1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
|
/*
* Copyright 2011-2013 Blender Foundation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef __UTIL_COLOR_H__
#define __UTIL_COLOR_H__
#include "util/util_math.h"
#include "util/util_types.h"
#if !defined(__KERNEL_GPU__) && defined(__KERNEL_SSE2__)
# include "util/util_simd.h"
#endif
CCL_NAMESPACE_BEGIN
ccl_device uchar float_to_byte(float val)
{
return ((val <= 0.0f) ? 0 :
((val > (1.0f - 0.5f / 255.0f)) ? 255 : (uchar)((255.0f * val) + 0.5f)));
}
ccl_device uchar4 color_float_to_byte(float3 c)
{
uchar r, g, b;
r = float_to_byte(c.x);
g = float_to_byte(c.y);
b = float_to_byte(c.z);
return make_uchar4(r, g, b, 0);
}
ccl_device uchar4 color_float4_to_uchar4(float4 c)
{
uchar r, g, b, a;
r = float_to_byte(c.x);
g = float_to_byte(c.y);
b = float_to_byte(c.z);
a = float_to_byte(c.w);
return make_uchar4(r, g, b, a);
}
ccl_device_inline float3 color_byte_to_float(uchar4 c)
{
return make_float3(c.x * (1.0f / 255.0f), c.y * (1.0f / 255.0f), c.z * (1.0f / 255.0f));
}
ccl_device_inline float4 color_uchar4_to_float4(uchar4 c)
{
return make_float4(
c.x * (1.0f / 255.0f), c.y * (1.0f / 255.0f), c.z * (1.0f / 255.0f), c.w * (1.0f / 255.0f));
}
ccl_device float color_srgb_to_linear(float c)
{
if (c < 0.04045f)
return (c < 0.0f) ? 0.0f : c * (1.0f / 12.92f);
else
return powf((c + 0.055f) * (1.0f / 1.055f), 2.4f);
}
ccl_device float color_linear_to_srgb(float c)
{
if (c < 0.0031308f)
return (c < 0.0f) ? 0.0f : c * 12.92f;
else
return 1.055f * powf(c, 1.0f / 2.4f) - 0.055f;
}
ccl_device float3 rgb_to_hsv(float3 rgb)
{
float cmax, cmin, h, s, v, cdelta;
float3 c;
cmax = fmaxf(rgb.x, fmaxf(rgb.y, rgb.z));
cmin = min(rgb.x, min(rgb.y, rgb.z));
cdelta = cmax - cmin;
v = cmax;
if (cmax != 0.0f) {
s = cdelta / cmax;
}
else {
s = 0.0f;
h = 0.0f;
}
if (s != 0.0f) {
float3 cmax3 = make_float3(cmax, cmax, cmax);
c = (cmax3 - rgb) / cdelta;
if (rgb.x == cmax)
h = c.z - c.y;
else if (rgb.y == cmax)
h = 2.0f + c.x - c.z;
else
h = 4.0f + c.y - c.x;
h /= 6.0f;
if (h < 0.0f)
h += 1.0f;
}
else {
h = 0.0f;
}
return make_float3(h, s, v);
}
ccl_device float3 hsv_to_rgb(float3 hsv)
{
float i, f, p, q, t, h, s, v;
float3 rgb;
h = hsv.x;
s = hsv.y;
v = hsv.z;
if (s != 0.0f) {
if (h == 1.0f)
h = 0.0f;
h *= 6.0f;
i = floorf(h);
f = h - i;
rgb = make_float3(f, f, f);
p = v * (1.0f - s);
q = v * (1.0f - (s * f));
t = v * (1.0f - (s * (1.0f - f)));
if (i == 0.0f)
rgb = make_float3(v, t, p);
else if (i == 1.0f)
rgb = make_float3(q, v, p);
else if (i == 2.0f)
rgb = make_float3(p, v, t);
else if (i == 3.0f)
rgb = make_float3(p, q, v);
else if (i == 4.0f)
rgb = make_float3(t, p, v);
else
rgb = make_float3(v, p, q);
}
else {
rgb = make_float3(v, v, v);
}
return rgb;
}
ccl_device float3 xyY_to_xyz(float x, float y, float Y)
{
float X, Z;
if (y != 0.0f)
X = (x / y) * Y;
else
X = 0.0f;
if (y != 0.0f && Y != 0.0f)
Z = (1.0f - x - y) / y * Y;
else
Z = 0.0f;
return make_float3(X, Y, Z);
}
#ifdef __KERNEL_SSE2__
/*
* Calculate initial guess for arg^exp based on float representation
* This method gives a constant bias,
* which can be easily compensated by multiplication with bias_coeff.
