Use of Filters in B&W Photography

Filters are an essential part of the B&W analogue photographers toolkit, especially so when it comes to landscape. Indeed, I’d go as far as to say that it is rare that no filter is needed in B&W landscape photography, not least because without a filter the sky invariably comes out too bright.

Colour Filters

One of the features of doing B&W conversions from colour digital images is having complete control over the different colour channels, which makes it possible to radically alter how the B&W image looks; the analogue B&W photographer can’t exercise this control in post processing (or printing), and has to instead resort to manipulating the makeup of the light spectrum using colour filters while taking the picture. (TL;DR )

(Arguably, this can be a blessing in disguise, as the unlimited freedom in the digital conversion can, and very often does, produce unnaturally looking images that jar. It is a commonly overlooked fact that the human brain is used to operating in B&W in low light situations and so has evolved certain expectations about how objects and scenes should look; radically diverging from these expectations may at times produce striking images, but it rarely provides a satisfactory sensory experience.)

The colour filters are simple devices that can be used very effectively. Unfortunately, an online search on this subject throws up mostly misunderstanding. The one thing that is repeated again and again is that colour filters increase the contrast of an image … that is simply not true, and using them as if that was the case will lead to disappointing results.

So what does the colour filter actually do? It partially blocks light passing through it that is of a different wavelength than its own. How much of that light is blocked depends on how different the wavelengths are. This can be visualised by looking at a standard image of the spectrum of visual light that plots colour in dependency on wavelength, going from violet on the left (~380nm), across the blues, greens, yellows, oranges to the reds at the other end (~750nm): the further the colour of the light is away from the colour of the filter, the more it is blocked.[1]

So, for example, a green filter, the colour of which falls somewhere into the middle of the spectrum (~550nm), will largely leave greens alone, somewhat block cyans and yellows, quite a bit reduce blues and oranges, and dramatically cut violets and reds.

This is simple enough, but there are a couple of complications.

Firstly, colour is not an actual property of light directly corresponding to wavelength, but merely a neurological construct of the human brain, based on relative similarity with limited number of points of the visible spectrum. For example, an object that is orange can be reflecting either orange light per se (i.e., wave length of around 600nm) or it could be reflecting a mixture of yellow (~580nm) and red (~650nm) light. The human observer can’t tell the difference, but the two seemingly identical colours do not behave in an identical fashion, and will stop being perceived as identical if we stick a colour filter in front of the object. (Ever noticed how carefully matched clothing stops matching when viewed through sunglasses?) This effect is called metamerism, and applies not just to the colour of the object but equally to the colour of the filter itself, i.e., an ‘orange’ filter might not in fact be pure ‘orange’ colour at all.

The good news is that for the purposes of using colour filters in B&W photography metamerism doesn’t make a drastic practical difference when dealing with colours of the spectrum between blue and red. In that case when what we see is a mixture of two different colours, the two constituents are to the left and right of the perceived colour, the resulting is more or less an interpolation, and while the filter handles each component a bit differently, the combined final effect is similar enough to be able to get a reasonable idea what a filter of a given colour might do.

Where metamerism makes a huge difference is when it comes to violets. The human eye interprets a mixture of red and blue as violet, behaving as if the spectrum was a circle where violet lies between red and blue (i.e., the colour wheel), when in reality the spectrum is just a line: true violet light is as different from red light as it gets. Consequently a red filter (~650nm) will drastically block true violet light (~400nm), whereas it will not block the red component in red and blue mixture. So when dealing with violets, the effect of a red filter, and to a lesser extent of any colour filter, is completely unpredictable. There is a simple fix for this, which we will come to in a moment.

The second complication is that some colours simply don’t correspond to a single wavelength of light at all: there is no brown or magenta on the spectrum. This makes it more awkward to assess how such colours might respond to a given filter. (But gain, there is a simple fix.)

Those complications aside, now that we know what the filter does, we can leverage its selective reshaping of the spectrum of the light passing through it to our own advantage, and the most common use case of this is to increase separation between two different colours in the image.

For example, the Scottish landscape during the winter half of the year is characterised by a mixture of orange tones on the one hand, and some greens on the other. In colour photography these are nice contrasting colours to work with. However, this is not the case when wanting to take a B&W photograph: a spot meter will quickly reveal these tend to have similar levels of luminosity, so will end up a similar shade of grey. So let’s use a colour filter to alter that.

We have two basic options here: we can use a green filter, which will make the greens appear brighter and the oranges darker, or we can use an orange filter, which will make the oranges lighter and the greens darker. The choice will depend on how I visualise the final image, and also on what the rest of the landscape looks like, for example the orange filter will reduce blues more dramatically, so I will get a darker sky (as a side note, in my experience the green filter rarely works satisfactory in this specific case, the result tends not to feel right, I think principally because our brain doesn’t expect daylight to have a green tinge that would make greens on the whole brighter than oranges).

Now, when we do this, we are essentially looking to increase the contrast between the oranges and greens, which is why it is commonly asserted that filters increase contrast. But before we slap the filter onto the lens, let’s use the spot meter to carefully measure the luminosity of the orange and green areas; and let’s say the meter reveals the orange areas are in fact about 1/4 stop darker than the green areas. If I proceed here with the orange filter, what is likely to happen is that the relative brightening of the oranges will in fact bring the two luminosities even closer together, i.e., the over all contrast will be reduced. This is why I said at the beginning that the assertion that colour filters increase image contrast is incorrect. (I have a particular image from the early days of my LF photography that I managed to spoil this way, and which I really regret because of the very unique atmospheric conditions on the day.)

The singular conclusion arising out of this prolonged discussion is this: to use a colour filter effectively, and to accurately assess its true effect, it is necessary to use a spot meter to meter the scene through the filter.

