Browse Prior Art Database

Integrated Scheme for Dithering 24 Bits of True Color Down to 4/8 12 or 16 Bits with Reduced Error

IP.com Disclosure Number: IPCOM000112635D
Original Publication Date: 1994-Jun-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 4 page(s) / 110K

Publishing Venue

IBM

Related People

Lawless, WF: AUTHOR [+3]

Abstract

In graphics systems, often 24-bit true color RGB will be reduced down to 4, 8, 12 or 16 bits and then stored in the Video Random Acdess Memory (VRAM) while maintaining reasomable visual quality of the image. This reduces VRAM memory costs. The common technique used to accomplish this is called dithering. Most graphics systems products use a 4x4 matrix as a result of which extra bits of information are unnecessarily lost. This invention disclosure presents a new dither matrix that uses the full information range of the input Red, Green, Blue (RGB) values for incrementing most significant color values and an implementation scheme based on this matrix that permits 8/12/16/24-bit dither in an integrated manner.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Integrated Scheme for Dithering 24 Bits of True Color Down to 4/8
12 or 16 Bits with Reduced Error

      In graphics systems, often 24-bit true color RGB will be
reduced down to 4, 8, 12 or 16 bits and then stored in the Video
Random Acdess Memory (VRAM) while maintaining reasomable visual
quality of the image.  This reduces VRAM memory costs.  The common
technique used to accomplish this is called dithering.  Most graphics
systems products use a 4x4 matrix as a result of which extra bits of
information are unnecessarily lost.  This invention disclosure
presents a new dither matrix that uses the full information range of
the input Red, Green, Blue (RGB) values for incrementing most
significant color values and an implementation scheme based on this
matrix that permits 8/12/16/24-bit dither in an integrated manner.

      Background - The goal of this dithering is to increase the
apparent color resolution of the display by trading against its
spatial resolution.  24-bit RGB color (32-bit ARGB) is reduced to 8
bits (RGB=332), 12 bits (RGB=444) or 16 bits (RGB=565), nevertheless
yielding a realistic, photolike image.  The common algorithm is shown
in Fig. 1.  Only the red field is depicted, but the green and blue
fields should follow the same strain.

      Normally, a 4x4 dither matrix is used.  Therefore, the elements
of the dither matrix have a 4-bit range.  Bits <4:1> of the input red
color are compared with the 4-bit value in the dither matrix indexed
by 4 bits comprised by the 2 least significant bits of X and 2 least
significant bits of Y address of the pixel (relative to the window).
If the value indexed in the dither matrix is less, then the most
significant 3 bits of the input red color value is incremented (only
if these 3 bits are less than B'111'); otherwise, they are left
unchanged.  The resulting 3 bits are then the "dithered output."  One
bit of information, i.e., the LSB of the input color value is never
used.  For the green field, 2 bits of information will be lost.

      New Dither Matrix and Integrated Dithering Scheme - We present
an 8x8 dither matrix (shown below) indexed by 6 bits comprised of the
3 least significant bits of the X and Y addresses.  This dither
matrix fully utilizes all the information.

            \ 3LSB of Y (000,001.....111)
    3LSB of X
      000        0   2  8  40  3  35  11  43
      001        48 16 56  24 51  19  59  27
      010        12 44  4  36 15  47   7  39
      011        60 28 52...