In week nine of my color grading course at CalArts I’m looking over a student’s shoulder at her monitor. The shot is golden hour light coming through a kitchen window, everything drenched in amber. “This is too warm,” she says. “We need to cool it down.”
I ask her what she means by warm.
“It’s orange. Warm is orange.”
I ask her what the shot makes her feel.
She thinks about it. “Comfortable. Like being home.”
“Then why do you want to change it?”
Some version of that exchange happens every semester. The same assumptions come up every time, and they come from how we learn to talk about color. The vocabulary gets in the way of looking at the image.
Treating “warm” as physics
Color temperature language runs opposite to the physics. Hotter objects emit bluer light. A candle flame sits around 1,800 Kelvin and reads amber. Midday sun is 5,600 K and reads white. A clear blue sky pushes past 10,000 K. The hottest stars are blue.
We call the orange end warm because the metaphor comes from the body. Fire is orange and it warms your skin. Deep water is blue and it takes heat out of you.
The metaphor is useful. The trouble starts when beginning colorists treat it as physics, because then the vectorscope stops matching their intuitions. They push a shot toward orange to warm it up when the image needs magenta in the midtones, or a small lift in saturation with no hue shift at all.
The fix is to learn what the numbers mean and then look at the image instead of the numbers. That takes a lot of practice.
Assuming warm means pleasant
The early assumption is that warm tones are inviting and cool tones are clinical, so every scene with people in it gets graded toward amber whether or not the story calls for it.
Take a film about a character who is pulling away from her family. The dinner scenes come back glowing like a Hallmark card, and the reasoning is that it’s a family, and families are warm.
The character is suffering, and the grade says she isn’t. The scene needs to be warm enough to read as domestic, with something slightly wrong in it: a green cast in the highlights, or shadows a little deeper than the room accounts for. The storytelling is in the gap between what we expect a family dinner to look like and what this family feels like.
Warm as pleasant is also a local assumption. In the Chinese five-element system red maps to joy, white to grief, and green to anger. Victor Turner’s work on Ndembu ritual describes a color system organized around red, white and black, in which red stands for both life-force and spilled blood.
Assuming the audience reads color the way you do
New colorists grade as though everyone watching shares their associations.
White means purity in Western traditions: wedding dresses, angels, clean slates. Across much of East Asia white is the color of mourning. A colorist who floods a scene with white light to signal innocence has coded it with death for part of the audience.
In the suite this means asking who is going to watch the film, where, and what visual language they already carry. A colorist serves the director and the audience at the same time, and the audience is not one culture.
Grading by anxiety
This one does the most damage to footage, and it comes from a good place. People new to the tools want to demonstrate competence. They have learned lift, gamma, gain, curves, qualifiers and power windows, and they want to use them, so they correct footage that was already exposed and lit the way it was meant to be.
I call it grading by anxiety. A beginning colorist looks at an image that is working, feels she should be doing something, and starts pushing. She neutralizes a natural color cast the DP waited hours for. She crushes the blacks to match a reference off YouTube. She puts a power window on a face that was deliberately lit half in shadow.
I run an exercise for this. I give the class beautifully shot footage and tell them to grade it. Then I put the original next to their version and ask them to vote on which is better. The original wins about half the time.
Often the best thing a colorist can do is leave the image alone. That takes longer to learn than any of the tools.
Trusting the scopes over the body
This is the hardest one to shift, and my training in clinical psychology is what I fall back on.
The body reads an image before the conscious mind has anything to say about it. When you look at a grade and something feels off, that feeling is data. Your retina converts photons into opponent-channel signals (red-green, blue-yellow, light-dark), compresses them, and sends them to a cortex that compares them against everything you have ever seen. This happens in milliseconds, below awareness, and what reaches you is a feeling.
I watch people override it constantly. They look at a grade, feel uncomfortable, check the scopes, see numbers inside the legal range, and accept the grade. But scopes measure signal, not meaning. A waveform monitor will tell you the highlights sit at 95 IRE. It cannot tell you whether they feel like four in the afternoon or like an office ceiling.
Ignoring that discomfort because the vectorscope looks right is like ignoring your ears because the meter says the level is fine. I tell the class to trust the tightness first, the sense that something is wrong before they can name it, and then use the tools to find the cause.
For a while I taught the cultural side of this as a two-layer model, with hardware underneath and culture on top. Red arouses you, and your culture decides whether that arousal means passion, danger, luck or sacred power. I stopped teaching it that way because the research is messier than that.
The studies behind the red claim are old and small. Wilson in 1966 found red more arousing than green, and Jacobs and Hustmyer in 1974 found red above blue and yellow. Both measured skin conductance in small Western samples, and Jacobs and Hustmyer found no effect on heart rate. The research on melanopsin, the photopigment in the retinal cells that set the body clock, points the other way. Cajochen and colleagues in 2005 found that blue light at 460 nanometers raised alertness and heart rate more than green light at 550 with the same photon count. So the body does respond to color before culture gets involved, but which color comes out as more arousing depends on what you measure.
The cross-modal findings are sturdier. People reliably match higher pitch to brighter, lighter colors. In the suite, a warm color grade under cold sound design produces tension because the picture and the sound are telling the viewer different things.
Culture does most of the remaining work, and a vectorscope shows none of it.
Back in week nine, the amber kitchen is still on her monitor. The vectorscope trace sits well off center toward orange, which is accurate, and it says nothing about what the shot means.
She told me the shot felt like being home, and then she told me it was too warm. The first response was her own. The second was vocabulary she had been taught.
What I want at that point is another minute of looking before anyone touches a wheel, because her first answer was the accurate one.
References
Turner, V. (1967). “Colour Classification in Ndembu Ritual,” in The Forest of Symbols: Aspects of Ndembu Ritual. Cornell University Press.
Wilson, G. D. (1966). “Arousal Properties of Red versus Green.” Perceptual and Motor Skills, 23(3), 947–949.
Jacobs, K. W., & Hustmyer, F. E. (1974). “Effects of Four Psychological Primary Colors on GSR, Heart Rate and Respiration Rate.” Perceptual and Motor Skills, 38(3), 763–766.
Cajochen, C., Münch, M., Kobialka, S., Kräuchi, K., Steiner, R., Oelhafen, P., Orgül, S., & Wirz-Justice, A. (2005). “High Sensitivity of Human Melatonin, Alertness, Thermoregulation, and Heart Rate to Short Wavelength Light.” The Journal of Clinical Endocrinology & Metabolism, 90(3), 1311–1316.
Spence, C. (2011). “Crossmodal correspondences: A tutorial review.” Attention, Perception, & Psychophysics, 73(4), 971–995. Collects Lawrence Marks’s work on light-and-sound associations, including the pitch-to-brightness correspondence.
Related reading
- The Timebase Corrector and the Traumatized Brain: signal, noise, and what a body registers before the instruments do
- Where the Feet Need to Be: NTSC’s bandwidth limit read as a palette instead of a prohibition