red filter doesn't block all light except red. It just lets less of the other colors through, so in the resulting photo things that are those other colors look a darker shade of gray compared to red objects that were the same brightness in the scene. – Michael C Jun 22 '17 at 13:44
  • Sure; we perceive pure-wavelength yellow light because that activates both the "red" and "green" cones, and record it because it passes through both the red and green filters. But I think the simplification is perfectly adequate for the explanation here. It certainly doesn't affect the basic point about a "black and white" filter. – mattdm Jun 22 '17 at 15:37
16

All color is a result of software processing. The only thing a sensor, be it film or semiconductor, can do is change state in response to incoming photons. Yes, a digital camera has color filters, but all they do is restrict the wavelengths which are passed to the sensing pixels. The output of each pixel is simply a bunch of electrons, which are then converted to a voltage, which in turn is measured and reported as a digital number.
How you choose to interpret those numbers is entirely up to you. A couple examples:

Load a RAW file into a math tool such as R or MATLAB, and you can generate a monochrome image based on the numeric values in the array.

Load an RGB file similarly. It consists (generally) of three equally-sized arrays of numbers which have been tagged as "R,G,B" layers. You can generate a monochrome image of each, or assign whatever hue & chromaticity you wish to each layer before combining into a color image.

Again, the important thing to understand is that your original question is in error: whether through digital data processing or thru use of developer chemicals and color vs. B&W print paper, the camera and its sensor know nothing whatsoever about color. It's how you process the data (digital or analog).

12

You can't add a physical filter, but you can remove a physical filter to convert any digital camera to a strictly monochrome camera.

The actual sensor on any DSLR knows nothing about color - each pixel records the total luminosity in all the wavelengths it's sensitive to. The way color is introduced is by adding a Bayer filter, which is basically small pieces of differently colored glass for each pixel: Now some pixels can only see blue, others only red, and the rest can only see green.

By removing the Bayer filter, you camera will go back to being monochrome, as some people have actually done:

enter image description here

  • There are also monochrome cameras on the market – Hagen von Eitzen Aug 16 '16 at 21:37
  • I only know about the Leica M Monochrome which is a bit pricey for me personally, sadly. – Josef Aug 17 '16 at 12:52
6

No.

Each colour camera has three types of sensitive material - pixels in digital cameras, pixel layers in Foveon sensors, layers in colour film. Image being monochrome means that all those types produce response with constant chromaticity with any incident light and that is NOT possible because they are engineered to produce different chromaticities.

  • Semi-true but misleading. Can you edit to read "...engineered to produce responses to different chromaticities" ? – Carl Witthoft Aug 15 '16 at 12:33
  • @carl-witthoft: do you mean that it is possible to interpret that as "each layer produces own chromaticity"? – Euri Pinhollow Aug 15 '16 at 14:14
  • Each layer records a bunch of photons whose wavelength allows them to pass thru the color filter (and fall within the detection range of the pixel, of course). The end user can assign whatever color to that layer he wishes. – Carl Witthoft Aug 15 '16 at 14:30
  • @carl-witthoft: does not answer my question. I cannot understand the idea behind "engineered to produce responses to different chromaticities". – Euri Pinhollow Aug 15 '16 at 14:43
  • 1
    Well, yes -- en.wikipedia.org/wiki/Chromaticity . You can't get chromaticity from a single color filter. What each color filter does is integrate spectral input over a defined wavelength range, with varying transmissivity across that bandwidth. How you map that to an axis of a CIE map is up to the user. – Carl Witthoft Aug 15 '16 at 14:45
6

It's sort of theoretically possible, but not it's not generally practical.

To do it, you need a relatively narrow band-pass filter that restricts the light that passes through to the point that only one color of the (usually) three detected by the sensor will be affected (at least to a degree that it has visible effects on the picture you take).

Such narrow-band filters have been built and are in regular use--for example, they're used on a regular basis in wave division multiplexing, which is used to send multiple signals over an optical fiber simultaneously. On the transmitting end, you take a number of signals, encode each as a single color of light, and mix the light together before transmitting.

