Sorting and Tapering

We have a seemingly insurmountable problem here in River City.

Pearlescent paints and coatings present some of the most difficult color control problems one faces as a colorist. By their very nature, in order to look iridescent, the carrier vehicle must be relatively translucent so one sees the reflective particles. The result is often hyper-sensitivity to the effects of contamination, color drift, and any yellowing or instability during the heating cycles of your process. Large scale, multiple opening, hot press lamination subjects sheets to significant heat history variations both within a single sheet and among those at different locations in the press load. Putting these situations together along with an application where adjacent panels touch often resulted in "off-color" complaints. It was agreed between the parties that visual sorting would be done with the sheets being serialized and used in color sorted sequence. Sounds like a workable solution doesn't it?

Anyone who has worked very long at approving colors is painfully aware of the simultaneous contrast nature of human color judgment. When presented with a group of samples, as few as three or four, one must not be too quick to jump to conclusions, since the orientation and order of placement affects our perception of the differences. Try viewing samples three at a time and you'll likely discover that, when the colors are close, the one in the middle usually looks out of place even after repositioning the samples. Even two samples, one on top of the other, usually reverse for lightness when inverted. Try this experiment sometime - take a small sample of uniformly colored material, say an 8 x 10 inch sample, cut it in half, and without disclosing what you have done, ask people to evaluate the quality of the color match. Odds are only about 50/50 or less they'll say they are the same. Be careful though, some people don't take too lightly to the humbling experience of making a wrong color call. Over the years I have had to deal with some angry people at times when the "trick" is revealed. It might be best to accept the comments and just say thanks. It does illustrate the point that applying numbers to small color differences is best left to machines.

Still not convinced. Take a look at the following examples. At first glance, there doesn't seem to be any really big differences between adjacent panels. One might note that the right end is lighter than the left, although the magnitude is hard to judge. These colors have been color tapered to minimize the differences between each swatch and its nearest neighbors.

The maximum color differences among these swatches are 1976 CIELab DE = 4.87, CIE DE2000 = 4.58, DECMC1:1 = 4.90, DECMC2:1 = 2.46. No, I didn't slip a decimal point! They are really that bad and, if not for color sorting, we might have a real mess. Try putting your cursor on one of the end swatches and dragging it to the other side. In fact, all of the swatches are drag capable so you can mix and match as you wish. Swatches on the left drag under those to the right. Those on the right drag over. Your browser refresh button will restore them to their original positions.

Above are the same patches without the 1 pixel wide, black margin. Removing it further enhances visual detection of differences, so try dragging and comparing this set. The differences will now start looking more like 3's and 4's.

The following are the above patches arranged in a different order to illustrate how effectively proper sorting can make a bad situation look better!

The visual concept illustrated here is what made the computer sorting system such a valuable tool. By properly sequencing a series of samples, one can utilize a group of widely varying colors in such a manner that the overall, final product is perfectly satisfactory.

Find out more about this sorting thing.