xenocara/doc/xorg-docs/specs/Xext/lbxalg.xml

151 lines
4.5 KiB
XML
Raw Normal View History

<?xml version="1.0" encoding="UTF-8" ?>
2012-06-09 03:07:41 -06:00
<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.3//EN"
"http://www.oasis-open.org/docbook/xml/4.3/docbookx.dtd"
[
<!ENTITY % defs SYSTEM "defs.ent"> %defs;
]>
<article id='lbxalg'>
2012-06-09 03:07:41 -06:00
<articleinfo>
<title>LBX X Consortium Algorithms</title>
<releaseinfo>X Version 11, Release &fullrelvers;</releaseinfo>
</articleinfo>
2012-06-09 03:07:41 -06:00
<sect1 id='Introduction'>
<title>Introduction</title>
<para>
The Low Bandwidth X extension allows for negotiating various algorithms used
by LBX. This document describes the algorithms used in the Consortium
implementation of LBX in the X11 Release 6.4.
</para>
</sect1>
2012-06-09 03:07:41 -06:00
<sect1 id='Streaming_Compression'>
<title>Streaming Compression</title>
<para>
LBX negotiates the use of a stream compressor. The consortium implementation defines a stream
compressor named XC-ZLIB, which is based on the Zlib version 1.0 compression library by
Gailly &amp; Adler.
</para>
<para>
The XC-ZLIB compressor is presented with a simple byte stream - the X and LBX message
boundaries are not apparent. The data is broken up into fixed sized blocks. Each block is compressed
using zlib, then a two byte header is prepended, and then the entire packet is transmitted.
The header has the following information:
</para>
<para><programlisting>
out[0] (length &amp; 0xfff) >> 8 | ((compflag) ? 0x80 : 0);
out[1] = length &amp; 0xff;
</programlisting></para>
<para>
If the compflag is false, then the contents of the block are not compressed.
</para>
</sect1>
2012-06-09 03:07:41 -06:00
<sect1 id='Bitmap_Compression'>
<title>Bitmap Compression</title>
<para>
LBX also negotiates for bitmap compression. The consortium
implementation defines a bitmap compressor named XC-FaxG42D,
which uses the CCITT Group 4 2D compression algorithm.
</para>
</sect1>
2012-06-09 03:07:41 -06:00
<sect1 id='Pixmap_Compression'>
<title>Pixmap Compression</title>
<para>
LBX allows for the negotiation of pixmap compression. The
consortium implementation does not define a pixmap compression
algorithm. (A run-length encoding algorithm was proposed, but
experimentation proved it was less efficient than allowing the
stream compressor to compress the image.
</para>
</sect1>
2012-06-09 03:07:41 -06:00
<sect1 id='Colormap_Algorithm'>
<title>Colormap Algorithm</title>
<para>
LBX negotiates for use of a colormap algorithm, used for color
matching when the proxy allocates pixels in a grabbed colormap.
The consortium implementation defines an algorithm named
XC-CMAP. This algorithm consists of three parts, resolving to a
hardware color, finding the closest existing color, and what free
cell to allocate.
</para>
<para>
The XC-CMAP algorithm resolves a color to a hardware color in the
following manner:
</para>
<para><programlisting>
#define RESCALE(x, nbits) (x &gt;&gt; (16 - nbits)) * 65535 / ((1 &lt;&lt; nbits) - 1)
#define GRAY(r, g, b) (30L * r + 59L * g + 11L * b) / 100
sigbits = pVisual-&gt;bitsPerRGB;
switch (pVisual->class) {
case PseudoColor:
case DirectColor:
case StaticColor:
/* rescale to rgb bits */
*red = RESCALE(*red, sigbits);
*green = RESCALE(*green, sigbits);
*blue = RESCALE(*blue, sigbits);
break;
case GrayScale:
/* rescale to gray then rgb bits */
*blue = *green = *red = RESCALE(GRAY(*red, *green, *blue), sigbits);
break;
case StaticGray:
/* rescale to gray then [0..limg] then [0..65535] then rgb bits */
*blue = *green = *red = RESCALE(RESCALE(GRAY(*red, *green, *blue),
pVisual>numPixelBits), sigbits);
break;
case TrueColor:
/* rescale to [0..limN] then [0..65535] then rgb bits */
*red = RESCALE(RESCALE(*red, pVisual->numRedBits), sigbits);
*green = RESCALE(RESCALE(*green, pVisual->numGreenBits), sigbits);
*blue = RESCALE(RESCALE(*blue, pVisual->numBlueBits), sigbits);
break;
}
</programlisting></para>
<para>
The XC-CMAP algorithm matches a color to an existing pixel in
static visuals by finding the pixel with the lowest color match
error, computed as follows:
</para>
<para><programlisting>
error = errRed * errRed + errGreen * errGreen + errBlue * errBlue
</programlisting></para>
<para>
The XC-CMAP algorithm selects a free pixel to allocate by selecting
the free pixel with the lowest index from the free pixels known to
the proxy. For direct visuals, it uses the lowest free or matching
pixel subfield known to the proxy for each color.
</para>
</sect1>
<sect1 id='Extensions'>
<title>Extensions</title>
<para>
LBX allows for extensions to LBX to enable additional compression
or short-circuiting. The consortium implementation does not define
any extensions to LBX.
</para>
</sect1>
</article>