On the receiving end you run that light through the same number of narrow band-pass filters so you can reconstruct the original data streams.

As to why it's not practical: two reasons. First of all, such filters can be fairly large and expensive. Second, (probably more important for photographic purposes) as you get a narrow band that's passed, you also typically get a fair amount of attenuation in the pass band. That is to say, along with getting rid of the light you don't want, you also typically lose quite a lot of the light you do want.

On a typical camera you're dealing with only three colors of sensors, distributed pretty widely in the spectrum. You'd typically want to keep the green light because 1) that's the range where people's eyes are normally the most sensitive, and 2) on a typical sensor, you have twice as many green sensor wells as either red or blue sensor wells.

Astronomers also use fairly narrow band-pass filters on a fairly regular basis. To be specific, one type of emission nebula emits light due to triply-ionized oxygen (aka "oxygen III"). The light emitted is at 496nm and 501nm, both of which are pretty close to the middle of the green range:

enter image description here

So, if we insert a filter to pass only those wavelengths of light, and stop essentially everything else, we get pictures that are pretty close to purely monochrome, regardless of the camera/sensor/film used to sense the light. Such filters are easily available (Googling for oxygen-III filter will turn up many choices). Just for an example, here's the response curve for one of these filters:

enter image description here

This particular one is a hydrogen-beta filter, but oxygen-III filters with similarly narrow band-pass are available. A few slightly wider band-pass filters (still usually called "narrow-band") are "tuned" to allow both hydrogen-beta emission (486 nm) and Oxygen-III emissions (496 and 501 nm). This one, however, would filter out most emission at 496 nm and essentially all of it at 501 nm, even though to most people's eyes all three colors are very similar (deep gree with just a hint of blue).

These filters, however, are generally designed for use on telescopes, not cameras. They're typically in the sizes (e.g., 2 inch) used for telescope eye pieces. They also block a lot of visible light, so they're typically recommended only for use on relatively large telescopes--at least 8 or 10 inches is the usual minimum for them to be of much use.

Even assuming you could get the filter mounted and could live with the amount of light being transmitted, you'd be left with one problem: although your picture would be (almost entirely) monochrome, unless you did some pre-processing, it wouldn't show up as shades of grey, it would show up as shades of green.

I can see one final problem to use of these filters: what you'd get probably wouldn't work well for most types of photography. Early black and white film had a fairly broad range of sensitivity, but was affected the most strongly by blue light, and only quite weakly by red light.

Later black and white file ("panchromatic film") was adjusted to have sensitivity across the visible spectrum that corresponded much more closely to normal vision. This was enough of an improvement that it fairly quickly replaced orthochromatic film for most typical photography.

In this case, you'd be detecting a much narrower range of light that orthochromatic film though--to the point that you probably would not be able to get results that were of much use for most typical purposes.

On the other hand, there are also a few up-sides to using such narrow-band filters under some circumstances. For one example, since the lens only has to focus one wavelength of light, chromatic aberration becomes essentially irrelevant. This can improve resolution (though the exact improvement will depend on how much chromatic aberration the lens had to start with).

  • Astronomers also, apparently, do this first. – wizzwizz4 Aug 16 '16 at 8:47
  • @wizzwizz4: Really serious astronomers mostly start with purpose-built cameras (e.g. that have coolers for the sensor to reduce noise). Some casual astronomers take pictures with unmodified cameras. And yes, some who are in between modify a normal camera. – Jerry Coffin Aug 16 '16 at 14:18
  • Many cooled astronomical cameras (dedicated ones for use connected to a laptop, not self contained units) are monochrome - as are some astronomical video cameras. Using a monochrome sensor increases sensitivity for luminance shots (since each pixel gets the full range of wavelengths) and allows higher colour resolution when combining multiple shots through R,G,B or different narrowband filters. – JerryTheC Mar 28 '17 at 0:39
  • The usual reason for modifying a DSLR for astronomical use is that the built in Infrared blocking filter also blocks around 80% of the deep red hydrogen alpha light - which is the red part of emission nebula images. Replacing the filter with one that passes h-alpha light greatly increases sensitivity to this, but gives a red cast to normal photos - which can be compensated for by a custom colour balance, or by using a suitable front of lens filter. – JerryTheC Mar 28 '17 at 0:54
  • To expand on that concept: filter out several narrow bands in parallel, and then coerce them all into the same output wavelength by either flourescence or heterodyning.... – rackandboneman Mar 18 at 14:58
3

