Newsgroups: comp.os.msdos.djgpp
From: tob AT world DOT std DOT com ("Tom Breton")
Subject: Re: About ray casting and look up tables
Message-ID: <rootsok9cpp1.fsf@world.std.com>
Sender: root AT pc
Organization: The World Public Access UNIX, Brookline, MA
References: <35A62FBB DOT AF821E92 AT cue DOT satnet DOT net>
Date: Fri, 10 Jul 1998 22:21:30 GMT
Lines: 70
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

"Ronald Patiņo G" <logan AT cue DOT satnet DOT net> writes:

> 
> Hi
> 
> I'm learning how to make a ray casting engine, i'm using
> djgpp, and a friend of mine sent me some source code, but
> i have a problem with look up tables.
> 
> I usually calculate them in this way
> 
> #include <math.h>
> float tan_table[320];
> float rad;
> for(ang=0;ang<320;ang++)
> {
>    rad= ang *(pi/180);

That seems to come out uneven. IE, you haven't filled the whole 360
degrees, tho you've filled the whole table.

>   tan_table[ang]=tan(rad);
> }
> 
> but i was cheking out my friend's code and found

Your friend writes nice code.
 
> 
> #include <math.h>
> #define ANG_360 1920

You seem to be expecting the same number of elements in the table as
degrees in an angle, but it hain't neccessarily so. ANG_360 is the
*number of array elements* in a *360* degree circle, which is the
whole array.

> #define ANG_0  0
> #define PI 3.1415927
> #define TWOPI  ( 2 * PI)
> #define TWOPIDIV360 ( TWOPI / ANG_360)

Here he defines his conversion ratio. Your friend seems to like to
write his stuff out explicitly, a habit I recommend.

> float *tan_table;
> 
> tan_table = (float *) malloc( sizeof(float) * ANG_360);

For stuff like this, I like to have an extra entry so that tan_table[
ANG_360 ] == tan_table[ 0 ], because that way there's always a
rounded-up element and a rounded-down element, which saves some tests.

> 
> for(angulo=ANG_0;angulo<=ANG_360;angulo++)
> 
> {
> 
>     ang_rad = ang * TWOPIDIV360 ;
> 
>    tan_table[ang] =  (float)tan(ang_rad);
> 
> }
> 

That's pretty much the same as you did. He's done the conversion
separately, and (properly) cast the result of tan ().


	Tom