Scott #63 Color Variations

Also a New York cancel.



Makes me suspect that a batch of these were sold there.

The ultramarine shade has been controversial, if it exists. The most recent, March-April 2018 Volume 98 - Number 2 The Collectors Club Philatelist article "Ink & Paper of the U.S. 1869 3¢ Pictorial (Scott 114)" part 2 compares Scott 114 with Scott 63a examples said to be ultramarine. The article also references Steve Pacetti's article debunking the existence of a 63a.

While one or two of the stamps resembled certified 63a examples, all of them, inlcuding the one I own, may be a dull shade of blue. The examples in the article were found to be prussian blue when a Fourier Transform Infrared Spectroscopy (FITR) test was applied.

While it may be possible to find new examples of some rare stamps, it is impossible to find stamps that never existed.

I thought that I had finally located a copy of what Doug is calling pastel blue. When I got it in it was considerably darker than the online image showed. It is the stamp in the middle with a dark blue to the right and an ordinary though somewhat light blue #63 to the left. I will call my new stamp a bright blue. It is a beauty.



Same stamps shown with a different but ordinary #63 at left.

I agree, the color is amazingly blue. It is also interesting how different the middle stamps' hue appears just by the change in background hues from the yellow to white inj your two photos/scans. The top one seems darker to me.

I agree - from what I can tell, Bright Blue seems to be the right choice for your stamp.

It is amazingly beautiful.
There are some really pretty blues on this stamp for sure.

I was wondering about the results of coping the image(s) from this posting and saving it in BMP format (i.e. "Paint"), and then extracting the RGB matrices and comparing the colors of each of the blue and green matrices in that matter.

This would eliminate the factors of human perception of the eye/brain, and the differences in computer monitors.

Has anyone tried this?

stagedew
I tried what you suggested and as a test I compared the resultant HEX Color Codes by sampling the 2 middle stamp in Sinclair2010's post above. They are the same stamp though I am unsure whether the same scanner was used. Oddly enough, the same stamp returned different Hex Codes as the RGB #'s varied. I then tried to pick up the color 3 times from the same stamp (the middle lower stamp) and it not surprisingly returned three different RGB numbers for each hue. I'm not sure where this gets us as apparently this Paint tool method would indicate each stamp has different hues all over the stamp and so can't be used as a comparative tool unless somehow ranges for the hues can be developed.

???????? You are sampling an image, not the stamp. There is no value in spending time with evaluating an image.

Consider how the image was generated. You have a scanner, firmware which runs the scanner including the driver, and then there is the software which generates the file. If the user saved the image as a JPG or any other compressed file format, you then have software changing the original image. And finally, if the user 'optimized' the image to upload you have another layer of software manipulation.

And if you could get past all the way the iage was generated, you then have to understand and deal with the way teh image is being presented on YOUR computer. Yikes. You are now many, many layers away from the actual stamp. This is why it is useless.

Don

there is also the question of excactly what was sampled - only the ink or the ink and paper in wha would undoubtedly be different proportions each time

Very interesting, bookmarking!

I have begun work on extracting the RGB from the stamp, or more exactly from a tiny most heavily inked portion of the stamp with some initial success

Because of this topic I am working with deep and dark blues, 63 and 63b including the "indigo" essay and what the specialists call near indigo.

Should we discuss this here or would a new topic on color be a better place?

Start a new topic. It should be very interesting.

Before spending time on trying to sample stamp colors the first thing that needs to be remembered as that you are not sampling the stamp color, you are sampling and analyzing an IMAGE. This then begs the question of what are you actually sampling? Sorry, but you have to get into a lot of geeky stuff to understand computer graphics. Making any color claims based upon computer graphics demands that we understand this techy geeky stuff or we make a large number of incorrect assumptions and statements.

So let's start with the most simple, basic computer color understanding. Computers use pixels. A pixel is a sample of an original image; more samples typically provide more accurate representations of the original. The intensity of each pixel is variable. With computers images, a color is typically represented by three or four component intensities such as red, green, and blue, (RGB) or cyan, magenta, yellow, and black (CMYK). Computers typically work with 16 million colors. To see what 16 million colors looks like you can download this large TIFF file and open it in your graphics application http://stampsmarter.com/webimages/RGB16Million.tif This TIFF file contains exactly one pixel of each of the 16,777,216 possible color values. The image is 4096 × 4096 = 16 million (16,777,216) pixels arranged as 256 slices of the RGB color cube. The uncompressed TIFF file is a 48 Mb.

When you save this file as a standard JPG (which is a compression algorithm) the file size drops to 29.28 Mb.

When you save this file to an optimized JPG (an even stronger compression algorithm) the file size drops even further to 3.84 Mb (shown above).

Yet if we were to display all three of this files on your screen, they would all look exactly the same. This is because the computer can 'see' far more colors then our human eyes can detect. If our eyes cannot detect the 16,777,216 possible color differences, there is little point in saving huge files on our computers. Online websites like this forum certainly do not want users to have to upload, and then consume hard drive space, huge files. This is why when you save a scanned image to a JPG the file size is reduced. This is why Bobby asks us to use the 'image optimizer' on this forum, it uses an even stronger JPG compression algorithm, making the file size even smaller than a standard JPG.

Ok, now we have a basic understanding, let's explore what the heck 'compression algorithm' means. An algorithm is simply a calculation set of rules to be followed by a computer. So it follows that a 'compression algorithm' is a set of rules which allows a computer to compress an image. The compression algorithm compresses the image using something called 'quantization'. (Is this geeky enough yet!?!)

Quantization is used in mathematics and digital signal processing and is defined as the process of mapping input values from a large set to a smaller set of output values. In other words; think rounding, averaging and truncation. So a compression algorithm reduces the image file size by using rounding, averaging and truncation on the pixels. Obviously the reduction of the file size means the image file loses data.

So what does this rounding, averaging and truncation of the image file mean if you try to sample a single spot of a stamp image? Here is a 5 x 5 pixel example of what your computer does at a pixel level when you save a uncompressed TIFF file to a compressed JPG file.

As you can see, the JPG compression algorithm as 'averaged' the pixel, reducing the number of colors. This data lose is how the image file size is made smaller. If you zoom back out and away from the pixel level, our eyes can barely detect any change in the overall image. This explains why so many people use JPG (or other) file formats.

When I scan a stamp image the computer has it in a uncompressed 16,777,216 colors on my monitor; but as soon as I save the file as a JPG (or other) file, I introduce all kinds of pixel averaging and rounding. If I run the forum image optimizer it modifies the image even more.

If I then try to detect the RGB value of either a single pixel or a group of pixels, what am I analyzing? I am analyzing some software engineers compression algorithm. Is this close to the original, actual stamp color? It is probably close but is it exact? No.

And all of the above is simply the 'input' side of computer image. If we were to then factor in the 'output' side (how the video controller, monitor, and your computer displays things) your head might explode. But it is safe to say that no two computers and monitors display the exact same image.

Knowing the geeky stuff above helpes us understand why analyzing an IMAGE color often has little to do with an actual stamp color.
Don

Great explanation Don! Very interesting AND eye-opening,

Peter

Don, I appreciate and need your advise on what I am attempting to do. I am not analyzing an image of a stamp but directly of the stamp itself. I am using a color meter app which I believe is crude, but if I measure two stamps in the same way under the same light I hope to compare the differences.

I also have begun comparing the RGB to a reading of a color chip in Ridgway.

Am I fooling myself? Tilting windmills?

So far if I enter the RGB result into a program that returns the color image, the match seems good.

I value your comments.