I struggled to classify this post… my most recent “series” post has been on lipstick destashing- actually detailing the process of determining dupes, via Excel, in my collection. I’ve felt the need to explain color models recently, as different color models hold different advantages (or disadvantages) when it comes to finding dupes.
I wasn’t sure if I should include this post as a Science! post or as part of TGLD… in the end, I’m going with TGLD because the strengths and weaknesses I’ll be talking about here are related to lipstick destashing.
This post isn’t going to get into how to determine dupes. This post is just about determining what color model is best to work with.
RGB Color Model: Using Red, Green, Blue Components
- 256^3 (nearly 18 million) color options- great resolution
- Easy to conceptualize and find
- Standard across all devices; can easily put an RGB color into any document, so you can organize your lipsticks in an Excel or Word inventory with a “color swatch.”
- Formula to find “similarity” can be created (distance between two colors is a single line that is easily computed)
- Easily sorted
- Focuses only on color components, so it’s not as easy to see hue (color family) or how muted a color is.
- Need to do other calculations to see how light/dark a color is
- May be too many colors- hard to classify where to draw the line between different colors
HSV (HSB) Color Model: Using Hue, Saturation, and Value
- Separates colors into what we look for most: hue (color family), lightness, and saturation/purity.
- This makes it easier to determine similar colors
- Allows us to look at what is “warm” vs “cool” in a color family
- Not as many colors, so there’s less “overlap” to look for (may make analysis easier)
- Can establish categories for color family, lightness, and purity to make analysis even easier.
- Not all programs use the same scale for this model
- Converting between RGB and HSV can be time consuming
- Value/lightness is not always consistent- eg, the lightest blue at hue 180 still looks very dark to our eyes.
- Because of the lightness and saturation/purity, colors can have very similar hues but look quite different
HSL Color Model: Using Hue, Saturation, and Luminosity
This model is very similar to HSV, but Luminosity is slightly different and Saturation is more dependent upon Luminosity
- All the strenghts of HSV, plus
- Luminosity also considers the darkest color component, so it may be more accurate to our vision.
- A luminosity of 1 (or 100%) = pure white and 0 = pure black. This makes it easier to compare lightness/darkness (IMO).
- Saturation level depends upon lightness, so that may be more accurate as well.
- More difficult to convert
- Often is not used
- When it is used, gets confused with HSV/HSB because hue remains the same.
HSI Color Model: Using Hue, Saturation, and Intensity
Similar to HSV other HS_ models, but Intensity is just an average of all color intensities. This makes it easy to see how far a color is to white or black. This model also has a different version of saturation which maintains some difficulty
HCL/LCH/LUV Color Model: Using Hue, Chroma, and Luma
This color model, is touted as the “best” one, often adjusting for the weaknesses in the HSV/HSI/HSL models. It is actually very closely related to CIELUV, another color model, but it translates it to polar- which makes it a little more intuitive for us to use.
- Best model to characterize colors (so this is the best for intuitive analysis for our purpose)
- Keeps hue on the same 360 color wheel
- Adjusts problems related to saturation and perceptual lightness (now called chroma)
- Adjusts problems related to intensity, luminosity, or value (now called luma)
- Device independent, which doesn’t really matter for our purposes, tbh…
- Conversion between RGB and HCL/LCH/CIELUV are more complex and difficult to easily write in an Excel formula
- You can easily find a calculator online, but this means you’d have to type everything in manually, which takes a while.
- There is a formula for Luma based off of RGB that is pretty straight-forward; however, this formula is more of an approximation. It’s a simple weight-based formula to account for the difference in perceived lightness in the red, green, and blue components.
- This formula is: Luma = (0.299*R+0.587*G+0.114*B)/255 (if using RGB values from 0-255)
- Not common in “laymen” programs. While this color space is very common in professional settings, it’s much less common to find settings to adjust it- so for us, we’d just be working with some base numbers and not be able to “see” what those numbers themselves mean.
A Word About Other Models (CMY/K, CIE-LUV, XYZ, etc)
Other color models do exist. For instance, CMY/K (cyan, magenta, yellow, and black) is very common for printers. For many of these models, application plays a role in what model you select. For our purposes, I’ve determined most of these models simply aren’t ideal
So, what model should you work in for a swatch analysis?
You can choose your own, of course, but it really depends upon just how close you want this to come to finding dupes and how much time you an invest in it. It is my personal belief that the “best” model is to use hue and saturation from the HSL model and the approximate Luma formula from the HCL/LCH.
My reasoning for this is simple: HSL is easy to compute from RGB, and the approximate Luma formula is easy as well. To me, the time it takes to calculate out HSL and Luma is well worth the “better” values. While LCH is, definitely, better, the time it takes to get there is just not worth it- remember, we aren’t looking for anything to be perfect, just “good enough.”
RGB is, honestly, the worst color space to work in because it requires a certain “eye” to be able to tell how close colors actually are and determine things like color purity or mutedness and intensity/lightness.
All of that said, it’s your choice. Work using a color model that you are comfortable in!