Topic: Secrets of the graphic standards GX1
Author: Fenev SV
---------------------------------------
This article can serve as Dep
vnoy point for programmers who wish to use in their work
graphic standard GX1.
Graphic images appearing on the screen, cause sometimes
people carrying leading admiration for programmers professional
interest. Which graphics Shell, that the format, what is the
length of file that stores the image -
These and many other questions arise in
such situations, people who have represented
resistivity of the graphic capabilities inherent in the
computer.
Specificity of the Soviet program, "Bazaar" - the spread of
extremely vulnerable to copying software products -
substantially restricted access to the world market of software
Among wandering on our computers
versions of most graphic editors
popular, perhaps, Paint
Brush Company ZSOFT. Used it
Packaging image format GX1 supported in many other software
Drugs, encountered in some games
Remember how often you met Faye
ly with the extension. gx1.
Beyond doubt the feasibility packaging
graphic image, because for
store all the information from the screen to display adapter
EGA mode 640 * 350 items, 16 colors, requires 128
KByte.
"Picture" with the normal intensity
images compressed in a format GX1,
placed in a file length of approximately 15-25
KByte. Without giving comparative analysis format used (this
Material can be found in one of the closest
issue of electronic journal FAULTLESS)
note the compactness of this standard:
graphic design, packaged in GX1,
It takes approximately one and a half times
less memory capacity than the same figure, packaged in a format
PCX.
Format structure GX1.
File format GX1 has the following structure
structure:
- Header file, a length of 3 bytes
- Information about the used panel
(For EGA)
- Data describing the image.
So, the file header:
- 1 byte - FA - graphic, FB - tex
The quantum regime
- 2 bytes - the screen mode in a 16-chnom form
- 3 bytes - the number of bytes in the color palette and
Background:
0 - CGA 40 * 25 monohr.tekst.
1 - CGA 40 * 25 color. text.
2 - CGA 80 * 25 monohr.tekst.
(4 senior level)
3 - CGA 80 * 25 color. text.
0 - Dark
green-red-brown
4 - CGA 320 * 200 4 tsvet.graf.
1 - Light
green-red-yellow
...
2 - Dark
blue-pink-white
8 - HGC 720 * 348 monohr.graf.
3 - Light
blue-pink-white
9 - EGA 640 * 350 monohr.graf.
4 - dark
blue-red-white
A - EGA 320 * 200 16 tsvet.graf.
5 - Bright
blue-red-white
B - EGA 640 * 200 16 tsvet.graf.
6 - monochrome graph.
CGA mode
C - EGA 640 * 350 4 tsvet.graf.
640 * 200
D - EGA 640 * 350 16 tsvet.graf.
...
(4 LSB)
0 - black 8 - Gray
1 - blue 9 - bright blue
2 - Green A - Light Green
3 - Blue B - Light Blue
4 - Red C - light red
5 - Pink D - light pink
6 - Brown E - yellow
7 - White F - bright white
The presence of the third byte must
even under conditions in which he has no
sense. Just above 3 bytes are not
are described below package.
If you try to sort out the string
kture files with the extension. gx1, prosmat
By considering their content, it is likely
These attempts will fail. But the principle
ciples used in image compression, are rather trivial.
Compression algorithm.
When you create a file in a format GX1 are two of the
algorithm. At the base of the first lies an indisputable fact:
the horizontal line is often very slightly different from the
next row image located above and below. Originally dedicated to
the memory buffer is placed first upper row of images. After
that each subsequent line compared with the previous byte, for
each line is formed by the so-called
map differences. So for graphics display adapter CGA mode 320 *
200 elements every mapping string occupies 80 bytes. Packing
CA is as follows: map the differences formed a length of 10
bytes each bit map corresponds to a byte string iso
mapping. If this bit - 1, hence
but the bytes in the current row is different from
byte of the previous line. New byte value is placed after the
map. Number of bytes equal to the number of bits with value 1.
This algorithm is called vertical compression.
Once the package is executed every
mapping, in the buffer data
compressed, is used in this version
familiar to many of the group method of coding. These are
divided into groups, each of which begins with the total number
meaning the number of repetitions tions of a byte. If the byte
value occurs in a series of more than three times, then there
is a "byte repetition." This Byte MSB unlike all the other
installed in the unit, and the remaining bits indicate the
number of repetitions bytes whose value can be found in the
following byte. To make it clearer, we illustrate
this with specific examples. So, the next group (16-dimensional
representation) 84 00 indicates the presence of four well
lei:
2-ary representation:
1000 0100 0000 0000
> Bit-sign
And FF 00 - respectively 127 zeros.
By the way, a number of repeated
values can be encoded in two bytes using the method described
above group coding.
Together, these methods
provides a very compact
representation of graphic images.
In this case, there is a seal of information because of its
specificity as a vertical and horizontally. When packing the
screen in text mode only works the second method. At the same
value of the first byte header, and hence, total
file is equal to FB.
Subject: The compression algorithm LZSS
Author: Unknown
---------------------------------------
METHOD LZSS
This scheme uses the algorithm of Lempel
(Lempel) and Ziv (Ziv). Slightly modified version described
Storer (Storer) and Szymanski (Szymanski). Development, use
forming a binary tree to accelerate the search suggested that
Bell (Bell).
We introduce the following definitions used
the description of the method.
1. Input stream - is any file (data set), which should be
compressed.
2. Output - is a file that
recorded the results of the method.
3. The circular buffer is an array of some length, whose "end
is connected to beginning ", that is, if working with it a
pointer to jump beyond the end of this buffer, then it is
automatically set to the top Similarly, if a pointer to cross
the border beginning of the buffer, it will be installed in the
end (the last item in the buffer).
4. The size of the ring buffer is usually 4,8,16, or 32 KB of
data.
5. Maximum length sequences - a long sequence of characters
that uniquely encode the packer. If the length of the sequence
exceeds this value, then the packer will divide it into pieces
no longer than the maximum.
6. Current index is a marker in the ring buffer, indicating
the beginning of another uncompressed text.
7. Forward-pointer - a marker of
circular buffer indicating the byte
following the last symbol is read from the input stream.
Forward-decree Tel differs from the current pointer
maximum length sequences. In
this method, the maximum length sequence is 264, so
forwardukazatel will be greater than the current 264.
8. A pointer to the beginning of the coincidence - it
marker in the ring buffer, which decrees
cates the beginning of a byte sequence is the sequence
bytes, the first symbol is set to the current index.
Ring buffer initially filled with blanks. From the input
stream in a circular buffer read sequence of bytes, the number
which is equal to difference between forward-pointer from the
current pointer in this case 264 bytes.
Current index is set to
beginning of the buffer and forward-pointer - in
the end of the read sequence and exact
it to the byte immediately following the last
symbol of the read data.
We then proceed according to the algorithm:
1. We are looking back on the length of the buffer (we assume
that the buffer length is equal to 4 KB) naibolschuyu sequence
of characters coinciding with the current, ie with the
sequence, which is the beginning of the current schy pointer.
2. Analyze the following situations:
a) The length of the repeating sequence is equal to zero or
one. The output stream to record: 1 bit (equal to unity -
a sign that more is nezapakovanny bytes), 8 bits (the current
byte). Current Index increased by one.
If you do not reach the end of the file, then read
Th symbol from the input stream and write
at forward-ukazatelya.Forward-pointer increment. Otherwise from
input stream, nothing to read and
forward-pointer does not increase.
If the current pointer is forwardukazatelyu, the process is
complete, otherwise go to step 1.
b) The length of the repeating sequences of more than one.
In the output stream to write: 1 bit (equal to zero - a sign
that is packed on the sequence), 12 bits (offset back from the
current pointer to an identical sequence sequence, the maximum
displacement is 2 ^ 12 = 4096), the length of the repeating
sequence:
1) in 3 bits, if the length from 2 to 8
(Coded 000 .. 110);
2) 3 8 bits if the length from 9 to 264.
The first 3 bits contain a combination of 9 (111)
which means: the true length is
in the next 8 bits (length 9 .. 264, coded 00000000 ..
11111111).
Current Index increased by
the length of repetitive sequences.
If you do not reach the end of the file, then consider
from the input stream for a number of
characters, equal to the length of repetitive sequences at
forward-pointer. Forward-pointer to increase by the same length
as the current index.
If end of file met earlier than
been read the required number of bytes, the
forward-pointer to read by increasing the number of bytes.
Otherwise, the input stream does not read, forward-pointer
does not change.
If the current pointer is forwardukazatelyu, then complete
the process, otherwise go to p.1.Esli current index is
forward-pointer, then complete the process, otherwise go to
step 1.
It is worth noting a few nuances
in search: 1. If you implement the usual
search back at 4 Kbytes (for
this case) under which originally sought to match the first
symbol, and If matched, then the second, etc., then this
means that the packer must for every byte of text to make at
least 4097 comparisons: 4096 comparisons back and one
Comparison vpered.Netrudno guess what
package will be extremely slow
slowly. 2. We note an important point: as
uses a circular buffer, then if
looking back index jumped search
through the buffer, it automatically
Move to the end of the buffer. 3.Pri
compared to the coincidence index of coincidence can jump
through the current pointer, as long as the top match was
earlier this ukazatelya.Privedem example:
sequence length
with and on for a while about a
current pointer
with and on for a while about a
pointer to start matching
and the text looks like this:
and b but with a W on to with a shock of ASP and Art and
Design Bureau
We give two examples of this
method:
Example 1:
Source text:
m and to m and m and m s for a while TT TT TT TT and
Compressed text:
1, "m", 1, "a", 1 "to", 0, 4,
1b. 8b. 1b. 8b. 1b. 8b. 1b. 12b.
0, 0, 2, 0,
36. 16. 126. 36.
1, "", 1, "m", 1, "s", 1 "to"
1b. 8b. 1b. 8b. 1b. 8b. 1b. 8b.
1, "B", 0, 6, 1, 0, 0,
16. 86. 16. 126. 36. 16. 126.
6, 1, "and"
3b. 1b. 8b.
Percentage compression = 160 bit/200 bit = 80.0%
Example 2:
Source text:
and b p and b to a to b and b to b to abrabkabkabk
Compressed text:
1, "a", 1, "b", 1, "p",
1b. 8b. 1b. 8b. 1b. 8b.
0, 3, 0, 1, "k", 0, 3, 4,
16. 126. 36. 16. 86. 16. 12. 36.
0, 5, 2, 0, 14, 7, 2
1b. 12b. 3b. 1b. 12b. 3b. 8b.
Percentage compression = 108 bit/216 bit = 50%
---------------------------------------
Sample from FRACTINT.DOC
Fractint Version 15.0 Page 27
Barnsley IFS Fractals (type = ifs,
ifs3d) One of the most remarkable
spin-offs of fractal geometry is the
ability to "encode" realistic images in
very small sets of numbers-parameters
for a set of functions that map a region
of two-dimensional space onto itself. In
principle (and increasingly in
practice), a scene of any level of
complexity and detail can be stored as a
handful of numbers, achieving amazing
"Compression" ratios ... how about a
super-VGA image of a forest, more than
300,000 pixels at eight bits apiece,
from a 1-KB "seed" file? Again, Michael
Barnsley and his co-workers at the
Georgia Institute of Technology are to
be thanked for pushing the development
of these iterated function systems
(IFS).
-Sean Burke
Los Angeles, December 1991
