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
|
/*
* Copyright 2017 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.
*/
#ifdef WITH_NANOVDB
# include "nanovdb/NanoVDB.h"
# include "nanovdb/util/SampleFromVoxels.h"
#endif
/* w0, w1, w2, and w3 are the four cubic B-spline basis functions. */
ccl_device float cubic_w0(float a)
{
return (1.0f / 6.0f) * (a * (a * (-a + 3.0f) - 3.0f) + 1.0f);
}
ccl_device float cubic_w1(float a)
{
return (1.0f / 6.0f) * (a * a * (3.0f * a - 6.0f) + 4.0f);
}
ccl_device float cubic_w2(float a)
{
return (1.0f / 6.0f) * (a * (a * (-3.0f * a + 3.0f) + 3.0f) + 1.0f);
}
ccl_device float cubic_w3(float a)
{
return (1.0f / 6.0f) * (a * a * a);
}
/* g0 and g1 are the two amplitude functions. */
ccl_device float cubic_g0(float a)
{
return cubic_w0(a) + cubic_w1(a);
}
ccl_device float cubic_g1(float a)
{
return cubic_w2(a) + cubic_w3(a);
}
/* h0 and h1 are the two offset functions */
ccl_device float cubic_h0(float a)
{
/* Note +0.5 offset to compensate for CUDA linear filtering convention. */
return -1.0f + cubic_w1(a) / (cubic_w0(a) + cubic_w1(a)) + 0.5f;
}
ccl_device float cubic_h1(float a)
{
return 1.0f + cubic_w3(a) / (cubic_w2(a) + cubic_w3(a)) + 0.5f;
}
/* Fast bicubic texture lookup using 4 bilinear lookups, adapted from CUDA samples. */
template<typename T>
ccl_device T kernel_tex_image_interp_bicubic(const TextureInfo &info, float x, float y)
{
CUtexObject tex = (CUtexObject)info.data;
x = (x * info.width) - 0.5f;
y = (y * info.height) - 0.5f;
float px = floor(x);
float py = floor(y);
float fx = x - px;
float fy = y - py;
float g0x = cubic_g0(fx);
float g1x = cubic_g1(fx);
float x0 = (px + cubic_h0(fx)) / info.width;
float x1 = (px + cubic_h1(fx)) / info.width;
float y0 = (py + cubic_h0(fy)) / info.height;
float y1 = (py + cubic_h1(fy)) / info.height;
return cubic_g0(fy) * (g0x * tex2D<T>(tex, x0, y0) + g1x * tex2D<T>(tex, x1, y0)) +
cubic_g1(fy) * (g0x * tex2D<T>(tex, x0, y1) + g1x * tex2D<T>(tex, x1, y1));
}
/* Fast tricubic texture lookup using 8 trilinear lookups. */
template<typename T>
ccl_device T kernel_tex_image_interp_bicubic_3d(const TextureInfo &info, float x, float y, float z)
{
CUtexObject tex = (CUtexObject)info.data;
x = (x * info.width) - 0.5f;
y = (y * info.height) - 0.5f;
z = (z * info.depth) - 0.5f;
float px = floor(x);
float py = floor(y);
float pz = floor(z);
float fx = x - px;
float fy = y - py;
float fz = z - pz;
float g0x = cubic_g0(fx);
float g1x = cubic_g1(fx);
float g0y = cubic_g0(fy);
float g1y = cubic_g1(fy);
float g0z = cubic_g0(fz);
float g1z = cubic_g1(fz);
float x0 = (px + cubic_h0(fx)) / info.width;
float x1 = (px + cubic_h1(fx)) / info.width;
float y0 = (py + cubic_h0(fy)) / info.height;
float y1 = (py + cubic_h1(fy)) / info.height;
float z0 = (pz + cubic_h0(fz)) / info.depth;
float z1 = (pz + cubic_h1(fz)) / info.depth;
return g0z * (g0y * (g0x * tex3D<T>(tex, x0, y0, z0) + g1x * tex3D<T>(tex, x1, y0, z0)) +
g1y * (g0x * tex3D<T>(tex, x0, y1, z0) + g1x * tex3D<T>(tex, x1, y1, z0))) +
g1z * (g0y * (g0x * tex3D<T>(tex, x0, y0, z1) + g1x * tex3D<T>(tex, x1, y0, z1)) +
g1y * (g0x * tex3D<T>(tex, x0, y1, z1) + g1x * tex3D<T>(tex, x1, y1, z1)));
}
#ifdef WITH_NANOVDB
