tag:blogger.com,1999:blog-1840985738235902482.post5724395428358688700..comments2024-03-27T07:14:48.488-04:00Comments on John the Math Guy: RGB into LabJohn Seymourhttp://www.blogger.com/profile/11350487038873935295noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-1840985738235902482.post-37394040334175966122019-12-09T07:10:20.113-05:002019-12-09T07:10:20.113-05:00I think for my purposes of making wall art that ma...I think for my purposes of making wall art that matches decor in the room, I can compare Munsell values on the smartphone and come up with a pretty decent result in a mixed paint at Lowes. <br />The challenges of measuring color are many, and I won't live long enough to solve them all. I studied art with Henry Hensche, even made a movie about him that's on YouTube. His view of art history is that we are now at a time, based on Claude Monet's work; showing the same scene in different times of day and weather conditions, that we MUST be aware of the "time of day" to really communicate the truth of what we're looking at. Colors are affected by that, no doubt.<br />Your experiments really show how lighting affects the color. <br />Our human desire to do things "perfectly" seems out of place when all we're looking to do is produce a wall art that fits a certain space and decor.<br />Have you seen on the Munsell color blog the lady that incorporates Munsell values into her designing of carpets and rugs? She found a guy in India I think that has swatches it many colors, but nowhere near the many possibilities that Munsell designs.<br />Keep up the good work. If we live forever we can solve these problems, but until Jehovah God's Kingdom comes, we'll have to just do our best. For more information about God's Kingdom, check out jw.org.Phillip St. Johnhttps://www.blogger.com/profile/16497798864875800938noreply@blogger.comtag:blogger.com,1999:blog-1840985738235902482.post-81281870310432371562014-07-17T11:30:28.760-04:002014-07-17T11:30:28.760-04:00Thanks Dave.
I am going to disagree on your comme...Thanks Dave.<br /><br />I am going to disagree on your comment about doing a "credible job".<br /><br />Once upon a time, there was a color control system for printing presses that used an RGB camera. It did a fairly decent job at measuring density, but when ISO 12647 started pushing people toward measuring CIELAB, I undertook a one-year project to try to squeeze accurate CIELAB out of an RGB camera. <br /><br />I had the benefit of ten man-years of development of a solid RGB photometer (not a simple task), and I had the benefit of two other smart guys who each put about six months of effort into this specific goal. Based on this, I claim that I have met Niels Bohr's criteria for being an expert in the art of RGB to CIELAB transforms. ("An expert is someone who has made all the mistakes possible in his narrow field.")<br /><br />The end result is that 12 man years of effort says "it won't be good enough". We spent ~$1M on developing a true spectro.<br /><br />The problems are that a) while the collection of all CMY spectra is theoretically three dimensional, it's not linear, b) the addition of K makes it at least four dimensional, c) dot gain is not linear, so a change in dot gain introduces a bunch more spectra, d) gloss of the paper will in effect add "white" as another dimension, and e) the color of the paper adds yet another dimension. Of... and then there's spot colors!<br /><br />ISO 13655 defines that a spectro for measuring the color of print must have at least 15 channels. <br /><br />That said... I determined (in later work) that within a press run, the deviation of color can be characterized into delta L, a, and b by an RGB camera to within a reasonable tolerance. That means that if there is some other mechanism for bringing the color up during makeready, then the RGB camera can monitor color from there. But, the big savings for a printer is during makeready - saving waste by getting to color ok quickly.John Seymourhttps://www.blogger.com/profile/11350487038873935295noreply@blogger.comtag:blogger.com,1999:blog-1840985738235902482.post-44066373066105244952014-07-17T09:47:03.805-04:002014-07-17T09:47:03.805-04:00Great article John, as always - there is so much g...Great article John, as always - there is so much good stuff here.<br /><br />One thing from near the beginning, I think that when properly configured, and RGB camera can do a credible job of printer control. "Properly configured" means that the camera characterization understands the (presumably 1:1) mapping of CMY to color. Then it is "just" a question of mapping camera RGBs to printer CMYs. Not trivial, but possible, and the delta E's should stay well below 100.<br /><br />Also, for anyone who want to delve further into use of three sensors to predict arbitrary color, ask The Google about the "Luther Condition."<br /><br />- DWAnonymousnoreply@blogger.com