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
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
|
/*
* Copyright 2014, 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.
*/
/* Triangle/Ray intersections.
*
* For BVH ray intersection we use a precomputed triangle storage to accelerate
* intersection at the cost of more memory usage.
*/
CCL_NAMESPACE_BEGIN
/* Workaround stupidness of CUDA/OpenCL which doesn't allow to access indexed
* component of float3 value.
*/
#ifndef __KERNEL_CPU__
# define IDX(vec, idx) \
((idx == 0) ? ((vec).x) : ( (idx == 1) ? ((vec).y) : ((vec).z) ))
#else
# define IDX(vec, idx) ((vec)[idx])
#endif
/* Ray-Triangle intersection for BVH traversal
*
* Sven Woop
* Watertight Ray/Triangle Intersection
*
* http://jcgt.org/published/0002/01/05/paper.pdf
*/
/* Precalculated data for the ray->tri intersection. */
typedef struct IsectPrecalc {
/* Maximal dimension kz, and orthogonal dimensions. */
int kx, ky, kz;
/* Shear constants. */
float Sx, Sy, Sz;
} IsectPrecalc;
/* Workaround for CUDA toolkit 6.5.16. */
#if defined(__KERNEL_CPU__) || !defined(__KERNEL_CUDA_EXPERIMENTAL__) || __CUDA_ARCH__ < 500
# if (defined(i386) || defined(_M_IX86))
ccl_device_noinline
# else
ccl_device_inline
# endif
#else
ccl_device_noinline
#endif
void triangle_intersect_precalc(float3 dir,
IsectPrecalc *isect_precalc)
{
/* Calculate dimension where the ray direction is maximal. */
int kz = util_max_axis(make_float3(fabsf(dir.x),
fabsf(dir.y),
fabsf(dir.z)));
int kx = kz + 1; if(kx == 3) kx = 0;
int ky = kx + 1; if(ky == 3) ky = 0;
/* Swap kx and ky dimensions to preserve winding direction of triangles. */
if(IDX(dir, kz) < 0.0f) {
int tmp = kx;
kx = ky;
ky = tmp;
}
/* Calculate the shear constants. */
float inv_dir_z = 1.0f / IDX(dir, kz);
isect_precalc->Sx = IDX(dir, kx) * inv_dir_z;
isect_precalc->Sy = IDX(dir, ky) * inv_dir_z;
isect_precalc->Sz = inv_dir_z;
/* Store the dimensions. */
isect_precalc->kx = kx;
isect_precalc->ky = ky;
isect_precalc->kz = kz;
}
/* TODO(sergey): Make it general utility function. */
ccl_device_inline float xor_signmast(float x, int y)
{
return __int_as_float(__float_as_int(x) ^ y);
}
ccl_device_inline bool triangle_intersect(KernelGlobals *kg,
const IsectPrecalc *isect_precalc,
Intersection *isect,
float3 P,
uint visibility,
int object,
int triAddr)
{
const int kx = isect_precalc->kx;
const int ky = isect_precalc->ky;
const int kz = isect_precalc->kz;
const float Sx = isect_precalc->Sx;
const float Sy = isect_precalc->Sy;
const float Sz = isect_precalc->Sz;
/* Calculate vertices relative to ray origin. */
const float4 tri_a = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+0),
tri_b = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+1),
tri_c = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+2);
const float3 A = make_float3(tri_a.x - P.x, tri_a.y - P.y, tri_a.z - P.z);
const float3 B = make_float3(tri_b.x - P.x, tri_b.y - P.y, tri_b.z - P.z);
const float3 C = make_float3(tri_c.x - P.x, tri_c.y - P.y, tri_c.z - P.z);
const float A_kx = IDX(A, kx), A_ky = IDX(A, ky), A_kz = IDX(A, kz);
const float B_kx = IDX(B, kx), B_ky = IDX(B, ky), B_kz = IDX(B, kz);
const float C_kx = IDX(C, kx), C_ky = IDX(C, ky), C_kz = IDX(C, kz);
/* Perform shear and scale of vertices. */
const float Ax = A_kx - Sx * A_kz;
const float Ay = A_ky - Sy * A_kz;
const float Bx = B_kx - Sx * B_kz;
const float By = B_ky - Sy * B_kz;
const float Cx = C_kx - Sx * C_kz;
const float Cy = C_ky - Sy * C_kz;
/* Calculate scaled barycentric coordinates. */
float U = Cx * By - Cy * Bx;
int sign_mask = (__float_as_int(U) & 0x80000000);
float V = Ax * Cy - Ay * Cx;
if(sign_mask != (__float_as_int(V) & 0x80000000)) {
return false;
}
float W = Bx * Ay - By * Ax;
if(sign_mask != (__float_as_int(W) & 0x80000000)) {
return false;
}
/* Calculate determinant. */
float det = U + V + W;
if(UNLIKELY(det == 0.0f)) {
return false;
}
/* Calculate scaled z−coordinates of vertices and use them to calculate
* the hit distance.
