By Anne Thomas-Davis - Chief Technology Officer, Formatt-Hitech
Anne Thomas-Davis has been with Formatt-Hitech for nearly 20 year as our Technical Director, prior to the sale of Formatt-Hitech to Kenko-Tokina, and remains as our Chief Technical Officer and Executive Advisor.
Anne and has been the technical lead behind all of our filters, including our best loved products such as the Prostop IRND, a market leader in it’s time, and more recently with Firecrest where we were the first manufacturer to bring neutral density coating technology to the photographic market in 2014. We remain ahead of the field with Firecrest Ultra, a product unmatched in terms of durability, clarity and neutrality thanks to Anne’s vision in bringing cinema quality manufacturing standards to the photographic market.
She remain “hands-on” on the factory floor, overseeing production and maintaining quality controls. Anne brings with her many years of experience in colour technology from diverse fields of coating and polymer industries and remains at the forefront of new product development.
James Stamp - Head of Sales & Marketing
The following paper has been written to give anyone interested in photography and ND filter usage an insight into how filter design has developed over the years. I have tried to keep it as informative as possible without being overly technical.
How We See Colour
The eye contains cone and rod cells.
Rod cells are efficient at absorbing light and enable us to see in low light conditions (night vision). Rod cells have very little sensitivity to colour. This is one reason that night vision looks black and white because almost all visible light waves are absorbed at night.
Cone cells contain three different colour sensitive receptors. The colour sensitive cones react to wavelengths;
- Red – Long wavelength
- Green – Middle wavelength
- Blue – Short wavelength
Therefore all colours seen by us are simply combinations of these wavelengths.
The human eye can normally see colour between 400nm and 700nm.
Colour and the Eye
How does colour information reach the eye?
When light hits an object, the light is either absorbed, scattered, or reflected. Any light that is absorbed will not be seen. Various intensities of reflected or scattered light enter the eye and depending on the degree of red, green and blue lightwaves present, variations of different colours are seen.
The various quantities of light needed to produce each colour we see are known as the Tristimulus values. Colours matched this way are known as Metamers. Note that colours of different spectral powers can produce the same colour response in the human eye.
It is worth noting that for every combination of colour that can be seen an exact spectral match is available.
What is Colour
All objects absorb, reflect and scatter light waves in different ways and it is the way the light interacts with the surface of an object that produces the colours we see. Put another way, it is the varying intensities of light waves reflected that enter the eye to produce the colours we see.
For example, a tomato absorbs blue and green light waves and reflects red light waves.
So we see a red tomato.
RGB Colour Model
How do we measure colour?
The eye sees colour using the RGB (red, green, blue) colour model. This is also known as additive colour or additive primaries.
Early colour photography was based on this system. Colour film was coated with layers of emulsion that were sensitive to red, green and blue. Television and computer screens also use this method.
When equal amounts of each wavelength are added, white light is reflected.
The cross over colours of yellow, magenta and cyan are known as subtractive colour or subtractive primaries.
Subtractive primaries absorb different light waves, hence equal amounts of each will absorb all light and will appear black.
This is the method historically used to produce photographic filters for all areas of photography:
- Magenta is used to filter green.
- Cyan is used to filter red.
- Yellow is used to filter blue.
We have established that the eye can only interpret colour based on the three different colour receptors in the eye; red, green and blue. All colours are reduced to these three sensory values based on broad wavelengths.
Colours can also be measured by reading their spectral curves using a spectrophotometer. This measures the spectral power distribution across all wave lengths. Incidentally, every colour will have an exact spectral match that will be constant in all environments.
In order to fully understand this, some information about Metamerism needs to be covered;
Metamerism is the matching of colours of different spectral power distributions. It occurs because the cone cells in the eye will give the same colour response to various combinations of light waves across the spectrum.
Put another way, two light sources made up of different wavelengths can appear to be the same colour irrespective of their spectral power distribution because they produce the same Tristimulus values.
As we have established, colours of different spectral power distributions can visually have the same colour appearance to the eye. Metameric failure occurs because of this.
The different types of metameric failure are discussed overleaf.
Main types of metameric failure
Directional metameric failure; This is when two samples match when viewed from one angle but look different when viewed from another. One example would be pearlescent coatings.
- Observer metameric failure; Varying degrees of colour blindness.
