PNG 16:  Technical
Version 0.50, Rich Franzen, August 1998
NOTE:  this page and the PNG 16 Overview have cross-linked sections.

Visit the png-16 Forum at Deja News

  • Introduction
  • The png-16 image format stores imagery in 16-bit pixels.  It is being developed to offer several unique benefits: Four formats are proposed:  fc16, mc16, gc16, and ic16.  The bit patterns for each are: In each case, the high-order bit is on the left.  Just as in PNG, network byte order is used to store the pixels, regardless of the native byte-order of the computer which generates the image.

    The file extension "pn6" is to be used by png-16 images.  The pronunciation is "ping sixteen".  A compliant png-16 program must handle both normal PNG images, as well as pn6 images. Once png-16 has proved itself in the real world, my hope is that it become a formal part of the PNG specification.  Until then, it is simply a PNG-like image format.

  • fc16  (full color, 16 bits)
  • The primary format of png-16 is fc16.  In one sense it is a partial indexed representation of 24-bit color space.  This is a special indexing, though, wherein most of the index values having an inherent and obvious relationship to the colors they represent.  In another sense, fc16 is a superset, containing 12-bit grey in addition to true color.  Although the format is basically consistent with itself, some of the features require a degree of unorthogonality.  For example, absolute consistency would require that black exist 1024 times.  Obviously one true black is enough.

  • mc16  (many color, 16 bits)
  • gc16  (grey with color, 16 bits)

  • ic16  (indexed color, 16 bits)

  • 8-bit Indexing
  • Lots of LUTs
  • The Canvass


    this page still


    Rich Franzen 
    Color Space Overview 
    PNG 16 Overview 
    PNG 16 Technical 
    First published August 1998,   counter   hits since 2004-04-05,   last modified 2010-11-21