mirror/ioccc
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #254 from xexyl/1994-tvr

Browse Source 
Fixes in 1994/tvr/README.md
This commit is contained in:
Landon Curt Noll authored and GitHub committed 2023-04-02 16:26:03 -07:00
2 parents cde73d96fd d819988f1d
commit ee5ab83ba5
1 file changed
+155 -159
Download Patch File Download Diff File Expand all files Collapse all files
1994/tvr/README.md
+155 -159
View File
@@ -1,217 +1,213 @@
# Best X11 Program:
Teemu Rantanen
Helsinki University of Technology
Melkonkatu 5 A 8
00210 Helsinki
Finland
Teemu Rantanen
Helsinki University of Technology
Melkonkatu 5 A 8
00210 Helsinki
Finland
## To build:
make all
## To run
## To run:
./tvr mode screensize/2 < colormapfile
```sh
./tvr mode screensize/2 < colormapfile
```
## Judges' comments
## Try:
Try:
./tvr 0 128 < tvr.color (for color displays)
./tvr 0 128 < tvr.bw (for Black & White displays)
```sh
./tvr 0 128 < tvr.color (for color displays)
./tvr 0 128 < tvr.bw (for Black & White displays)
```
Mode may be a value from 0 to 12.
Mode may be a value from 0 to 12.
This entry requires the use of X11. It will form two windows
(hopefully your window mangler will ask you to place these
your display) which you should place side by side. Move the
cursor around in one of the two windows and watch what happens.
## Alternative code:
The fractally minded may be able to detect that mode 0 does not
calculate Mandelbrot/Julian sets correctly. Can you find the bug?
Better still, can you fix it without breaking something else?
The author has provided us a spoiler version:
The author has provided us a spoiler version:
make tvr.alt
./tvr.alt altmode screensize/2 < colormapfile
where altmode 1-4 correspond to mode 0-3 in the original entry
and altmode 0 calculates Mandelbrot/Julian sets correctly.
## Author's comments
Interactive Fractals for X Window System
where `altmode` `1` - `4` correspond to mode `0` - `3` in the original entry and
`altmode 0` calculates Mandelbrot/Julian sets correctly.
This program contains 4 pairs of fractals/effects and 3 methods of
animating them. There are two windows for a pair of fractals, one
of which contains static fractal (eg. Mandelbrot set) and the
other an animating one (eg. Julian set) which is derived from the
same formula as the static one. Only one pair of fractals can be
displayed on one process.
## Judges' comments:
This entry requires the use of X11. It will form two windows (hopefully your
window manager will ask you where to place these in your display) which you
should place side by side. Move the cursor around in one of the two windows and
watch what happens.
The fractally minded may be able to detect that mode 0 does not calculate
Mandelbrot/Julian sets correctly. Can you find the bug? Better still, can you
fix it without breaking something else?
## Author's comments:
### Interactive Fractals for X Window System
To animate the fractal, you must move your mouse on the static
fractal -window and the animated fractal will be calculated based
on mouse coordinates. If you are on one of the two non-stable
modes (more about them later) you can stabilize the picture by
pressing a mouse button. Pressing it does nothing on stable mode.
This program contains four pairs of fractals/effects and three methods of
animating them. There are two windows for a pair of fractals, one of which
contains static fractal (e.g. Mandelbrot set) and the other an animating one
(e.g. Julian set) which is derived from the same formula as the static one. Only
one pair of fractals can be displayed in one process.
To animate the fractal, you must move your mouse on the static
fractal window and the animated fractal will be calculated based
on mouse coordinates. If you are on one of the two non-stable
modes (more about them later) you can stabilize the picture by
pressing a mouse button. Pressing it does nothing on stable mode.
The three modes are:
- non-stable mode with simple 'attractor searcher'
- non-stable mode
- stable mode
On Mandelbrot/Julian mode the stable mode calculates normal Julian set according
to the Mandelbrot coordinates (mouse position). On non-stable modes the picture
slowly transforms to the same picture as on the stable mode but transformation
can take some time. Usually, and especially if changes are relatively small
between mouse movements or frame rate is high, the non-stable mode makes much
more interesting effects IMHO. The difference between the two non-stable modes
is that the first tries to stabilize the picture. It doesn't stabilize much, but
it helps in some cases quite a lot. One notable case where help can be seen is
when you first move your mouse to non-black areas of the Mandelbrot set, then
back to near the center. In some cases 'attractor searcher' can be seen as
noise.
