Some old code I recently saw again

Thu, 02 Feb 2006 20:34:01 -0600

Here is some code I wrote tht may be interesting to some of you. I was trying to work out how random-dot stereograms worked and decided to write some code to test some methods.

It's ugly code. ;] But it kinda works, and i think it shows how to do them. Oh yeah....it's all text.

to compile it, save it as stereo.c and then



$ cc -o stereo stereo.c -lm

$ ./stereo > stereo.txt

You may have to mess around a little to get the sizes you want.





#define HEIGHT 50

#define WIDTH 55



unsigned char screen[HEIGHT][WIDTH];

unsigned char map[HEIGHT][WIDTH];

unsigned char fill[11];



char f[]="!@#$%^&*()_+1234567890";





mkfill(char *s)

{

 int i;

 for(i=0;i<sizeof(fill);s[i]?i++:i)

 {

  s[i]=rand()%90 + ' ';

 }

}





int fn(int x,int y)

{

 double a,b,c;

 int ret;

 a=(WIDTH/2)-x;

 b=(HEIGHT/2)-y;

 c=(sqrt(a*a/4 + b*b)/4);

 ret=c>4?0:8-c;

 if(a>0)ret=-ret;

 return (sizeof(fill)+ret)%sizeof(fill);

}



main(int argc,char *argv[])

{

 int h,w;

 srand(time(0)^getpid());

 for(h=0;h<HEIGHT;h++)

 {

  for(w=0;w<WIDTH;w++)

  {

   map[h][w]=fn(w,h);

  }

 }





 if(argc<2)

 for(h=0;h<HEIGHT;h++)

 {

  int o;

  mkfill(fill);

  for(w=0;w<WIDTH;w++)

  {

   o = map[h][w];

   screen[h][w]=fill[(o+w)%sizeof(fill)];

  }

  printf("\t");

  printf("%s%s",fill,fill);

  printf("%s%s",fill,fill);

  for(w=0;w<WIDTH;w++)

  {

   printf("%c",screen[h][w]);

  }

  printf("%s%s",fill,fill);

  printf("%s%s",fill,fill);

  printf("\n");

 }

 else

 {

  for(h=0;h<HEIGHT;h++)

  {

   printf("\t");

   for(w=0;w<WIDTH;w++)

   {

    printf("%c",map[h][w]+'0');

   }

   printf("\n");

  }

 }

}

Coming soon (maybe)... animated ascii random dot stereograms.
(not sure yet if the maybe is the 'soon' or the 'coming' part.

I guess I'll put a some code here...

Fri, 14 Oct 2005 12:34:03 -0500

30 Byte Demo

Sometime in the last year or so a friend on IRC mentioned he had written something for a sort of contest to write to coolest MS-DOS graphics demo. The catch was you only had 32 bytes. The contest was over by the time I heard about it, and it's been a while since then and I have no idea which site it was now, but here is some code I wrote that, IMNSHO, kicked the butts of all the entries in that contest.

In order to fully appreciate it, make sure you let it color cycle once thry the 256 colors. You'll know when it's getting good. ;]

                                                           

 mov ax,13h           ; al =0x13 , ah = 0

 int 10h              ; set video mode 0x13 (320x200x8bit)

 mov bx,0a000h        ; address a000:0000  == start of VGA memory

 mov ds,bx            ; so set the data segment to be that

 xor cx,cx            ; give me something to be 0

top_of_loop:

 mov al,[si]          ; grab one byte from video memory

 inc ax               ; increment the word, not the byte

 add al,ah            ; and add any overflow to the value (clever part ;])

 mov [si], al         ; store it back in video ram

 inc si               ; curpixel++

 cmp si, 0fa00h       ; curpixel != 64000 == 320 * 200 ?

 jnz top_of_loop      ; if so, do the next pixel

 mov si,cx            ; otherwise, reset curpixel

 jmp top_of_loop      ; and start from top of screen

I wrote this in a short space of time, and didn't really debug it very well. Now that I look at the commented version I just made of it, I can
see a few subtle things that could've made it better.

I'll quickly make some changes, and see how it looks.

.... some times passes ;] ...

Ok... Now it is a
28 Byte Demo

                                                           

 mov ax,13h           ; al =0x13 , ah = 0

 int 10h              ; set video mode 0x13 (320x200x8bit)

 mov bx,0a000h        ; address a000:0000  == start of VGA memory

 mov ds,bx            ; so set the data segment to be that

init_loop:

 xor si,si            ; intialize the screen pointer

top_of_loop:

 mov al,[si]          ; grab one byte from video memory

 inc ax               ; increment the word, not the byte....

 add al,ah            ; and add any overflow to the value (clever part ;])

 mov [si], al         ; store it back in video ram

 inc si               ; curpixel++

 cmp si, 0fa00h       ; curpixel != 64000 == 320 * 200 ?

 jnz top_of_loop      ; if so, do the next pixel

 jmp init_loop      ; and start from top of screen

All I did was optimize out the the temporary cx variable.

How it works



/* 

 * You shouldn't expect this to actually work as C, but

 * it should give you a rough idea.

*/

setGraphicsMode(0x13);

char *a000=address_at_start_of_16bit_segment(0xa000);



while(1) // nearly forgot the important part

{

 for(i=0;i<64000;i++)

 {

  ax[1]=a000[i];

  ax++;    // ax == (ah<<8)|al   if al > 0xff, then al = 0, ah++

  al=al+ah; 

  a000[i]=al;

 }

}

Notice that ah is never being reset... once it starts to increase, it will continue to... lets imagine a simple screen where all the values are up to 0xff



| 0|   1    2    3

0xff 0xff 0xff 0xff  <- video memory

SI=0x0000

mov al,[si] -> AX=0x00FF  AH=0x00  AL=0xFF

inc ax      -> AX=0x0100  AH=0x01 AL=0x00

add al,ah   -> AX=0x0101

mov [si],al

inc si

so now the state looks like this:



  0  | 1|   2    3

0x01 0xff 0xff 0xff

SI=0x0001

mov al,[si] -> AX=0x01FF  AH=0x01  AL=0xFF

inc ax      -> AX=0x0200  AH=0x02 AL=0x00

add al,ah   -> AX=0x0202

mov [si],al

inc si

and one more time, so you see the pattern...



  0    1  | 2|   3

0x01 0x02 0xff 0xff

SI=0x0002

mov al,[si] -> AX=0x00FF  AH=0x00  AL=0xFF

inc ax      -> AX=0x0300  AH=0x03 AL=0x00

add al,ah   -> AX=0x0303

mov [si],al

inc si

  0    1    2  | 3|

0x01 0x02 0x03 0xff

The next time around, things get even more complex. Iterative generated chaotic stuff.

This is what makes it interesting. ;]

OpenRCE: Blog

Some old code I recently saw again

I guess I'll put a some code here...