*/
const float T = (U * A_kz + V * B_kz + W * C_kz) * Sz;
const float sign_T = xor_signmast(T, sign_mask);
if((sign_T < 0.0f) ||
(sign_T > isect->t * xor_signmast(det, sign_mask)))
{
return false;
}
#ifdef __VISIBILITY_FLAG__
/* visibility flag test. we do it here under the assumption
* that most triangles are culled by node flags */
if(kernel_tex_fetch(__prim_visibility, triAddr) & visibility)
#endif
{
/* Normalize U, V, W, and T. */
const float inv_det = 1.0f / det;
isect->prim = triAddr;
isect->object = object;
isect->type = PRIMITIVE_TRIANGLE;
isect->u = U * inv_det;
isect->v = V * inv_det;
isect->t = T * inv_det;
return true;
}
return false;
}
/* Special ray intersection routines for subsurface scattering. In that case we
* only want to intersect with primitives in the same object, and if case of
* multiple hits we pick a single random primitive as the intersection point.
*/
#ifdef __SUBSURFACE__
ccl_device_inline void triangle_intersect_subsurface(
KernelGlobals *kg,
const IsectPrecalc *isect_precalc,
Intersection *isect_array,
float3 P,
int object,
int triAddr,
float tmax,
uint *num_hits,
uint *lcg_state,
int max_hits)
{
const int kx = isect_precalc->kx;
const int ky = isect_precalc->ky;
const int kz = isect_precalc->kz;
const float Sx = isect_precalc->Sx;
const float Sy = isect_precalc->Sy;
const float Sz = isect_precalc->Sz;
/* Calculate vertices relative to ray origin. */
const float4 tri_a = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+0),
tri_b = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+1),
tri_c = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+2);
const float3 A = make_float3(tri_a.x - P.x, tri_a.y - P.y, tri_a.z - P.z);
const float3 B = make_float3(tri_b.x - P.x, tri_b.y - P.y, tri_b.z - P.z);
const float3 C = make_float3(tri_c.x - P.x, tri_c.y - P.y, tri_c.z - P.z);
const float A_kx = IDX(A, kx), A_ky = IDX(A, ky), A_kz = IDX(A, kz);
const float B_kx = IDX(B, kx), B_ky = IDX(B, ky), B_kz = IDX(B, kz);
const float C_kx = IDX(C, kx), C_ky = IDX(C, ky), C_kz = IDX(C, kz);
/* Perform shear and scale of vertices. */
const float Ax = A_kx - Sx * A_kz;
const float Ay = A_ky - Sy * A_kz;
const float Bx = B_kx - Sx * B_kz;
const float By = B_ky - Sy * B_kz;
const float Cx = C_kx - Sx * C_kz;
const float Cy = C_ky - Sy * C_kz;
/* Calculate scaled barycentric coordinates. */
float U = Cx * By - Cy * Bx;
int sign_mask = (__float_as_int(U) & 0x80000000);
float V = Ax * Cy - Ay * Cx;
if(sign_mask != (__float_as_int(V) & 0x80000000)) {
return;
}
float W = Bx * Ay - By * Ax;
if(sign_mask != (__float_as_int(W) & 0x80000000)) {
return;
}
/* Calculate determinant. */
float det = U + V + W;
if(UNLIKELY(det == 0.0f)) {
return;
}
/* Calculate scaled z−coordinates of vertices and use them to calculate
* the hit distance.
*/
const float T = (U * A_kz + V * B_kz + W * C_kz) * Sz;
const float sign_T = xor_signmast(T, sign_mask);
if((sign_T < 0.0f) ||
(sign_T > tmax * xor_signmast(det, sign_mask)))
{
return;
}
/* Normalize U, V, W, and T. */
const float inv_det = 1.0f / det;
(*num_hits)++;
int hit;
if(*num_hits <= max_hits) {
hit = *num_hits - 1;
}
else {
/* reservoir sampling: if we are at the maximum number of
* hits, randomly replace element or skip it */
hit = lcg_step_uint(lcg_state) % *num_hits;
if(hit >= max_hits)
return;
}
/* record intersection */
Intersection *isect = &isect_array[hit];
isect->prim = triAddr;
isect->object = object;
isect->type = PRIMITIVE_TRIANGLE;
isect->u = U * inv_det;
isect->v = V * inv_det;
isect->t = T * inv_det;
}
#endif
/* Refine triangle intersection to more precise hit point. For rays that travel
* far the precision is often not so good, this reintersects the primitive from
* a closer distance. */
/* Reintersections uses the paper:
*
* Tomas Moeller
* Fast, minimum storage ray/triangle intersection
* http://www.cs.virginia.edu/~gfx/Courses/2003/ImageSynthesis/papers/Acceleration/Fast%20MinimumStorage%20RayTriangle%20Intersection.pdf
*/
ccl_device_inline float3 triangle_refine(KernelGlobals *kg,
ShaderData *sd,
const Intersection *isect,
const Ray *ray)
{
float3 P = ray->P;
float3 D = ray->D;
float t = isect->t;
#ifdef __INTERSECTION_REFINE__
if(isect->object != OBJECT_NONE) {
if(UNLIKELY(t == 0.0f)) {
return P;
}
#ifdef __OBJECT_MOTION__
Transform tfm = sd->ob_itfm;
#else
Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_INVERSE_TRANSFORM);
#endif
P = transform_point(&tfm, P);
D = transform_direction(&tfm, D*t);
D = normalize_len(D, &t);
}
P = P + D*t;
const float4 tri_a = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+0),
tri_b = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+1),
tri_c = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+2);
float3 edge1 = make_float3(tri_a.x - tri_c.x, tri_a.y - tri_c.y, tri_a.z - tri_c.z);
float3 edge2 = make_float3(tri_b.x - tri_c.x, tri_b.y - tri_c.y, tri_b.z - tri_c.z);
float3 tvec = make_float3(P.x - tri_c.x, P.y - tri_c.y, P.z - tri_c.z);
float3 qvec = cross(tvec, edge1);
float3 pvec = cross(D, edge2);
float rt = dot(edge2, qvec) / dot(edge1, pvec);
P = P + D*rt;
if(isect->object != OBJECT_NONE) {
#ifdef __OBJECT_MOTION__
Transform tfm = sd->ob_tfm;
#else
Transform tfm = object_fetch_transform(kg, isect->object, OBJECT_TRANSFORM);
#endif
P = transform_point(&tfm, P);
}
return P;
#else
return P + D*t;
#endif
}
/* Same as above, except that isect->t is assumed to be in object space for
* instancing.
*/
ccl_device_inline float3 triangle_refine_subsurface(KernelGlobals *kg,
ShaderData *sd,
const Intersection *isect,
const Ray *ray)
{
float3 P = ray->P;
float3 D = ray->D;
float t = isect->t;
#ifdef __INTERSECTION_REFINE__
if(isect->object != OBJECT_NONE) {
#ifdef __OBJECT_MOTION__
Transform tfm = sd->ob_itfm;
#else
Transform tfm = object_fetch_transform(kg,
isect->object,
OBJECT_INVERSE_TRANSFORM);
#endif
P = transform_point(&tfm, P);
D = transform_direction(&tfm, D);
D = normalize(D);
}
P = P + D*t;
const float4 tri_a = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+0),
tri_b = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+1),
tri_c = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+2);
float3 edge1 = make_float3(tri_a.x - tri_c.x, tri_a.y - tri_c.y, tri_a.z - tri_c.z);
float3 edge2 = make_float3(tri_b.x - tri_c.x, tri_b.y - tri_c.y, tri_b.z - tri_c.z);
float3 tvec = make_float3(P.x - tri_c.x, P.y - tri_c.y, P.z - tri_c.z);
float3 qvec = cross(tvec, edge1);
float3 pvec = cross(D, edge2);
float rt = dot(edge2, qvec) / dot(edge1, pvec);
P = P + D*rt;
if(isect->object != OBJECT_NONE) {
#ifdef __OBJECT_MOTION__
Transform tfm = sd->ob_tfm;
#else
Transform tfm = object_fetch_transform(kg,
isect->object,
OBJECT_TRANSFORM);
#endif
P = transform_point(&tfm, P);
}
return P;
#else
return P + D*t;
#endif
}
#undef IDX
CCL_NAMESPACE_END
|