Planetarion Forums

Structural Integrity · 1 Dec 2002, 22:10

Is there a function with which I can calculate a cos/sin in degrees instead of radians?
If not, can anyone tell me in which header file PI is defined... it doesn't seem to be in <math.h>

TY

Ragnarak · 1 Dec 2002, 22:33

can't you just use the deg2rad and rad2deg functions?

not 100% if they are in any of the headers, i seem to remember having to write them myself when learning what functions did but they're very simple anyway

Ragnarak · 1 Dec 2002, 22:36

oh and i don't think Pi is defined in any of the standard headers, we always had to define it at the start

SnAzBaZ · 1 Dec 2002, 23:06

#define PI 3.14159.......

int degrees = 30;
sin((degrees * PI)/180);

meglamaniac · 1 Dec 2002, 23:31

M_PI is defined in math.h under linux gcc (at least, i think it's called M_PI).

What I've allways wanted to know is how do you do inverse sin/cos/tan in C/C++? I've not seen a function that does it. Presumably its a bit of simple maths, but I havn't a clue how you do it...

MT · 1 Dec 2002, 23:51

Code:

[email protected] [10:54pm]  '~'
> $ cat megla.c
#include <math.h>
#include <stdio.h>

int main() {
  double angle=M_PI/2,sine;
  sine = sin(angle);
  printf("sine of %f is %f\n",angle,sine);
  printf("asin(%f) is %f\n",sine,asin(sine));
  return 0;
}
[email protected] [10:55pm]  '~'
> $ gcc -o megla megla.c -Wall -lm
[email protected] [10:55pm]  '~'
> $ ./megla
sine of 1.570796 is 1.000000
asin(1.000000) is 1.570796

meglamaniac · 1 Dec 2002, 23:59

You the daddeh.
I really need to dredge up some of that maths 'stuff' I learnt at AS. Oh well, at least I got the Pi constant right.

W · 2 Dec 2002, 01:26

Quote:

Originally posted by meglamaniac
M_PI is defined in math.h under linux gcc (at least, i think it's called M_PI).

What I've allways wanted to know is how do you do inverse sin/cos/tan in C/C++? I've not seen a function that does it. Presumably its a bit of simple maths, but I havn't a clue how you do it...

You have atan, which you can derive asin and acos from

Structural Integrity · 2 Dec 2002, 07:29

I do remember using PI once.... I can't remember if it was under Borland or VC++ tho (or was that in Java, hmmm). deg2rad isn't a standard function either.

BTW, anyone an opinion about an acceptable precision of PI? I'm using it for a rotation matrix and in combination with (float)'s.

Nodrog · 2 Dec 2002, 16:04

Quote:

Originally posted by Structural Integrity

BTW, anyone an opinion about an acceptable precision of PI? I'm using it for a rotation matrix and in combination with (float)'s.

Since theres no performance bonus whatsoever by using lower precision, youd be as well just using 3.141592653589796 or something. I doubt you'd actually need greater precision than 3.141592 in 90% of cases though.

mikay · 2 Dec 2002, 20:35

Quote:

Originally posted by Structural Integrity
I do remember using PI once.... I can't remember if it was under Borland or VC++ tho (or was that in Java, hmmm). deg2rad isn't a standard function either.

BTW, anyone an opinion about an acceptable precision of PI? I'm using it for a rotation matrix and in combination with (float)'s.

Way back when I dabbled in 3D (about 10 years ago using assembler) I just used a lookup table. 1 byte = 0-127 signed then rotate the right 8 bits off. It gives 1% accuracy which is fine for most things, especially if it's moving about 20-30fps or faster. Dunno how efficiently this would translate to C tho and maths performance is a lot better nowadays.

MT · 3 Dec 2002, 00:43

Quote:

Originally posted by Structural Integrity
... deg2rad isn't a standard function either.

Code:

#include <math.h>
#include <stdio.h>
#ifndef _RAD_DEG
#define _RAD_DEG
#define DEG2RAD(x) ((x) * M_PI / 180)
#define RAD2DEG(x) ((x) * 180 / M_PI)
#endif

int main() {
  double angle=90.0,sine;
  sine = sin(DEG2RAD(angle));
  printf("sine of %f is %f\n",angle,sine);
  printf("asin(%f) is %f\n",sine,RAD2DEG(asin(sine)));
  return 0;
}

etc.

W · 3 Dec 2002, 00:59

Oh, and a precalculated table is of course much much faster if you don't need 100% accuracy.

Cyp · 3 Dec 2002, 08:47

I normally just type
#define PI 3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825
#define PI2 (2.*PI)

(Or sometimes just cut and paste, if I can remember the name of a program where I used π and could be bothered to open the code. (Usually can't, so just type it.))

Anyway, then you can use sin(θ*(PI/180.)) and cos(θ*(PI/180.))

Quote:

Originally posted by Nodrog
Since theres no performance bonus whatsoever by using lower precision, youd be as well just using 3.141592653589796 or something. I doubt you'd actually need greater precision than 3.141592 in 90% of cases though.

I suppose you meant 3.141592653589793...

Quote:

Originally posted by Structural Integrity
I do remember using PI once.... I can't remember if it was under Borland or VC++ tho (or was that in Java, hmmm). deg2rad isn't a standard function either.

BTW, anyone an opinion about an acceptable precision of PI? I'm using it for a rotation matrix and in combination with (float)'s.

I think 90 decimal places is usually sufficient for most purposes.

Quaternions can be useful for rotation - rather homogenous, and no sin/cos involved. And easy to combine two rotations, without turning them into matrixes first. (Depends on what you want to rotate, and how you want to rotate it, I guess.)