Before graphics capabilities of personal computers became
commonplace, printing pictures on the computer center's lineprinter
has been very popular. This was done by replacing pixels in an
original picture by characters, usually overprinted by other
characters in order to obtain roughly the correct gray value.
Overprinting was done line by line (by issuing only a carriage
return at the end of a line instead of carriage return /
linefeed).
An example of a dataset for lineprinter graphics is the text file tiger.txt [2] which I found on a reel
of paper tape. Viewing this file on a computer screen by means of a text
editor doesn't produce a convincing result. Therefore I wrote a
TurboPascal program tiger4.pas which
translates the character data back to gray-valued pixels.
I used the 5*7 dots character set of the Aritma 130 mosaic
printer [3] for calculating the gray values corresponding to the
various character combinations appearing in the file. The relevant
mosaic patterns are in the file patterns.
With a few adaptations the program can be used for reconstructing
pictures from any text file of this kind.
The reconstructed picture, in 16-bit 'raw' format, has only 158
*132 non-square (i.e. 5*7) pixels: 2597 lines, many of them
overprinted resulting in 158 separate lines.
The version shown here was -after rotation, inversion and cropping-
enlarged about three times (with bilinear interpolation), taking
into account that square pixels are used in the final picture.
[1] Jaarverslag Stichting Mathematisch Centrum, Amsterdam 1971.
[2] The files tiger.txt, tiger4.pas and patterns are in this zip archive.
[3] This character set is described in A. Ralston (ed), Encyclopedia of Computer Science, Van Nostrand Reinhold 1976.