This simple trick addresses all the issues we run into: metering through the filter eliminates effects of metamerism, it does away with the complications caused by colours that don’t exist on the spectrum, and, not least, it allows us to work out accurate exposure instead of relying on the the generic correction factors stamped out on the filter. (These correction factors are based on some sort of a ‘typical’ landscape metered using an averaging meter and can lead to significant error in exposure if the scene is ‘atypical’, which is not that uncommon; this is particularly the case for the filters that have larger correction factors, where it is very easy to end up with badly overexposed negatives.)

The only issue that metering through the filter doesn’t address is the spectral characteristics of the film being used, i.e., the meter sensor never behaves exactly like film and different films stocks have different spectral characteristics. There is only one thing for it, you need to get to know the film you are working with. (Some suggest that even with metering through the filter an additional exposure correction should be applied for filters that significantly reduce blues, to account for the prevalence of blue tones in shadows. I am not convinced this is universally the case.)

Metering through a filter can produce rather unexpected results; when that happens it is often worth to repeat this with other filters I might intuitively not consider using; the above discussion should make it clear that what we think we see is not always what is actually out there in terms of light spectrum.

A final note: stacking different colour filters together will generally not produce the result one might intuitively expect — if you must, make sure to meter through the stacked filters.

Notes:

[1] Some of the Wratten filters, including 8, 12, 22 and 25, are said to be long pass (i.e., only filtering the shorter wavelengths. My experience suggests that in practice filters for common photography are actually band pass. There is a rough eyeball test that can be done for this: take two colour samples, one that is just touch warmer than the colour of the filter (i.e., toward red), and one that is red. For a genuinely long pass filter the contrast between the two colours should not change when viewed through the filter. But I doubt this makes practical difference anyway, perhaps at the light yellow it might.

Polarisers

To fully leverage the potential of polarising filters requires a good grasp of different types of light sources and how light reflects from different surfaces, something that every photographer should familiarise themselves with, but it’s a subject for a book, rather than a few paragraphs — see for example Light - Science and Magic by Hunter, Biver, Fuqua, and Reid.

That said, a landscape photographer might get away with just knowing that a polarising filter can be used to control (i.e., increase or decrease) reflections. For example when photographing a body of water, it is often possible to alter the luminosity of the water surface from near pure white to near total black. This is useful, as, generally speaking, an undisturbed water surface tends to look wrong in B&W when it is too bright.

In colour photography polarisers tend to increase colour saturation and tend to be used to de-haze an image. But I find that for B&W they have virtually no impact on the appearance of the image outside of rendering reflections, notably they make very little difference to how dark a sky will appear.

When working with manual focus lenses, a linear polarising filter can be used; these are cheaper, as well as more effective than the alternative circular polarisers, but linear polariser cannot be used with modern autofocus cameras.

It its perfectly fine, and often necessary, to stack a polariser with a colour filter. The order of the filters doesn’t matter for their effect, but the polariser is easier to adjust when it is at the front of the stack.

A good polarising filter for LF use should have a numbered scale on it; this makes it possible to select the filter orientation by looking directly through the filter, and then accurately reproduce it once the filter is attached to the camera. This is by far the simplest way to set a polariser on an LF camera, adjusting the selection on a ground glass can be quite difficult, especially when stacked with another filter.

Other Filters

Other filters, such as neutral density ones, exist. I have not found it necessary to use these for my B&W analogue photography. There is no benefit to using a graduated ND filter, as it’s simple to produce the same effect in the darkroom by burning, and I dare say, with much better control over the result.

Given the long exposure intrinsic to LF cameras, I have not needed to use a solid ND filter either, but I am not a huge fan of very long exposures, they might be useful to others.

There are also specialist centre ND filters for LF wide angle lenses, these are tied to a specific lens, stupidly expensive, and virtually impossible to lay hands on. Their usefulness is hotly debated.

Building Up a Filter Set

Filter quality varies hugely, and it’s a false economy to skimp on them, as the optical degradation caused by poor quality filter is immediately obvious — rather than buying a complete set of cheap or midrange filters (I won’t name names, but if it’s made of plastic and/or lacks multicoating, it falls into that category), I’d recommend to gradually build up a set of good quality filters as required.

To avoid having to cary multiple filters of the same colour, it’s best to make a decision on the biggest size required, and get a set of reduction rings, which can be had very cheaply. For a large format 4x5 camera 77mm is a good size to go for, as that will cover most of the common lenses.

Some of the best filters are made by Heliopan, they are the filter equivalent of the Zeiss lens (and made from the same glass). Alas the colour options of their filters are nowadays rather limited, lacking things like greens.

Here are the filters I use regularly, ordered by the frequency of their use:

  • Orange (O22): good for separating greens and browns, and for darkening the sky, but the effect on the greens can be undesirable when greens dominate,
  • Mid yellow (Y12): good for darkening the sky (close to O22), and separation between yellow-greens and deep greens, but with less undesirable impact on scenes rich in green,
  • Light yellow (Y8): good for some darkening the sky (enough for effective burning later), with relatively few side-effects in most landscapes (many photographers in the past had a light yellow filter pretty much permanently on their lenses),
  • Linear Polariser: for controlling water reflections,
  • Yellow-Green (Hoya X0): useful to accentuate young foliage, but not as generally useful as yellow filters.

Filters I use rarely:

  • Green (Hoya X1): useful for working with green foliage but often has too many undesirable side effects in scenes where other colours are significant; Y-G filter is often preferable.

Filters I almost never use:

  • Red (R25): can be used to quite dramatically darken blue skies, but it has significant impact on most of the visible spectrum (you are basically killing anything but red and oranges).