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
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
|
/*
* Adapted from code Copyright 2009-2010 NVIDIA Corporation
* Modifications Copyright 2011, 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.
*/
CCL_NAMESPACE_BEGIN
/*
* "Persistent while-while kernel" used in:
*
* "Understanding the Efficiency of Ray Traversal on GPUs",
* Timo Aila and Samuli Laine,
* Proc. High-Performance Graphics 2009
*/
/* bottom-most stack entry, indicating the end of traversal */
#define ENTRYPOINT_SENTINEL 0x76543210
/* 64 object BVH + 64 mesh BVH + 64 object node splitting */
#define BVH_STACK_SIZE 192
#define BVH_NODE_SIZE 4
#define TRI_NODE_SIZE 3
/* silly workaround for float extended precision that happens when compiling
* without sse support on x86, it results in different results for float ops
* that you would otherwise expect to compare correctly */
#if !defined(__i386__) || defined(__SSE__)
#define NO_EXTENDED_PRECISION
#else
#define NO_EXTENDED_PRECISION volatile
#endif
__device_inline float3 bvh_inverse_direction(float3 dir)
{
/* avoid divide by zero (ooeps = exp2f(-80.0f)) */
float ooeps = 0.00000000000000000000000082718061255302767487140869206996285356581211090087890625f;
float3 idir;
idir.x = 1.0f/((fabsf(dir.x) > ooeps)? dir.x: copysignf(ooeps, dir.x));
idir.y = 1.0f/((fabsf(dir.y) > ooeps)? dir.y: copysignf(ooeps, dir.y));
idir.z = 1.0f/((fabsf(dir.z) > ooeps)? dir.z: copysignf(ooeps, dir.z));
return idir;
}
__device_inline void bvh_instance_push(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *idir, float *t, const float tmax)
{
Transform tfm = object_fetch_transform(kg, object, OBJECT_INVERSE_TRANSFORM);
*P = transform_point(&tfm, ray->P);
float3 dir = transform_direction(&tfm, ray->D);
float len;
dir = normalize_len(dir, &len);
*idir = bvh_inverse_direction(dir);
if(*t != FLT_MAX)
*t *= len;
}
__device_inline void bvh_instance_pop(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *idir, float *t, const float tmax)
{
if(*t != FLT_MAX) {
Transform tfm = object_fetch_transform(kg, object, OBJECT_TRANSFORM);
*t *= len(transform_direction(&tfm, 1.0f/(*idir)));
}
*P = ray->P;
*idir = bvh_inverse_direction(ray->D);
}
#ifdef __OBJECT_MOTION__
__device_inline void bvh_instance_motion_push(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *idir, float *t, Transform *tfm, const float tmax)
{
Transform itfm;
*tfm = object_fetch_transform_motion_test(kg, object, ray->time, &itfm);
*P = transform_point(&itfm, ray->P);
float3 dir = transform_direction(&itfm, ray->D);
float len;
dir = normalize_len(dir, &len);
*idir = bvh_inverse_direction(dir);
if(*t != FLT_MAX)
*t *= len;
}
__device_inline void bvh_instance_motion_pop(KernelGlobals *kg, int object, const Ray *ray, float3 *P, float3 *idir, float *t, Transform *tfm, const float tmax)
{
if(*t != FLT_MAX)
*t *= len(transform_direction(tfm, 1.0f/(*idir)));
*P = ray->P;
*idir = bvh_inverse_direction(ray->D);
}
#endif
/* intersect two bounding boxes */
#ifdef __HAIR__
__device_inline void bvh_node_intersect(KernelGlobals *kg,
bool *traverseChild0, bool *traverseChild1,
bool *closestChild1, int *nodeAddr0, int *nodeAddr1,
