New lightmapper

-Added LocalVector (needed it)
-Added stb_rect_pack (It's pretty cool, we could probably use it for other stuff too)
-Fixes and changes all around the place
-Added library for 128 bits fixed point (required for Delaunay3D)

core/math/basis.cpp

@@ -878,3 +878,114 @@ Basis Basis::slerp(const Basis &target, const real_t &t) const {
 
 	return b;
 }
+
+void Basis::rotate_sh(real_t *p_values) {
+
+	// code by John Hable
+	// http://filmicworlds.com/blog/simple-and-fast-spherical-harmonic-rotation/
+	// this code is Public Domain
+
+	const static real_t s_c3 = 0.94617469575; // (3*sqrt(5))/(4*sqrt(pi))
+	const static real_t s_c4 = -0.31539156525; // (-sqrt(5))/(4*sqrt(pi))
+	const static real_t s_c5 = 0.54627421529; // (sqrt(15))/(4*sqrt(pi))
+
+	const static real_t s_c_scale = 1.0 / 0.91529123286551084;
+	const static real_t s_c_scale_inv = 0.91529123286551084;
+
+	const static real_t s_rc2 = 1.5853309190550713 * s_c_scale;
+	const static real_t s_c4_div_c3 = s_c4 / s_c3;
+	const static real_t s_c4_div_c3_x2 = (s_c4 / s_c3) * 2.0;
+
+	const static real_t s_scale_dst2 = s_c3 * s_c_scale_inv;
+	const static real_t s_scale_dst4 = s_c5 * s_c_scale_inv;
+
+	real_t src[9] = { p_values[0], p_values[1], p_values[2], p_values[3], p_values[4], p_values[5], p_values[6], p_values[7], p_values[8] };
+
+	real_t m00 = elements[0][0];
+	real_t m01 = elements[0][1];
+	real_t m02 = elements[0][2];
+	real_t m10 = elements[1][0];
+	real_t m11 = elements[1][1];
+	real_t m12 = elements[1][2];
+	real_t m20 = elements[2][0];
+	real_t m21 = elements[2][1];
+	real_t m22 = elements[2][2];
+
+	p_values[0] = src[0];
+	p_values[1] = m11 * src[1] - m12 * src[2] + m10 * src[3];
+	p_values[2] = -m21 * src[1] + m22 * src[2] - m20 * src[3];
+	p_values[3] = m01 * src[1] - m02 * src[2] + m00 * src[3];
+
+	real_t sh0 = src[7] + src[8] + src[8] - src[5];
+	real_t sh1 = src[4] + s_rc2 * src[6] + src[7] + src[8];
+	real_t sh2 = src[4];
+	real_t sh3 = -src[7];
+	real_t sh4 = -src[5];
+
+	// Rotations.  R0 and R1 just use the raw matrix columns
+	real_t r2x = m00 + m01;
+	real_t r2y = m10 + m11;
+	real_t r2z = m20 + m21;
+
+	real_t r3x = m00 + m02;
+	real_t r3y = m10 + m12;
+	real_t r3z = m20 + m22;
+
+	real_t r4x = m01 + m02;
+	real_t r4y = m11 + m12;
+	real_t r4z = m21 + m22;
+
+	// dense matrix multiplication one column at a time
+
+	// column 0
+	real_t sh0_x = sh0 * m00;
+	real_t sh0_y = sh0 * m10;
+	real_t d0 = sh0_x * m10;
+	real_t d1 = sh0_y * m20;
+	real_t d2 = sh0 * (m20 * m20 + s_c4_div_c3);
+	real_t d3 = sh0_x * m20;
+	real_t d4 = sh0_x * m00 - sh0_y * m10;
+
+	// column 1
+	real_t sh1_x = sh1 * m02;
+	real_t sh1_y = sh1 * m12;
+	d0 += sh1_x * m12;
+	d1 += sh1_y * m22;
+	d2 += sh1 * (m22 * m22 + s_c4_div_c3);
+	d3 += sh1_x * m22;
+	d4 += sh1_x * m02 - sh1_y * m12;
+
+	// column 2
+	real_t sh2_x = sh2 * r2x;
+	real_t sh2_y = sh2 * r2y;
+	d0 += sh2_x * r2y;
+	d1 += sh2_y * r2z;
+	d2 += sh2 * (r2z * r2z + s_c4_div_c3_x2);
+	d3 += sh2_x * r2z;
+	d4 += sh2_x * r2x - sh2_y * r2y;
+
+	// column 3
+	real_t sh3_x = sh3 * r3x;
+	real_t sh3_y = sh3 * r3y;
+	d0 += sh3_x * r3y;
+	d1 += sh3_y * r3z;
+	d2 += sh3 * (r3z * r3z + s_c4_div_c3_x2);
+	d3 += sh3_x * r3z;
+	d4 += sh3_x * r3x - sh3_y * r3y;
+
+	// column 4
+	real_t sh4_x = sh4 * r4x;
+	real_t sh4_y = sh4 * r4y;
+	d0 += sh4_x * r4y;
+	d1 += sh4_y * r4z;
+	d2 += sh4 * (r4z * r4z + s_c4_div_c3_x2);
+	d3 += sh4_x * r4z;
+	d4 += sh4_x * r4x - sh4_y * r4y;
+
+	// extra multipliers
+	p_values[4] = d0;
+	p_values[5] = -d1;
+	p_values[6] = d2 * s_scale_dst2;
+	p_values[7] = -d3;
+	p_values[8] = d4 * s_scale_dst4;
+}