Color: why we'll never get it exactly right

[ReducedAirFact]

First let's establish that they're your planes and you can paint them any way you want. If you want a copper and violet Albatros with flames along the wing edges, go for it. It'll draw a lot of attention at conventions, it'll be fun to fly, and guess which plane the new-to-the-game pilots will fight over?

If you're aiming to match historical colors (or "colours" for our Commonwealth friends), you'll never get the perfectly correct color either, and that's the topic of this post. But there's close and there's not-close: we may debate over the best match to the RFC's khaki brown-green PC10, but it's surely not a bright lime green color. So the goal is to get reasonably, plausibly close to historical colors (as best we can) without getting too pedantic about getting it "exactly right".

There are many reasons we're not going to get it perfect. Some of them are driven by color science, others by logistics, others by statistics, others by lack of historical data.

Variances in the Original Colors: Let's pretend we could somehow travel back to 1916 with a colorimeter. I'm sure if we went from plane to plane, we'd find variances even where colors are supposed to be consistent. There were variances in the factories where the paints were manufactured, either due to material shortages and substitutions to simply not being overly worried about being perfectly exact. Formulas changed mid-war, e.g. the British changed the red in their roundels in 1918. The colors were hand-painted or spray-painted onto the plane, and variations in thickness or texture can affect the paint's look. They were painted on different underlying materials, e.g. linen or plywood or metal (and if you've ever switched from black to white primer, you know how that can change things). Once the paints were dry, the were frequently varnished-over, and the varnish formula and thickness can change the surface color. Finally, they were weathered to various degrees, whether it's exposure to oxygen, sunlight, rain, engine oil, or wind-friction. So even in 1916 you're going to find plane-to-plane variances.

Aging of our Samples: We modern historians don't have it that easy, though, we're left with a scattering of original fabric, and it's roughly 100 years old. If you've seen old flags or dresses in museums, you know how aging can affect color. So the surviving samples we have all have seen various (and unknown) levels of fading and color-shift.

Metamers: Even if a color were 100% consistent and we had a sample stored in a time-vault, there are problems of color science itself. The color of a surface will change depending on the lighting conditions. Two paints can look exactly the same under one spectrum of lighting (e.g. noon sunlight) and very different under a different spectrum of lighting (metamerism). A white ball and a red ball will look exactly the same under a pure red monochromatic light. But if take that light just 5% toward yellow, it's very evident the difference. So even if we could find an *exact* match for a historical color under daylight conditions, it probably isn't an exact match under artificial lighting, and it probably even looks different under cloudy conditions or dawn/dusk sunlight. For an exact match, you'd probably have to formulate the paint using the exact same ingredients, e.g. 250 parts iron oxide to 1 part lamp-black by weight mixed in cellulose or oil varnish. (I doubt even museums go to that length.)

Two balls look exactly the same under red monochromatic light. Take that 5% toward yellow, and a difference is evident; under white light it's obvious (this is just an oversimplified illustration of the point).


The color spectra of various "white" light sources (from olympusmicro.com):


Perception plays a role. Everyone's eyes and color perceptions are a little different, so what might be greenish-blue to me might be bluish-green to you. Hopefully we can minimize that problem by avoiding subjective, textual descriptions of colors such as "Sea Green". Plus, the lens in your eye darkens as you age, which will change your color perception.

The Scale Effect: Furthermore, with scale models, to match the way you perceive full-scale models at a distance, some modellers recommend you reduce the contrast and add to the "whiteness" of the colors, increasing that effect the smaller the model is. One reference cites adding 23% white (or neutral gray) to a 1/144 model to simulate what you would perceive if the model were full size. Some model paint companies even formulate the colors to already compensate for the scale effect (at the most common scale for the paint in question). [Author's note: I'm a bit of a skeptic on this one.]

---

Describing color to each other: But let's put all that aside and say I had, sitting at my desk, an exact, perfect, unaged copy of a paint from a 1918 SPAD 13. How do I communicate that color to you? A computer monitor makes a poor communication mechanism. For one thing, your monitor and my monitor would have to be precisely calibrated; otherwise what looks like bluish-green to me might look like yellowish-green to you. For another thing, your monitor and mine are both limited by the colors in the LEDs they use -- they can only show colors that are linear combinations of those three colors (the "gamut" of the monitor), and the gamut of monitors is much smaller than what a human can perceive. And -- even if we get the hue correct -- the brightness of the color probably varies depending on the monitor and the room lighting.

The yellow triangle might represent colors colors your monitor can produce; the whole colored area represents colors a human eye can perceive.


Furthermore, as you move your head to the left and right or up and down, you'll probably see the colors shift on your monitor. Hence, if you're using a monitor to assess a color, you may see it differently when it's on the top of the monitor than when it's on the bottom. Computer monitors are a terrible way to try to communicate colors.

Summary: Probably the best we can do is to communicate via commercially-available paint and ink colors. There are a lot of different color books and standards and fan decks and web sites with paint and ink chips (though the most popular one for WWI colors, the Methuen Handbook of Colour, has been out of print for a long time). Some of the hobby paint manufacturers are kind enough to include more detailed color information and/or translation tables (though they frequently change their formulations). But even Pantone (a big commercial color firm) recommends replacing their color-chip books every two years to avoid aging. (Which is no doubt lucrative for them as well as scientific). And most commercial paint companies change their paint selection and/or formulation every few years.

My hope is to describe a color and where it was used, tell the source of that color description (e.g. "Windsock Datafile 39, Junkers J.I, says the Purple/Mauve used on the upper surfaces was Methuen 15D4/15E2/15E3/17F8/18D4/18D6 or FS595 37100/37144") and translate that into various other forms, e.g. closest commercial paints (both hobby [e.g. Humbrol & Vallejo] and house paints [which come with handy, free chips from the store]), closest RGB values, standards (RAL Standard, British Standard, Federal Standard 595, CIELab), etc. I'll be using sources such as Windsock, a copy of Methuen, and a Nix color sensor to try to put together some of this data. (And of course any help is appreciated.)

Again, when it comes to historical paint color, there is no perfectly-correct match -- the goal is to just get close and say, "good enough!"