Kinematograph is a generative computer installation about the relationship between encoding and generation.

Its current version consists of two elements. The first is a print displaying a frequency matrix, which has been obtained by analyzing a short video sequence recorded in a crowded street of Hong Kong’s Mong Kok district, one of the most densely populated areas in the world. Each entry (i, j) of the matrix represents the number of times during the course of the video that the brightness of any pixel has changed from i to j.

The second component is a computer that generates pixels in real time and displays them on a monitor. The system has previously encoded the matrix data in the form of a Huffman tree and uses this tree to generate pixels with the same probabilities as those of the original street footage.

Kinematograph explores the artistic possibilities of entropy encoding algorithms, which employ the frequencies of a set of data to achieve the most efficient possible representation of that data.

More specifically, the work focuses on the Huffman encoding algorithm. This technique was originally designed in the 1950s as an optimal method of lossless data compression. In addition, the Huffman algorithm can be also be used as a generator of random numbers with a non-uniform distribution. The concept of the work highlights the intimate connection between data compression and random data generation.

A more detailed description of the algorithm can be found here.

Presented by the Writing Machine Collective and the Hong Kong Museum of Art