- Field-size metameric failure; Occurs when the cone cells vary from the centre of the eye to the periphery. So small colour areas can look different to large colour areas.
- Illuminate metameric failure; When colours viewed under one light source look different under another. This is the type of metameric failure that has caused cameramen and filter developers the most problems, due to varying light sources Unless the light source is pure white this type of metameric failure is unavoidable.
The Goblin shark is an extreme example of this, which in normal light appears to be a shocking pink colour. In its natural habitat, in deep water, it is a well camouflaged predator. This is because the only light waves available are blue and as blue absorbs red the Goblin shark appears black.
Neutral Density (ND) Filters
What is an ND filter?
It is a grey filter that is an intermediate neutral colour between black and white. In other words, a colour without colour. Historically, a combination of all three subtractive primaries are used to produce Neutral Density filters.
Metameric matches of ND filters worked well when using RGB sensitive film but this is no longer the case for modern digital cameras.
Modern camera chips are sensitive to slight colour shifts and any slight variance in the balance of the three subtractive primaries become evident.
The camera sensors are also responsive to varying degrees of Near infra-red (NIR) and longer wave bands of infra-red (IR).
NIR and IR pollution is a problem due to the popularity of high density ND usage. This is because so much visible light is blocked that the effects of NIR and IR become noticeable.
A Basic Understanding of the Spectral Curves
At this point, a basic understanding of the spectral curves would be useful. The following information explains how the various colour wavelengths are computed in the graphs.
As stated earlier, colours can also be measured by reading their spectral curves using a spectrophotometer. This measures the spectral power distribution across all wavelengths.
Here are the colour band widths in nanometres:
- Below 400nm - UV
- 400nm to 450nm - violet.
- 450nm to 490nm - blue.
- 490nm to 560 - green.
- From 560nm to 590nm - yellow.
- 590nm to 635nm - orange.
- 635nm to 700nm - red.
- From 700nm is the infra-red range starting with near infra-red up to 2500nm.
The full IR range goes much higher but the NIR range is the important one for this purpose.
400nm and below is near ultra-violet light. When high levels of UV are present, this can be seen as violet and give a bluish look to photographs. It is corrected by using a UV filter. UV dyes used to absorb ultra-violet light, depending on absorption characteristics, usually have a yellow tint.
For example a UV absorber blocking up to 420nm will have a yellow tint. A UV absorber below 400nm will appear clear because the human eye cannot see colour below 400nm under normal conditions. This UV band of light (300nm to 400nm) is seen by birds, insects, reptiles and fish. With regards to how fish have evolved this ability, this makes perfect sense because as pointed out earlier, the deeper you go in the sea the more wave lengths are absorbed until only the blue wavelengths remain. Incidentally specially formulated UV filters are also extremely useful for under water filming.
There is also middle UV and far UV whose wavelengths are below 300nm. These are filtered by the atmosphere except for a small amount of middle UV so are not relevant for this exercise.
Absorption Colours for Wavelengths
The graphs I have used to show the spectral data of neutral density filters show the light absorption at these wavelengths.
Note that photographic filters, as stated earlier, are made using subtractive primaries to block wavelengths at specific points.
The graph shows the levels of the colour absorption:
- 300nm to 400nm – UV absorbers.
- 400nm to 450nm – yellow.
- 450nm 490nm – orange/red.
- 490nm to 560 – magenta.
- 560nm to 590nm – violet.
- 590nm to 635nm – blue.
- 635nm to 700nm - blue/cyan.
- Above 700nm IR dyes are used. The ones reading near to the 700nm range are cyan in colour.Graphs Explained
Before we look at the Neutral density graphs I will explain how to read them. I mentioned earlier that subtractive primaries are used to manufacture Filters. Below are representative graphs of the three colours used to make traditional ND filters.
Because yellow is used to absorb blue the above graph is showing how much blue is being absorbed by the yellow.
If you look at the vertical reading (Abs) you will see a scale from 0.0000 to 3.000 each 0.3 increment of absorption is equal to a density reading of 1 stop of light. Respectively 0.6 absorption equals a density of 2 stops.
The horizontal line (wavelength nm) shows the wavelengths from 300nm to 800nm.
Note that the curve/broad band width of this yellow dye will remain the same but the absorption will increase or decrease depending on the level of dye present.
The same rules, as the yellow readings, apply to the magenta graph. This is showing a peak of 540nm. The magenta is absorbing green wave lengths.
