• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //---------------------------------------------------------------------------------
2 //
3 //  Little Color Management System
4 //  Copyright (c) 1998-2023 Marti Maria Saguer
5 //
6 // Permission is hereby granted, free of charge, to any person obtaining
7 // a copy of this software and associated documentation files (the "Software"),
8 // to deal in the Software without restriction, including without limitation
9 // the rights to use, copy, modify, merge, publish, distribute, sublicense,
10 // and/or sell copies of the Software, and to permit persons to whom the Software
11 // is furnished to do so, subject to the following conditions:
12 //
13 // The above copyright notice and this permission notice shall be included in
14 // all copies or substantial portions of the Software.
15 //
16 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
18 // THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
19 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
20 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
21 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
22 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 //
24 //---------------------------------------------------------------------------------
25 //
26 
27 #include "lcms2_internal.h"
28 
29 
30 #define DSWAP(x, y)     {cmsFloat64Number tmp = (x); (x)=(y); (y)=tmp;}
31 
32 
33 // Initiate a vector
_cmsVEC3init(cmsVEC3 * r,cmsFloat64Number x,cmsFloat64Number y,cmsFloat64Number z)34 void CMSEXPORT _cmsVEC3init(cmsVEC3* r, cmsFloat64Number x, cmsFloat64Number y, cmsFloat64Number z)
35 {
36     r -> n[VX] = x;
37     r -> n[VY] = y;
38     r -> n[VZ] = z;
39 }
40 
41 // Vector subtraction
_cmsVEC3minus(cmsVEC3 * r,const cmsVEC3 * a,const cmsVEC3 * b)42 void CMSEXPORT _cmsVEC3minus(cmsVEC3* r, const cmsVEC3* a, const cmsVEC3* b)
43 {
44   r -> n[VX] = a -> n[VX] - b -> n[VX];
45   r -> n[VY] = a -> n[VY] - b -> n[VY];
46   r -> n[VZ] = a -> n[VZ] - b -> n[VZ];
47 }
48 
49 // Vector cross product
_cmsVEC3cross(cmsVEC3 * r,const cmsVEC3 * u,const cmsVEC3 * v)50 void CMSEXPORT _cmsVEC3cross(cmsVEC3* r, const cmsVEC3* u, const cmsVEC3* v)
51 {
52     r ->n[VX] = u->n[VY] * v->n[VZ] - v->n[VY] * u->n[VZ];
53     r ->n[VY] = u->n[VZ] * v->n[VX] - v->n[VZ] * u->n[VX];
54     r ->n[VZ] = u->n[VX] * v->n[VY] - v->n[VX] * u->n[VY];
55 }
56 
57 // Vector dot product
_cmsVEC3dot(const cmsVEC3 * u,const cmsVEC3 * v)58 cmsFloat64Number CMSEXPORT _cmsVEC3dot(const cmsVEC3* u, const cmsVEC3* v)
59 {
60     return u->n[VX] * v->n[VX] + u->n[VY] * v->n[VY] + u->n[VZ] * v->n[VZ];
61 }
62 
63 // Euclidean length
_cmsVEC3length(const cmsVEC3 * a)64 cmsFloat64Number CMSEXPORT _cmsVEC3length(const cmsVEC3* a)
65 {
66     return sqrt(a ->n[VX] * a ->n[VX] +
67                 a ->n[VY] * a ->n[VY] +
68                 a ->n[VZ] * a ->n[VZ]);
69 }
70 
71 // Euclidean distance
_cmsVEC3distance(const cmsVEC3 * a,const cmsVEC3 * b)72 cmsFloat64Number CMSEXPORT _cmsVEC3distance(const cmsVEC3* a, const cmsVEC3* b)
73 {
74     cmsFloat64Number d1 = a ->n[VX] - b ->n[VX];
75     cmsFloat64Number d2 = a ->n[VY] - b ->n[VY];
76     cmsFloat64Number d3 = a ->n[VZ] - b ->n[VZ];
77 
78     return sqrt(d1*d1 + d2*d2 + d3*d3);
79 }
80 
81 
82 
83 // 3x3 Identity
_cmsMAT3identity(cmsMAT3 * a)84 void CMSEXPORT _cmsMAT3identity(cmsMAT3* a)
85 {
86     _cmsVEC3init(&a-> v[0], 1.0, 0.0, 0.0);
