how do i generate a random float between 0 and 1??

[thre3dee]

Hi, i need to generate a random float number between 0.0 and 1.0 for use in my OpenGL program but i can't seem to get it to work.

i tried
float r = (float) rand () / (GLfloat) RAND_MAX;
but doesn't work coz my red colour ends up being 0.

[Ejaz]

#include <time.h>
#include <iostream>

using std::cout;
using std::endl;

int main()
{
    srand( (unsigned)time( NULL ) );

    for (int i = 0; i < 5; i++) 
	{
		cout << (float) rand()/RAND_MAX << endl;
    }

    return 0;
}

[NMTop40]

RAND_MAX is often 32767. If you want to make use of the full 23 bits of mantissa then:

const unsigned long RAND_MAX_PLUS_1 = RAND_MAX + 1;

unsigned long bigRand()
{
   return rand() * (RAND_MAX_PLUS_1) + rand();
}

the new RAND_MAX will be (RAND_MAX+1)²-1

const unsigned long BIG_RAND_MAX = RAND_MAX_PLUS_1 * RAND_MAX_PLUS_1 - 1;
float rand0to1()
{
   return static_cast<float>(bigRand() ) / BIG_RAND_MAX;
}

(You can also make BIG_RAND_MAX a float or a double instead of an unsigned long).
Note if you want to make use of the precision of double, you could require 4 calls to rand().

[Marc G]

To get a better random number generator search for "Mersenne twister".

[RoboTact]

It's not good to rely on certain value of RAND_MAX constant. It's better to do smth like this:

float random()
{
  float scale=RAND_MAX+1.;
  float base=rand()/scale;
  float fine=rand()/scale;
  return base+fine/scale;
}

[NMTop40]

Yes it is better to use RAND_MAX as a float or double because my unsigned long may well overflow.

The compiler may be able to optimise your division by scale but I wouldn't count on it. If you make scale a const float it is more likely to.

I don't know if you can assume that RAND_MAX will always be (2^N) - 1 for some N.

[Marc G]

With all this playing with rand_max etc you are not sure if the resulting random numbers will be truly (pseudo-)random. Therefore, I suggest to use a Mersenne twister which can generate a floating point random number in 0-1.

[Improving]

boost random numbers library

[thre3dee]

no i fixed it i was setting it to a local float not the class member float so the class member waasn't actually getting the value.

[KevinHall]

Read this FAQ.

[flaviorodriguez]

My humble solution


float rand_FloatRange(float a, float b)
{
return ((b-a)*((float)rand()/RAND_MAX))+a;
}

Work fine with any range. All I did was to treat rand() as % and multiply that result by range between 'a' and 'b'