It's not a filter--not removable and definitely not reversible--but any digital camera can be converted to greyscale by scraping the color filters off the sensor and processing the RAW image. Without the color filters, the sensor only collects brightness information. The camera will continue to process the pixels as if the color filter matrix was still there, so you have to capture the RAW images and process them yourself. Never tried it myself, but I heard about it when CVS (US pharmacy chain) first started selling use-and-return digital cameras.

Thread with examples: http://photo.net/digital-camera-forum/00CM0R

More about the color filter matrix: https://en.wikipedia.org/wiki/Bayer_filter

Hope this helps!

2

In cameras the incident light is filtered to three coordinates of RGB spectra and then captured using chemical reaction (film cameras), CCD or CMOS chip (digital cameras).

The only way how you can physically disable the camera to capture colour images is to use monochromatic film or remove the filter mask from the CMOS chip. This procedure will kill your camera 999 999 times of 1 000 000 attempts.

When you set your camera to monochromatic capture it "ignores" the filtering and sums up the signal from all 3 channels. In postprocessing, the program will calculate the mean value from the channels.

If you want to capture IR images you have to have IR-compatible optics and IR sensitive detector. You will, probably, get brand new chip and customized AF sensors.

0

No. You have to understand that there is no such thing as the wavelength of the white light, so there is no physical property on which such filter could be based.

If you don't like physics, think of a logical example: white light is a wider set which includes lights of all other colours as subsets. So your question effectively sounds as

Is there a filter that can extract fruits from apples?

Again, the answer is NO.

  • I'd suggest: Is there a filter which can extract fruits from apples, oranges, and cherries? or similar. – mattdm Aug 17 '16 at 16:55
  • 1
    You could extract a generic "mixed fruit juice" from apples, oranges and cherries though :) – rackandboneman Mar 18 at 14:55
0

I'm going to go against the grain, and say YES, WE CAN... if you expand the meaning of "physical filter" as follows:

The filter is an active camera which displays its output in black and white on its own display (by having no colour filters on its sensor, desaturating in software, using a monochrome display etc), perhaps with some optics to simulate a focus farther away.

Your camera then takes a photo of the filter's display, thinking that it's the real world. And it's in black and white :-)

If that sounds outrageous, consider that in 2011, the movie Olive was reported as being the first film to be shot entirely on Smartphone. But how did they get the wonderful bokeh and depth of field? By filming the image projected on ground glass by an $800 Canon L Series 24-70mm lens! Cheating?

  • 1
    en.bugemos.com/?q=node/12 – Crowley Aug 18 '16 at 7:32
  • I agree with you, I don't understand how can a filter add colours (bayer filter) and why there shouldn't be an exact opposite filter to remove them. – MeV Aug 18 '16 at 14:27
  • 1
    @Mev: See my answer. The Bayer filter array does not add colors. It actually removes everything but certain (broad swaths) of wavelengths in a pattern which makes it possible to reconstruct full-color information in a way that roughly matches to the human vision system. Since we're showing the results to humans, this works out. – mattdm Jun 21 '17 at 20:17
  • And for this answer itself: I don't think it's meaningful or helpful to expand the meaning of "physical filter" in this way. – mattdm Jun 21 '17 at 20:19
  • Color filters do not remove everything except the color of the filter. Some of the entire visible spectrum gets through each of all three color filters. It's just that more, often much more, of the colors closest to the color of each filter gets through. A good bit of red gets through the green filter and vice-versa. Some green gets through the blue filter and vice-versa. Even a small amount of blue and red get through the other colored filters. That is the way human vision works, that is the way color film works, and that is the way digital cameras work. – Michael C Jun 22 '17 at 13:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.