The first of the fractal pairs is Mandelbrot/Julian set as mentioned before. The
second is a variation of that. The third makes nice 3d-look-a-like effects and
the fourth is a variation of that with more nice effects.
The three modes are:
- non-stable mode with simple 'attractor searcher'
- non-stable mode
- stable mode
### Usage:
On Mandelbrot/Julian mode the stable mode calculates normal Julian
set according to the Mandelbrot coordinates (mouse position). On
non-stable modes the picture slowly transforms to the same picture
as on the stable mode but transformation can take some
time. Usually, and especially if changes are relatively small
between mouse movements or frame rate is high, the non-stable mode
makes much more interesting effects IMHO. The difference between
2 non-stable modes is that the first tries to stabilize
picture. It doesn't stabilize much, but it helps on some cases
quite a lot. One notable case where help can be seen is when you
first move your mouse to non-black areas of Mandelbrot set, then
back to near the center. On some cases 'attractor searcher' can be
seen as noise.
```sh
./tvr mode screensize/2 < colormapfile
```
First of the fractal pairs is Mandelbrot/Julianset as mentioned
before. Second is a variation of that. Third makes nice
3d-look-a-like effects and fourth is variation of that with more
nice effects.
The first parameter is the operating-mode of the program. There are 4 fractal
pairs, 3 display modes, making total of 12 modes. `Screensize` is half of the
actual `screenlength/width`. I.e. if you use 128 as a second parameter, you will
get two 256x256 windows.
The program needs reasonable values for both of these parameters to
operate. A good first try could be "0 128" or "8 128". `Colormap` has
to be piped into the programs standard input.
Note that you will need approximately `(second_parameter^2)*64` bytes of free
main memory or the machine will swap heavily (memory references are quite random
and may confuse paging algorithms easily). This means that for 512x512 windows
you will need 4M of memory.
### `Colormapfile`:
Usage:
tvr mode screensize/2 < colormapfile
This is the file from which program reads colors for screen `colormap`. The
first line of the file must be a number between `3..254`. This informs the
program how many colors there are to be expected. This is also the maximum
number of iterations used on picture calculations. Lines after that are X11
color names. The program expects to find, and it uses, the number of colors
reported on the first line.
The first parameter is the operating-mode of the program. There
are 4 fractal pairs, 3 display modes, making total of 12
modes. Screensize is half of the actual screenlength/width. ie. If
you use 128 as a second parameter, you will get two 256x256
windows.
Program needs reasonable values for both of these parameters to
operate. Good first try could be "0 128" or "8 128". Colormap has
to be piped into the programs standard input.
Note that you will need approximately (secondparameter^2)*64 bytes
of free main memory or machine will swap heavily (memory
references are quite random and may confuse paging algorithms
easily). This means that for 512x512 windows you will need 4M of
memory.
There are two `colormapfiles` to start with. One with colors and another with
only black and white. If you have enough colors on your display you can use the
one with colors (even on `grayscale`). The other is meant mainly for a one bit
plane display (on which the picture looks terrible compared to a 8 bit plane
display).
Colormapfile:
### Compilation:
This is the file from which program reads colors for screen
colormap. First line of the file must be a number between
3..254. This informs the program how many colors there are to be
expected. This is also the maximum number of iterations used on
picture calculations. Lines after that are X11 color
names. Program expects to find, and uses, the number of colors
reported on first line.
You should compile this entry with an optimizing compiler with the
highest level of optimization possible. You will also need an ANSI
compatible compiler (`gcc -ansi -pedantic` will do fine).
There are 2 colormapfiles to start with. One with colors and
another with only black and white. If you have enough colors on
your display you can use the one with colors (even on
grayscale). The other is meant mainly for 1 bitplane display
(on which the picture looks terrible compared to 8 bitplane
display).
This entry has been tested on wide variety of machines including:
```
Machine OS
------- --
Decstation 5000/240 Ultrix 4.3A
DECAlpha 3000/300 OSF/1 v1.3 and v2.0
IBM RS6000 AIX v3
SPARCStation IPX SunOS 4.1.3
i486 Linux v1.1.8 (binarysize 3576 bytes
using `gcc -O6 -s -N` ;)
HP 9000/730 HP-UX
```
Also, many of the X server and X library versions have been tested and the
program should be highly portable. The program may slow down considerably on
machines where table references are slow as the program makes extensive use of
them. Also, due to the huge amount of data sent to the X server, considerable
slowdown can be noticed if the X server is slow or especially if the connection
between the X server and the client is slow. (Don't try this on 9600 bps modem
line. I did ;) )
Compilation:
You should compile this entry with optimizing compiler with the
highest level of optimization possible. You will also need an ansi
compatible compiler (gcc -ansi -pedantic will do fine).