The cyan graph has a peak of approximately 670nm and is absorbing red wave lengths. When all three colours are blend in equal proportions you produce a neutral density dye.
See the following neutral density graph:
Neutral Density Graph
This ND is made up of the yellow, magenta and cyan subtractive colours.
The vertical scale of absorption show the degree of blue green and red that are being absorbed. In this instance an absorption of 1.8 or 6 stops. The horizontal wave lengths scale show the wave lengths being from 350nm to 800nm. If you look at where the line starts to decline you can see that more red light is being passed or, put another way, there is less cyan present.
As stated earlier. This method worked fine for film and lower densities when used with digital cameras.
NDs for film cameras
So to recap. The graph seen here is a spectral curve of an ND that worked well on film cameras and on digital cameras at lower densities.
Note again that there is a slow drop off as we get to the red end of the graph.
At lower densities this shift is minimal. As density increases, however, this starts to give problems with NIR and IR pollution because the sensor can now see more of the ‘invisible’ red end of the spectrum once the visible light has been reduced. Modern camera sensors also record this information giving the magenta or red/brown cast to the images.
Any variance in the absorption of each wave length will also be intensified.
Digital cameras (Part 1)
The first solution to dealing with this problem was by combining a Hot Mirror with a Neutral Density filter. This worked well on the early Red camera as most NIR and IR pollution was removed, but the red channel was maintained in order to ensure accurate reproduction of skin tones. The downside was that the Hot Mirror caused vignetting in wide angle shots.
You can see in the graph (i) how the Hot Mirror gradually fills in from approximately 700nm. Next is a competitor’s graph (ii) and their approach to rectifying this problem.
Note that different filter manufacturers have approached this NIR and IR problem slightly differently. Some have used IR dyes combined with visible dyes. Notice the high levels used of NIR/IR cut dyes/coatings regardless of the density of the ND.
In tests by cameramen it has been noted that clipping of the red channel sometimes occurs because of the high levels used to correct the NIR and IR problem.
Digital cameras (Part 2)
Another approach we have used to solving this problem was to blend a combination of dyes which gives even attenuation at all wavelengths up to 750nm approximately. This was done to try to emulate a true grey with minimal variation across the wavelengths and to avoid clipping the red channel due to excessive levels of cyan. This method, working with the camera’s built in NIR/IR filter, has been extremely successful. If you compare to the other graphs shown you can see the improvement.
Note that the attenuation is pretty even across the wave lengths, especially from 650nm to above 700nm.
Digital cameras (Part 3)
Historically there has not been a ‘one size fits all’ filter to solve IR pollution as the chips in digital cameras vary from one manufacturer to another. Some cameras respond better that others when using different manufacturers filters.
Note also that cameras made before 2004 had no or very little IR filtration built in.
So, how can we improve further to prevent minor colour balance variations, varying results from camera to camera and metameric failure?
The only way is to produce an exact spectral match that responds precisely to varying light conditions and cameras requiring the use of neutral density filtration.
Due to the dyes that have been available to the photographic filter manufacturer, it has proved extremely difficult to produce true spectral matches that have even attenuation across all bandwidths and under different light sources.
The aim of the above graph is to get equal proportions of all lightwaves so that the readings are as close to the centre as possible. See above for how additive colour works.
Formatt Hitech has undertaken a lot of research into finding a true spectral match that gives an even absorption and reflection across all wave lengths rather than having to rely on metameric matches. This has been achieved by using Nano technology to coat an optically clear glass.
If you look at the two graphs opposite you will notice in graph (i) how straight the line is. And again in graph (ii) from 300 to 1100 nm.
The graph speaks for itself. This is the closest neutral density (non-colour) spectral match available. It absorbs all colour evenly across the spectrum and into the IR range.
The beauty of this method is that we have a ND that not only solves NIR and IR problems, but does not suffer from metameric failure under different lighting - and works with any camera.
Formatt Hitech were the pioneers in bringing this technology to the photographic and broadcast market and continue to develop its filter range for a continually demanding and changing market.
Note: Firecrest and Firecrest Ultra use the same technology to produce a high quality ND filter. Both also use the same optical grade of polished glass. The difference is that the Ultra range goes through a secondary polishing process, known as "Lapping & Polishing" to produce a filter which is optically flat and so delivers a high level of image sharpness previously only available for the broadcast market.