float3 P, float3 idir, float t, uint visibility, int nodeAddr, float difl, float extmax)
{
float hdiff = 1.0f + difl;
float ldiff = 1.0f - difl;
#else
__device_inline void bvh_node_intersect(KernelGlobals *kg,
bool *traverseChild0, bool *traverseChild1,
bool *closestChild1, int *nodeAddr0, int *nodeAddr1,
float3 P, float3 idir, float t, uint visibility, int nodeAddr)
{
#endif
/* fetch node data */
float4 n0xy = kernel_tex_fetch(__bvh_nodes, nodeAddr*BVH_NODE_SIZE+0);
float4 n1xy = kernel_tex_fetch(__bvh_nodes, nodeAddr*BVH_NODE_SIZE+1);
float4 nz = kernel_tex_fetch(__bvh_nodes, nodeAddr*BVH_NODE_SIZE+2);
float4 cnodes = kernel_tex_fetch(__bvh_nodes, nodeAddr*BVH_NODE_SIZE+3);
/* intersect ray against child nodes */
float3 ood = P * idir;
NO_EXTENDED_PRECISION float c0lox = n0xy.x * idir.x - ood.x;
NO_EXTENDED_PRECISION float c0hix = n0xy.y * idir.x - ood.x;
NO_EXTENDED_PRECISION float c0loy = n0xy.z * idir.y - ood.y;
NO_EXTENDED_PRECISION float c0hiy = n0xy.w * idir.y - ood.y;
NO_EXTENDED_PRECISION float c0loz = nz.x * idir.z - ood.z;
NO_EXTENDED_PRECISION float c0hiz = nz.y * idir.z - ood.z;
NO_EXTENDED_PRECISION float c0min = max4(min(c0lox, c0hix), min(c0loy, c0hiy), min(c0loz, c0hiz), 0.0f);
NO_EXTENDED_PRECISION float c0max = min4(max(c0lox, c0hix), max(c0loy, c0hiy), max(c0loz, c0hiz), t);
NO_EXTENDED_PRECISION float c1loz = nz.z * idir.z - ood.z;
NO_EXTENDED_PRECISION float c1hiz = nz.w * idir.z - ood.z;
NO_EXTENDED_PRECISION float c1lox = n1xy.x * idir.x - ood.x;
NO_EXTENDED_PRECISION float c1hix = n1xy.y * idir.x - ood.x;
NO_EXTENDED_PRECISION float c1loy = n1xy.z * idir.y - ood.y;
NO_EXTENDED_PRECISION float c1hiy = n1xy.w * idir.y - ood.y;
NO_EXTENDED_PRECISION float c1min = max4(min(c1lox, c1hix), min(c1loy, c1hiy), min(c1loz, c1hiz), 0.0f);
NO_EXTENDED_PRECISION float c1max = min4(max(c1lox, c1hix), max(c1loy, c1hiy), max(c1loz, c1hiz), t);
#ifdef __HAIR__
if(difl != 0.0f) {
if(__float_as_int(cnodes.z) & PATH_RAY_CURVE) {
c0min = max(ldiff * c0min, c0min - extmax);
c0max = min(hdiff * c0max, c0max + extmax);
}
if(__float_as_int(cnodes.z) & PATH_RAY_CURVE) {
c1min = max(ldiff * c1min, c1min - extmax);
c1max = min(hdiff * c1max, c1max + extmax);
}
}
#endif
/* decide which nodes to traverse next */
#ifdef __VISIBILITY_FLAG__
/* this visibility test gives a 5% performance hit, how to solve? */
*traverseChild0 = (c0max >= c0min) && (__float_as_int(cnodes.z) & visibility);
*traverseChild1 = (c1max >= c1min) && (__float_as_int(cnodes.w) & visibility);
#else
*traverseChild0 = (c0max >= c0min);
*traverseChild1 = (c1max >= c1min);
#endif
*nodeAddr0 = __float_as_int(cnodes.x);
*nodeAddr1 = __float_as_int(cnodes.y);
*closestChild1 = (c1min < c0min);
}
/* Sven Woop's algorithm */
__device_inline void bvh_triangle_intersect(KernelGlobals *kg, Intersection *isect,
float3 P, float3 idir, uint visibility, int object, int triAddr)
{
/* compute and check intersection t-value */
float4 v00 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+0);
float4 v11 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+1);
float3 dir = 1.0f/idir;
float Oz = v00.w - P.x*v00.x - P.y*v00.y - P.z*v00.z;
float invDz = 1.0f/(dir.x*v00.x + dir.y*v00.y + dir.z*v00.z);
float t = Oz * invDz;
if(t > 0.0f && t < isect->t) {
/* compute and check barycentric u */
float Ox = v11.w + P.x*v11.x + P.y*v11.y + P.z*v11.z;
float Dx = dir.x*v11.x + dir.y*v11.y + dir.z*v11.z;
float u = Ox + t*Dx;
if(u >= 0.0f) {
/* compute and check barycentric v */
float4 v22 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+2);
float Oy = v22.w + P.x*v22.x + P.y*v22.y + P.z*v22.z;
float Dy = dir.x*v22.x + dir.y*v22.y + dir.z*v22.z;
float v = Oy + t*Dy;
if(v >= 0.0f && u + v <= 1.0f) {
#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
{
/* record intersection */
isect->prim = triAddr;
isect->object = object;
isect->u = u;
isect->v = v;
isect->t = t;
}
}
}
}
}
#ifdef __HAIR__
__device_inline void curvebounds(float *lower, float *upper, float *extremta, float *extrema, float *extremtb, float *extremb, float p0, float p1, float p2, float p3)
{
float halfdiscroot = (p2 * p2 - 3 * p3 * p1);
float ta = -1.0f;
float tb = -1.0f;
*extremta = -1.0f;
*extremtb = -1.0f;
*upper = p0;
*lower = p0 + p1 + p2 + p3;
*extrema = *upper;
*extremb = *lower;
if(*lower >= *upper) {
*upper = *lower;
*lower = p0;
}
if(halfdiscroot >= 0) {
halfdiscroot = sqrt(halfdiscroot);
ta = (-p2 - halfdiscroot) / (3 * p3);
tb = (-p2 + halfdiscroot) / (3 * p3);
}
float t2;
float t3;
if(ta > 0.0f && ta < 1.0f) {
t2 = ta * ta;
t3 = t2 * ta;
*extremta = ta;
*extrema = p3 * t3 + p2 * t2 + p1 * ta + p0;
if(*extrema > *upper) {
*upper = *extrema;
}
if(*extrema < *lower) {
*lower = *extrema;
}
}
if(tb > 0.0f && tb < 1.0f) {
t2 = tb * tb;
t3 = t2 * tb;
*extremtb = tb;
*extremb = p3 * t3 + p2 * t2 + p1 * tb + p0;
if(*extremb >= *upper) {
*upper = *extremb;
}
if(*extremb <= *lower) {
*lower = *extremb;
}
}
}
__device_inline void bvh_cardinal_curve_intersect(KernelGlobals *kg, Intersection *isect,
float3 P, float3 idir, uint visibility, int object, int curveAddr, int segment, uint *lcg_state, float difl, float extmax)
{
float epsilon = 0.0f;
int depth = kernel_data.curve_kernel_data.subdivisions;
/* curve Intersection check */
float3 dir = 1.0f/idir;
int flags = kernel_data.curve_kernel_data.curveflags;
int prim = kernel_tex_fetch(__prim_index, curveAddr);
float3 curve_coef[4];
float r_st,r_en;
/*obtain curve parameters*/
{
/*ray transform created - this shold be created at beginning of intersection loop*/
Transform htfm;
float d = sqrtf(dir.x * dir.x + dir.z * dir.z);
htfm = make_transform(
dir.z / d, 0, -dir.x /d, 0,
-dir.x * dir.y /d, d, -dir.y * dir.z /d, 0,
dir.x, dir.y, dir.z, 0,
0, 0, 0, 1);
float4 v00 = kernel_tex_fetch(__curves, prim);
int k0 = __float_as_int(v00.x) + segment;
int k1 = k0 + 1;
int ka = max(k0 - 1,__float_as_int(v00.x));
int kb = min(k1 + 1,__float_as_int(v00.x) + __float_as_int(v00.y) - 1);
float4 P0 = kernel_tex_fetch(__curve_keys, ka);
float4 P1 = kernel_tex_fetch(__curve_keys, k0);
float4 P2 = kernel_tex_fetch(__curve_keys, k1);
float4 P3 = kernel_tex_fetch(__curve_keys, kb);
float3 p0 = transform_point(&htfm, float4_to_float3(P0) - P);
float3 p1 = transform_point(&htfm, float4_to_float3(P1) - P);
float3 p2 = transform_point(&htfm, float4_to_float3(P2) - P);
float3 p3 = transform_point(&htfm, float4_to_float3(P3) - P);
float fc = 0.71f;
curve_coef[0] = p1;
curve_coef[1] = -fc*p0 + fc*p2;
curve_coef[2] = 2.0f * fc * p0 + (fc - 3.0f) * p1 + (3.0f - 2.0f * fc) * p2 - fc * p3;
curve_coef[3] = -fc * p0 + (2.0f - fc) * p1 + (fc - 2.0f) * p2 + fc * p3;
r_st = P1.w;
r_en = P2.w;
}
float r_curr = max(r_st, r_en);
if((flags & CURVE_KN_RIBBONS) || !(flags & CURVE_KN_BACKFACING))
epsilon = 2 * r_curr;
/*find bounds - this is slow for cubic curves*/
float upper,lower;
float zextrem[4];
curvebounds(&lower, &upper, &zextrem[0], &zextrem[1], &zextrem[2], &zextrem[3], curve_coef[0].z, curve_coef[1].z, curve_coef[2].z, curve_coef[3].z);
if(lower - r_curr > isect->t || upper + r_curr < epsilon)
return;
/*minimum width extension*/
float mw_extension = min(difl * fabsf(upper), extmax);
float r_ext = mw_extension + r_curr;
float xextrem[4];
curvebounds(&lower, &upper, &xextrem[0], &xextrem[1], &xextrem[2], &xextrem[3], curve_coef[0].x, curve_coef[1].x, curve_coef[2].x, curve_coef[3].x);
if(lower > r_ext || upper < -r_ext)
return;
float yextrem[4];
curvebounds(&lower, &upper, &yextrem[0], &yextrem[1], &yextrem[2], &yextrem[3], curve_coef[0].y, curve_coef[1].y, curve_coef[2].y, curve_coef[3].y);
if(lower > r_ext || upper < -r_ext)
return;
/*setup recurrent loop*/
int level = 1 << depth;
int tree = 0;
float resol = 1.0f / (float)level;
/*begin loop*/
while(!(tree >> (depth))) {
float i_st = tree * resol;
float i_en = i_st + (level * resol);
float3 p_st = ((curve_coef[3] * i_st + curve_coef[2]) * i_st + curve_coef[1]) * i_st + curve_coef[0];
float3 p_en = ((curve_coef[3] * i_en + curve_coef[2]) * i_en + curve_coef[1]) * i_en + curve_coef[0];
float bminx = min(p_st.x, p_en.x);
float bmaxx = max(p_st.x, p_en.x);
float bminy = min(p_st.y, p_en.y);
float bmaxy = max(p_st.y, p_en.y);
float bminz = min(p_st.z, p_en.z);
float bmaxz = max(p_st.z, p_en.z);
if(xextrem[0] >= i_st && xextrem[0] <= i_en) {
bminx = min(bminx,xextrem[1]);
bmaxx = max(bmaxx,xextrem[1]);
}
if(xextrem[2] >= i_st && xextrem[2] <= i_en) {
bminx = min(bminx,xextrem[3]);
bmaxx = max(bmaxx,xextrem[3]);
}
if(yextrem[0] >= i_st && yextrem[0] <= i_en) {
bminy = min(bminy,yextrem[1]);
bmaxy = max(bmaxy,yextrem[1]);
}
if(yextrem[2] >= i_st && yextrem[2] <= i_en) {
bminy = min(bminy,yextrem[3]);
bmaxy = max(bmaxy,yextrem[3]);
}
if(zextrem[0] >= i_st && zextrem[0] <= i_en) {
bminz = min(bminz,zextrem[1]);
bmaxz = max(bmaxz,zextrem[1]);
}
if(zextrem[2] >= i_st && zextrem[2] <= i_en) {
bminz = min(bminz,zextrem[3]);
bmaxz = max(bmaxz,zextrem[3]);
}
float r1 = r_st + (r_en - r_st) * i_st;
float r2 = r_st + (r_en - r_st) * i_en;
r_curr = max(r1, r2);
mw_extension = min(difl * fabsf(bmaxz), extmax);
float r_ext = mw_extension + r_curr;
float coverage = 1.0f;
if (bminz - r_curr > isect->t || bmaxz + r_curr < epsilon || bminx > r_ext|| bmaxx < -r_ext|| bminy > r_ext|| bmaxy < -r_ext) {
/* the bounding box does not overlap the square centered at O.*/
tree += level;
level = tree & -tree;
}
else if (level == 1) {
/* the maximum recursion depth is reached.
* check if dP0.(Q-P0)>=0 and dPn.(Pn-Q)>=0.