Obfuscation is done on different levels. Almost none of it is made at the macro
level, so `cpp`ing the code won't give you much. Perhaps most notable is the
algorithmic obfuscation. Simple formula has been turned into something
completely different. Also it may be a surprise to find out which piece of the
code calculates non-stable fractals and how it does that. Many obfuscations were
left out of the program because speed of animation was one of the key issues.
Well, obfuscated or not, some of the effects are very nice ;)
### Memory refreshment:
This entry has been tested on wide variety of machines including:
Mandelbrot set is calculated from the formula
Machine OS
------- --
Decstation 5000/240 Ultrix 4.3A
DECAlpha 3000/300 OSF/1 v1.3 and v2.0
IBM RS6000 AIX v3
SPARCStation IPX SunOS 4.1.3
i486 Linux v1.1.8 (binarysize 3576 bytes
using gcc -O6 -s -N ;)
HP 9000/730 HP-UX
Also, many of X-server and Xlibrary version have been tested and
the program should be highly portable. Program may slow down
considerably on machines where table references are slow as the
program makes extensive use of them. Also, due to the huge amount
of data sent to the X-server, considerable slowdown can be noticed
if the Xserver is slow or especially if the connection between the
Xserver and the client is slow. (Don't try this on 9600 bps modem
line. I did ;)
Obfuscation is done on different levels. Almost none of it is made
at the macro level, so cpp:ing the code won't give you
much. Perhaps most notable is the algorithmic obfuscation. Simple
formula has been turned into something completely different. Also
it may be a surprise to find out which piece of the code
calculates non-stable fractals and how it does that. Many
obfuscations were left out of the program because speed of
animation was one of the key issues. Well, obfuscated or not,
some of the effects are very nice ;)
Memory refreshment:
Mandelbrot set is calculated from formula
2
Z = Z + Z
n+1 n 0
and Julian set is calculated from formula
and the Julian set is calculated from the formula
2
Z = Z + C
n+1 n
where C is the same throughout the picture. Both
calculation are finished when |Z| >= 2.0. For the other
fractals, the basic methods are the same but formula is
different (look from the source ;)
Note that Mandelbrot and Julian on this program aren't
exactly correct. Find out why ;)
where `C` is the same throughout the picture. Both calculation are finished
when `|Z| >= 2.0`. For the other fractals, the basic methods are the same but
the formula is different (look at the source ;)
Note that Mandelbrot and Julian on this program aren't exactly correct. Find out
why ;)
Bugs:
### Bugs:
- No error checking. If DISPLAY is not set program dumps core.
This may also happen if a request of some of the resources
fails. eg. getting colors.
- No expose event handling
- It may be possible that the program starts drawing before
window is mapped as there is no XSync(). In that case drawn
data is lost. Also, some window manager configurations places
the second window on top of the first and and again drawn data
will be lost. By default, windows are placed next to the other
in the topleft corner of the screen.
- Zooming is not possible
- Eats LOTS of memory on BIG pictures. Also lot's of CPU.
- Under mode 0, the boundary condition is not checked properly.
The condition |Z| >= 2.0 should be checked exactly (by comparing
x*x+y*y with 2*2), or else artifacts are introduced. The result
of this error is a "bubbly" appearance (at least in mono).
- No error checking. If DISPLAY is not set program dumps core. This may also
happen if a request of some of the resources fails e.g. getting colors.
- No expose event handling.
- It may be possible that the program starts drawing before window is mapped as
there is no XSync(). In that case drawn data is lost. Also, some window manager
configurations places the second window on top of the first and again drawn
data will be lost. By default, windows are placed next to the other in the
top left corner of the screen.
- Zooming is not possible.
- Eats LOTS of memory on BIG pictures. Also lots of CPU resources. Under mode
0, the boundary condition is not checked properly. The condition `|Z| >= 2.0`
should be checked exactly (by comparing `x*x+y*y` with `2*2`), or else artifacts
are introduced. The result of this error is a "bubbly" appearance (at least in
mono).
## Copyright and CC BY-SA 4.0 License: