A Biography of the Pixel

Alvy Ray Smith, 2021

Beaverton Library 777.7 SMI

The main message of the book is that "pixel" means an abstract, point-centered, weighted optical sample of an image; NOT a square swatch of color on a screen or a page, nor three binary numbers for red, green, and blue. A screen or print element grid can be computed from a pixel array, but the quality of the resulting image will depend on the "spreader" function that distributes the pixel across those grid elements.

Another message is that the concept of bandwidth limited sampling and construction (data points at least twice the distance or space frequency of the waveform) was developed by Vladimir A. Kotelnikov and proved (in Russian) in 1933, long before Claude Shannon's 1949 paper. Kotelnikov visited the US for 60 days in 1936, perhaps for a diplomatic cryptography project. Shannon did not claim credit for the Sampling Theorem, though it is frequently attributed to him in the west. It would not be wise to call it "The Shannon Sampling Theorem" within earshot of the author, or some readers of this book. For myself, the takehome message is to omit "Shannon" when saying "Sampling Theorem", but add it back in if my English-speaking audience doesn't know what I'm talking about --- which would include jingoistic US citizens.

Attribution and priority are frequent themes. Who made the first computer, the first graphic display, the first photograph, the first movie, the first digitized image, the first computer synthesized image, the first antaliased line? Or, at the focus of the book, who founded Pixar? Who managed Pixar? Who chose Pixar's mission, strategy, goals? It wasn't George Lucas. It wasn't Steve "moneybags" Jobs.

In a universe of trillions of galaxies with trillions of stars in each, the first instance and implementation of all these ideas happened long ago in a galaxy far far away ... but that information hasn't reached us yet. I'll rely on the best explanation with the most technical details. The explanations in Pixel are sorta OK, but devoid of algebra, so they aren't the best explanation for me.

Page 309, figure 7.1 is a computer rendered image of what looks like a (red) Apollo command and (black) service module (CSM), but is labeled Apollo Lunar Module and contributed by Jim Blinn. It is MISIDENTIFIED as a lunar module. The picture is (allegedly) from a GE brochure, claiming a 1962 date for the computer that produced it. In the Annotations (a weird PDF asking for passwords with some Linux pdf viewers, opens with firefox only), we are told this was made with the first color NASA-2 Electronic Scene Generator system in 1967. However ... brochures aren't screens, and a brochure artist can colorize black and white pictures. The earliest lunar mission proposals involved a single HUGE spacecraft, which landed on the moon with the command and service modules and the landing stage as one unit, a monster with the command module on the top and a wedding cake of tanks below ... WITH LANDING LEGS at the bottom, not what is shown in the picture. NASA soon realized that landing the return vehicle with all its Earth-Return consumables and a heavy heat shield was dangerous and expensive - better to send an orbiter and a lander. So ... this picture is a compounded mistake. This implies there may be many other mistakes in "Pixel" beyond my expertise.

The book mentions an interview with Ivan Sutherland in Portland, a very interesting takeaway from the book. Turns out, Ivan Sutherland moved to Portland, and researches asynchronous logic at Portland State University (Annotations p 217: "he told me, on Jan. 23, 2018, that he now knows how to handle it" ... Hmm ). Before COVID, I attended seminars at PSU, and he looks familiar.

All and all, about 20% fascinating, and 80% plodding detail, beating Sturgeon's Law by 100%. The book is in demand at the library, so back it goes, with no more notes.

Pixel (last edited 2022-01-14 06:56:42 by KeithLofstrom)