template<typename T>
ccl_device_inline T kernel_tex_image_interp_nanovdb(
const TextureInfo &info, float x, float y, float z, uint interpolation)
{
nanovdb::NanoGrid<T> *const grid = (nanovdb::NanoGrid<T> *)info.data;
const nanovdb::NanoRoot<T> &root = grid->tree().root();
const nanovdb::Coord off(root.bbox().min());
const nanovdb::Coord dim(root.bbox().dim());
const nanovdb::Vec3f xyz(off[0] + x * dim[0], off[1] + y * dim[1], off[2] + z * dim[2]);
typedef nanovdb::ReadAccessor<nanovdb::NanoRoot<T>> ReadAccessorT;
switch (interpolation) {
default:
case INTERPOLATION_LINEAR:
return nanovdb::SampleFromVoxels<ReadAccessorT, 1, false>(root)(xyz);
case INTERPOLATION_CLOSEST:
return nanovdb::SampleFromVoxels<ReadAccessorT, 0, false>(root)(xyz);
case INTERPOLATION_CUBIC:
return nanovdb::SampleFromVoxels<ReadAccessorT, 3, false>(root)(xyz);
}
}
#endif
ccl_device float4 kernel_tex_image_interp(KernelGlobals *kg, int id, float x, float y)
{
const TextureInfo &info = kernel_tex_fetch(__texture_info, id);
/* float4, byte4, ushort4 and half4 */
const int texture_type = info.data_type;
if (texture_type == IMAGE_DATA_TYPE_FLOAT4 || texture_type == IMAGE_DATA_TYPE_BYTE4 ||
texture_type == IMAGE_DATA_TYPE_HALF4 || texture_type == IMAGE_DATA_TYPE_USHORT4) {
if (info.interpolation == INTERPOLATION_CUBIC) {
return kernel_tex_image_interp_bicubic<float4>(info, x, y);
}
else {
CUtexObject tex = (CUtexObject)info.data;
return tex2D<float4>(tex, x, y);
}
}
/* float, byte and half */
else {
float f;
if (info.interpolation == INTERPOLATION_CUBIC) {
f = kernel_tex_image_interp_bicubic<float>(info, x, y);
}
else {
CUtexObject tex = (CUtexObject)info.data;
f = tex2D<float>(tex, x, y);
}
return make_float4(f, f, f, 1.0f);
}
}
ccl_device float4 kernel_tex_image_interp_3d(KernelGlobals *kg,
int id,
float3 P,
InterpolationType interp)
{
const TextureInfo &info = kernel_tex_fetch(__texture_info, id);
if (info.use_transform_3d) {
P = transform_point(&info.transform_3d, P);
}
const float x = P.x;
const float y = P.y;
const float z = P.z;
uint interpolation = (interp == INTERPOLATION_NONE) ? info.interpolation : interp;
const int texture_type = info.data_type;
#ifdef WITH_NANOVDB
if (texture_type == IMAGE_DATA_TYPE_NANOVDB_FLOAT) {
float f = kernel_tex_image_interp_nanovdb<float>(info, x, y, z, interpolation);
return make_float4(f, f, f, 1.0f);
}
if (texture_type == IMAGE_DATA_TYPE_NANOVDB_FLOAT3) {
nanovdb::Vec3f f = kernel_tex_image_interp_nanovdb<nanovdb::Vec3f>(
info, x, y, z, interpolation);
return make_float4(f[0], f[1], f[2], 1.0f);
}
#endif
if (texture_type == IMAGE_DATA_TYPE_FLOAT4 || texture_type == IMAGE_DATA_TYPE_BYTE4 ||
texture_type == IMAGE_DATA_TYPE_HALF4 || texture_type == IMAGE_DATA_TYPE_USHORT4) {
if (interpolation == INTERPOLATION_CUBIC) {
return kernel_tex_image_interp_bicubic_3d<float4>(info, x, y, z);
}
else {
CUtexObject tex = (CUtexObject)info.data;
return tex3D<float4>(tex, x, y, z);
}
}
else {
float f;
if (interpolation == INTERPOLATION_CUBIC) {
f = kernel_tex_image_interp_bicubic_3d<float>(info, x, y, z);
}
else {
CUtexObject tex = (CUtexObject)info.data;
f = tex3D<float>(tex, x, y, z);
}
return make_float4(f, f, f, 1.0f);
}
}
|