Newton and Color Wheel
In 1670, Sir Isaac Newton demonstrated that white |
light is composed of all the colors of the
spectrum by refracting sunlight through a
The visible spectrum displayed as a wheel.
The "primary" colors are made from equal amounts of the adjacent colors,
and colors on opposite sides of the wheel are exact chromatic
opposites. Mixing any two opposites anywhere on the wheel gives you grey.
For example, primary red is made of equal amounts of yellow and magenta, and is the chromatic opposite of cyan. Red plus cyan equals grey.
Additive and Substractive Colors
Colored light can be created in two ways. Starting with no light, and gradually adding red, green and blue until you get the color you want, or starting with white light and subtracting what you don't want with cyan, magenta and yellow filters.
Pigment-based color systems are subtractive (YCM), while light-based color systems are additive (RGB). Film is pigment-based. Your monitor (surprise) is light-based.
Offset printing in CMYK is subtractive, as was classic 3-strip Technicolor, which printed with YCM dyes. While motion picture film can be printed with either an additive or subtractive lamp house, an RGB light source is much more controllable.
Your eyes are capable of detecting about 450 luminance values.
Unfortunately, monitor phosphors are limited in dynamic range, and display only 256 values. This is why computer image files are in 8 bit color space.
Just as each digit in our base-10 numbering system is 10 times the one to its right, in a binary system each digit is twice the one to its right.
So an 8 bit "word" gives you a maximum of: 128 + 64 + 32 + 16 + 8 + 4 + 2 + 1 = 256 (with zero) possible luminance values.
Obviously, the larger the word, the more color information can be stored.
A 16 bit word can display 65536 luminance values, way more than can be seen or displayed. Because Cineon image structure is in log space, it only needs 10 bits (1024 values) to do the same thing.
The advantage is in areas of subtle gradation, where 8 bit hasn't enough values to show all the transitions, resulting in banding. Film grain can camoflage banding, but only to a point.
|16 bit color depth||8 bit color depth|
Surprisingly, tests have shown that an observer's perception of sharpness is more dependent on bit depth than on resolution.
Log and Linear Color Space
Of all the differences between how film reacts to light and your monitor displays it, the major one is this: film is logarithmic and the monitor is linear. So what does that mean?
On a monitor, there is a one-to-one correspondence between energy (think exposure) and brightness. Each time you increase the signal to the monitor by 1 volt, you get exactly the same incremental increase in brightness.
On film, however, the increase in brightness (emulsion density) is a result of the logarithm of the increase in exposure.
Here is the graph of the response of Eastman 5248 to light, called its characteristic curve.
Notice that there is both a maximum and minimum possible density
on the film, no matter how much or how little light strikes it, because
there is a finite amount of light-sensitive silver in the film, with a
minimum threshold sensitivity to light. When it is all exposed, all
the light in the universe won't make it any denser, nor will less
light than it can see make it thinner.
Notice, too, that this is not a linear change, but happens gradually as exposure eases in or out. This rolloff is called the "toe" of the curve for low densities, and the "shoulder" for highs.
As with most things, this will be more clear if we draw a picture...
|Linear Image||Log Image||Difference|
At left, the linear display of your monitor. In the center, the log sensitivity of film (exagerated for clarity). At right, the difference between the two.
Only when images must move from one medium to another are there problems.
|Original Image||Linear image in Log Viewer||Log image in Linear Viewer|
|Accurate tonality||Lows suppressed and highs accentuated.||Highs flattened and lows boosted.|
Moving images from one color space or bit depth to another requires the use of a "lookup table", common called a "lut", which accurately remaps the data. It works just like a graphic equalizer does for sound systems, when correcting for speaker flaws and room acoustics.
Monitor and Film Structure
|Monitors create color by shooting electrons onto a glass plate which is coated with adjacent colored phosphors which glow. Your eye blends the 3 primaries together.|
Film is 6 layers of emulsion, 2 each for yellow, cyan and
YELLOW gives BLUE
MAGENTA gives GREEN
CYAN gives RED Each layer works like this:
The yellow layer records color information where blue isn't, because blue is the sum of magenta and cyan, and the adsence of yellow.
If that's not confusing, nothing is.
Its important to understand that each medium has limitations on the accuracy of the colors it can represent. This is especially critical in the video realm, where colors outside the triangle either bloom or don't display at all, and are therefore termed "illegal".
To demonstrate this, we will use the CIE color co-ordinate system developed in France in 1931, which displays all the wavelengths of light visible to humans.
|CIE graph of visible light||TV and monitor phosphors||Colors seen by film|
Computer monitor phosphors are only slightly better than those in consumer TV's. Compared to these small sub-sets of visible color, the fidelity of film (left) is astonishing.
|Few concepts in colorimitry are as consistently misunderstood as gamma, which has several meanings, depending on the context.|
|Y change / X change = Film gamma||Amount of deflection = Monitor gamma|
"Gamma" in the film world refers to the slope of the straight-line portion of the characteristic curve, usually around .6 (although each color layer is slightly different). Middle grey (18%) is pegged, and the black and white points vary from emulsion to emulsion. A higher number means higher contrast.
On a monitor, however, "gamma" refers to the amount of deflection of a curve with its black and white points pegged. This deflection increases as you get further away from gamma 1 in either direction. Default gamma for most consumer computer devices is 1.7, which is a problem if you're making a movie at gamma one.
Like we are...
|Rendered Gamma 1||Adjusted paper printing Gamma 1.7|
The paper printer expects the image to be encoded for gamma 1.7. Since our renders are at gamma 1 for output to film, the image must be adjusted for this colorspace or they will print too dark.
In Photoshop, select IMAGE/ADJUST/CURVES, click anywhere on the line, and put 75 in "Input" and 128 in "Output". The image will appear washed out on your monitor, but will print correctly, matching your pre-gamma version.
By the way, if you're still wondering how a lookup table works, you just used one.
Flat Bed Scanner
|Gamma 1.7 scan||Gamma .58 compensation||Curves cancel||Image corrected to gamma 1|
Just the opposite happens when using imagery digitized on the flat bed scanner, which also operates at gamma 1.7.
Use the same procedure to apply the inverse of 1.7 (.58) to your scan.
This time, put "128" in the INPUT box and "75" in the OUTPUT box. This will correct the scanned image for gamma 1 colorspace. Now you can make whatever adjustments necessary to conform your element.
|Graph of a gamma 1.7 image "corrected" by using brightness.|
Top 5 Reasons Not To Blindly Trust Your Monitor (Even if it is calibrated...)
- Film is pigment-based YCM, monitors are light-based RGB.
- Monitor is linear color space, film is log.
- Phosphors have much less dynamic range than film.
- Displays 16 bit images as 8 bit.
- Shows more shadow detail than the film will.
by Greg Kimble