This is live entropy from the Photonic Instrument. The three lines represent the red, green and blue pixel values (0-255) of a continuous strip across the middle of the webcam’s sensor. There are 500 data points on this graph, containing a true yet deceptively low entropy of only ~11 bits, proportioned as $ \frac{500}{640 \times 480 \times 3} \times 21,000 $. The JPEG algorithm acts like a pseudo random number generator, $ \operatorname{JPEG}: \{0,1\}^{\approx 21,000 \times 8} \to \{0,1\}^{640 \times 480 \times 3 \times 8} $ on the source image file(s). The chart though only encompasses ~0.05% of the total entropy available from a single Photonic Instrument frame. This is based on a very conservative entropy assessment of 1 bit/jpg byte along files approximately 21kB in size.

Note 1: The chart is simply the output of the JPEG algorithm, and is therefore only representative of the entropy generated by the Photonic Instrument’s webcam sensor. It’s really a product of discrete cosine transforms, quantization and Huffman decoding. You can think of these lines as pseudo-random entropy. In terms of algorithmic complexity, the true entropy is the source JPG file itself. The decoding process produces a 32 fold expansion from the source file to the image.

Note 2: We endeavour to maintain this entropy source on-line at all times, but that’s not always possible. Sometimes maintenance is required, and it goes off-line when the instrument needs greasing. Most Bank holidays are out too. There will be a big red error box and an “An Error Has Occured…” message during the off times. Apologies.