Quantitative color calls

Some initial experiments

• CIE L* a* b* color space
• Reasonably priced ($240) colorimeter
• 4 or 5 part blog about statistical process control of color. Author is a color scientist.
• 2 part blog about multi-dimensional statistics (such as in LAB color space). Same author.
• can RGB be turned into LAB?
• cheap colorimeters?

Data gathered from images taken in the QC shack with a Sony RX100mk3 camera. Images were processed with a Python script that (1) cropped each image to a central region containing only glass, (2) randomly sampled 1000 pixel color values (in both RGB and LAB color spaces), (3) calculated the mean and standard deviation of each color channel, (4) saved this data table to disk, (5) created a 3D interactive scatter plot of color values in LAB space.

Additional images were collected from the Bullseye website of 171 "base" colors, including cathedrals and opals. They were processed identically, though with a different crop region.

Here are some plots.

1. The RGB color space, in 20 unit "steps" (eg (0,0,0)->(0,0,20)->(0,0,40)->...).
2. Styles from website
3. Opal color samples, from camera, slightly overexposed (probably).
4. Opal color samples, from camera, better exposure (probably).
5. Cathedral colors samples, from camera, questionable first trial.
6. Styles from website, 100 pixels from each image to show variation in pixel values within the image.
7. Opal color samples, from camera, 100 pixels from each image to show variation in pixel values within the image.

Camera settings