87     _cmsVEC3init(&a-> v[1], 0.0, 1.0, 0.0);
88     _cmsVEC3init(&a-> v[2], 0.0, 0.0, 1.0);
89 }
90 
91 static
CloseEnough(cmsFloat64Number a,cmsFloat64Number b)92 cmsBool CloseEnough(cmsFloat64Number a, cmsFloat64Number b)
93 {
94     return fabs(b - a) < (1.0 / 65535.0);
95 }
96 
97 
_cmsMAT3isIdentity(const cmsMAT3 * a)98 cmsBool CMSEXPORT _cmsMAT3isIdentity(const cmsMAT3* a)
99 {
100     cmsMAT3 Identity;
101     int i, j;
102 
103     _cmsMAT3identity(&Identity);
104 
105     for (i=0; i < 3; i++)
106         for (j=0; j < 3; j++)
107             if (!CloseEnough(a ->v[i].n[j], Identity.v[i].n[j])) return FALSE;
108 
109     return TRUE;
110 }
111 
112 
113 // Multiply two matrices
_cmsMAT3per(cmsMAT3 * r,const cmsMAT3 * a,const cmsMAT3 * b)114 void CMSEXPORT _cmsMAT3per(cmsMAT3* r, const cmsMAT3* a, const cmsMAT3* b)
115 {
116 #define ROWCOL(i, j) \
117     a->v[i].n[0]*b->v[0].n[j] + a->v[i].n[1]*b->v[1].n[j] + a->v[i].n[2]*b->v[2].n[j]
118 
119     _cmsVEC3init(&r-> v[0], ROWCOL(0,0), ROWCOL(0,1), ROWCOL(0,2));
120     _cmsVEC3init(&r-> v[1], ROWCOL(1,0), ROWCOL(1,1), ROWCOL(1,2));
121     _cmsVEC3init(&r-> v[2], ROWCOL(2,0), ROWCOL(2,1), ROWCOL(2,2));
122 
123 #undef ROWCOL //(i, j)
124 }
125 
126 
127 
128 // Inverse of a matrix b = a^(-1)
_cmsMAT3inverse(const cmsMAT3 * a,cmsMAT3 * b)129 cmsBool  CMSEXPORT _cmsMAT3inverse(const cmsMAT3* a, cmsMAT3* b)
130 {
131    cmsFloat64Number det, c0, c1, c2;
132 
133    c0 =  a -> v[1].n[1]*a -> v[2].n[2] - a -> v[1].n[2]*a -> v[2].n[1];
134    c1 = -a -> v[1].n[0]*a -> v[2].n[2] + a -> v[1].n[2]*a -> v[2].n[0];
135    c2 =  a -> v[1].n[0]*a -> v[2].n[1] - a -> v[1].n[1]*a -> v[2].n[0];
136 
137    det = a -> v[0].n[0]*c0 + a -> v[0].n[1]*c1 + a -> v[0].n[2]*c2;
138 
139    if (fabs(det) < MATRIX_DET_TOLERANCE) return FALSE;  // singular matrix; can't invert
140 
141    b -> v[0].n[0] = c0/det;
142    b -> v[0].n[1] = (a -> v[0].n[2]*a -> v[2].n[1] - a -> v[0].n[1]*a -> v[2].n[2])/det;
143    b -> v[0].n[2] = (a -> v[0].n[1]*a -> v[1].n[2] - a -> v[0].n[2]*a -> v[1].n[1])/det;
144    b -> v[1].n[0] = c1/det;
145    b -> v[1].n[1] = (a -> v[0].n[0]*a -> v[2].n[2] - a -> v[0].n[2]*a -> v[2].n[0])/det;
146    b -> v[1].n[2] = (a -> v[0].n[2]*a -> v[1].n[0] - a -> v[0].n[0]*a -> v[1].n[2])/det;
147    b -> v[2].n[0] = c2/det;
148    b -> v[2].n[1] = (a -> v[0].n[1]*a -> v[2].n[0] - a -> v[0].n[0]*a -> v[2].n[1])/det;
149    b -> v[2].n[2] = (a -> v[0].n[0]*a -> v[1].n[1] - a -> v[0].n[1]*a -> v[1].n[0])/det;
150 
151    return TRUE;
152 }
153 
154 
155 // Solve a system in the form Ax = b
_cmsMAT3solve(cmsVEC3 * x,cmsMAT3 * a,cmsVEC3 * b)156 cmsBool  CMSEXPORT _cmsMAT3solve(cmsVEC3* x, cmsMAT3* a, cmsVEC3* b)
157 {
158     cmsMAT3 m, a_1;
159 
160     memmove(&m, a, sizeof(cmsMAT3));
161 
162     if (!_cmsMAT3inverse(&m, &a_1)) return FALSE;  // Singular matrix
163 
164     _cmsMAT3eval(x, &a_1, b);
165     return TRUE;
166 }
167 
168 // Evaluate a vector across a matrix
_cmsMAT3eval(cmsVEC3 * r,const cmsMAT3 * a,const cmsVEC3 * v)169 void CMSEXPORT _cmsMAT3eval(cmsVEC3* r, const cmsMAT3* a, const cmsVEC3* v)
170 {
171     r->n[VX] = a->v[0].n[VX]*v->n[VX] + a->v[0].n[VY]*v->n[VY] + a->v[0].n[VZ]*v->n[VZ];
172     r->n[VY] = a->v[1].n[VX]*v->n[VX] + a->v[1].n[VY]*v->n[VY] + a->v[1].n[VZ]*v->n[VZ];
173     r->n[VZ] = a->v[2].n[VX]*v->n[VX] + a->v[2].n[VY]*v->n[VY] + a->v[2].n[VZ]*v->n[VZ];
174 }
175 
176 
177