* Gives better results for exponents near 1 (e. g. 4/5).
* exp = exponent, encoded as uint32_t
* e2coeff = 2^(127/exponent - 127) * bias_coeff^(1/exponent), encoded as uint32_t
*/
template<unsigned exp, unsigned e2coeff> ccl_device_inline ssef fastpow(const ssef &arg)
{
ssef ret;
ret = arg * cast(ssei(e2coeff));
ret = ssef(cast(ret));
ret = ret * cast(ssei(exp));
ret = cast(ssei(ret));
return ret;
}
/* Improve x ^ 1.0f/5.0f solution with Newton-Raphson method */
ccl_device_inline ssef improve_5throot_solution(const ssef &old_result, const ssef &x)
{
ssef approx2 = old_result * old_result;
ssef approx4 = approx2 * approx2;
ssef t = x / approx4;
ssef summ = madd(ssef(4.0f), old_result, t);
return summ * ssef(1.0f / 5.0f);
}
/* Calculate powf(x, 2.4). Working domain: 1e-10 < x < 1e+10 */
ccl_device_inline ssef fastpow24(const ssef &arg)
{
/* max, avg and |avg| errors were calculated in gcc without FMA instructions
* The final precision should be better than powf in glibc */
/* Calculate x^4/5, coefficient 0.994 was constructed manually to minimize avg error */
/* 0x3F4CCCCD = 4/5 */
/* 0x4F55A7FB = 2^(127/(4/5) - 127) * 0.994^(1/(4/5)) */
ssef x = fastpow<0x3F4CCCCD, 0x4F55A7FB>(arg); // error max = 0.17 avg = 0.0018 |avg| = 0.05
ssef arg2 = arg * arg;
ssef arg4 = arg2 * arg2;
x = improve_5throot_solution(x,
arg4); /* error max = 0.018 avg = 0.0031 |avg| = 0.0031 */
x = improve_5throot_solution(x,
arg4); /* error max = 0.00021 avg = 1.6e-05 |avg| = 1.6e-05 */
x = improve_5throot_solution(x,
arg4); /* error max = 6.1e-07 avg = 5.2e-08 |avg| = 1.1e-07 */
return x * (x * x);
}
ccl_device ssef color_srgb_to_linear(const ssef &c)
{
sseb cmp = c < ssef(0.04045f);
ssef lt = max(c * ssef(1.0f / 12.92f), ssef(0.0f));
ssef gtebase = (c + ssef(0.055f)) * ssef(1.0f / 1.055f); /* fma */
ssef gte = fastpow24(gtebase);
return select(cmp, lt, gte);
}
#endif /* __KERNEL_SSE2__ */
ccl_device float3 color_srgb_to_linear_v3(float3 c)
{
return make_float3(
color_srgb_to_linear(c.x), color_srgb_to_linear(c.y), color_srgb_to_linear(c.z));
}
ccl_device float3 color_linear_to_srgb_v3(float3 c)
{
return make_float3(
color_linear_to_srgb(c.x), color_linear_to_srgb(c.y), color_linear_to_srgb(c.z));
}
ccl_device float4 color_linear_to_srgb_v4(float4 c)
{
return make_float4(
color_linear_to_srgb(c.x), color_linear_to_srgb(c.y), color_linear_to_srgb(c.z), c.w);
}
ccl_device float4 color_srgb_to_linear_v4(float4 c)
{
#ifdef __KERNEL_SSE2__
ssef r_ssef;
float4 &r = (float4 &)r_ssef;
r = c;
r_ssef = color_srgb_to_linear(r_ssef);
r.w = c.w;
return r;
#else
return make_float4(
color_srgb_to_linear(c.x), color_srgb_to_linear(c.y), color_srgb_to_linear(c.z), c.w);
#endif
}
ccl_device float3 color_highlight_compress(float3 color, float3 *variance)
{
color += make_float3(1.0f, 1.0f, 1.0f);
if (variance) {
*variance *= sqr3(make_float3(1.0f, 1.0f, 1.0f) / color);
}
return log3(color);
}
ccl_device float3 color_highlight_uncompress(float3 color)
{
return exp3(color) - make_float3(1.0f, 1.0f, 1.0f);
}
CCL_NAMESPACE_END
#endif /* __UTIL_COLOR_H__ */
|