* dP* is reversed if necessary.*/
float t = isect->t;
float u = 0.0f;
if(flags & CURVE_KN_RIBBONS) {
float3 tg = (p_en - p_st);
float w = tg.x * tg.x + tg.y * tg.y;
if (w == 0) {
tree++;
level = tree & -tree;
continue;
}
w = -(p_st.x * tg.x + p_st.y * tg.y) / w;
w = clamp((float)w, 0.0f, 1.0f);
/* compute u on the curve segment.*/
u = i_st * (1 - w) + i_en * w;
r_curr = r_st + (r_en - r_st) * u;
/* compare x-y distances.*/
float3 p_curr = ((curve_coef[3] * u + curve_coef[2]) * u + curve_coef[1]) * u + curve_coef[0];
float3 dp_st = (3 * curve_coef[3] * i_st + 2 * curve_coef[2]) * i_st + curve_coef[1];
if (dot(tg, dp_st)< 0)
dp_st *= -1;
if (dot(dp_st, -p_st) + p_curr.z * dp_st.z < 0) {
tree++;
level = tree & -tree;
continue;
}
float3 dp_en = (3 * curve_coef[3] * i_en + 2 * curve_coef[2]) * i_en + curve_coef[1];
if (dot(tg, dp_en) < 0)
dp_en *= -1;
if (dot(dp_en, p_en) - p_curr.z * dp_en.z < 0) {
tree++;
level = tree & -tree;
continue;
}
/* compute coverage */
float r_ext = r_curr;
coverage = 1.0f;
if(difl != 0.0f) {
mw_extension = min(difl * fabsf(bmaxz), extmax);
r_ext = mw_extension + r_curr;
float d = sqrtf(p_curr.x * p_curr.x + p_curr.y * p_curr.y);
float d0 = d - r_curr;
float d1 = d + r_curr;
if (d0 >= 0)
coverage = (min(d1 / mw_extension, 1.0f) - min(d0 / mw_extension, 1.0f)) * 0.5;
else // inside
coverage = (min(d1 / mw_extension, 1.0f) + min(-d0 / mw_extension, 1.0f)) * 0.5;
}
if (p_curr.x * p_curr.x + p_curr.y * p_curr.y >= r_ext * r_ext || p_curr.z <= epsilon) {
tree++;
level = tree & -tree;
continue;
}
/* compare z distances.*/
if (isect->t < p_curr.z) {
tree++;
level = tree & -tree;
continue;
}
t = p_curr.z;
}
else {
float l = len(p_en - p_st);
/*minimum width extension*/
float or1 = r1;
float or2 = r2;
if(difl != 0.0f) {
mw_extension = min(len(p_st - P) * difl, extmax);
or1 = r1 < mw_extension ? mw_extension : r1;
mw_extension = min(len(p_en - P) * difl, extmax);
or2 = r2 < mw_extension ? mw_extension : r2;
}
/* --- */
float3 tg = (p_en - p_st) / l;
float gd = (or2 - or1) / l;
float difz = -dot(p_st,tg);
float cyla = 1.0f - (tg.z * tg.z * (1 + gd*gd));
float halfb = (-p_st.z - tg.z*(difz + gd*(difz*gd + or1)));
float tcentre = -halfb/cyla;
float zcentre = difz + (tg.z * tcentre);
float3 tdif = - p_st;
tdif.z += tcentre;
float tdifz = dot(tdif,tg);
float tb = 2*(tdif.z - tg.z*(tdifz + gd*(tdifz*gd + or1)));
float tc = dot(tdif,tdif) - tdifz * tdifz * (1 + gd*gd) - or1*or1 - 2*or1*tdifz*gd;
float td = tb*tb - 4*cyla*tc;
if (td < 0.0f){
tree++;
level = tree & -tree;
continue;
}
float rootd = sqrtf(td);
float correction = ((-tb - rootd)/(2*cyla));
t = tcentre + correction;
float w = (zcentre + (tg.z * correction))/l;
float3 dp_st = (3 * curve_coef[3] * i_st + 2 * curve_coef[2]) * i_st + curve_coef[1];
if (dot(tg, dp_st)< 0)
dp_st *= -1;
float3 dp_en = (3 * curve_coef[3] * i_en + 2 * curve_coef[2]) * i_en + curve_coef[1];
if (dot(tg, dp_en) < 0)
dp_en *= -1;
if(flags & CURVE_KN_BACKFACING && (dot(dp_st, -p_st) + t * dp_st.z < 0 || dot(dp_en, p_en) - t * dp_en.z < 0 || isect->t < t || t <= 0.0f)) {
correction = ((-tb + rootd)/(2*cyla));
t = tcentre + correction;
w = (zcentre + (tg.z * correction))/l;
}
if (dot(dp_st, -p_st) + t * dp_st.z < 0 || dot(dp_en, p_en) - t * dp_en.z < 0 || isect->t < t || t <= 0.0f) {
tree++;
level = tree & -tree;
continue;
}
w = clamp((float)w, 0.0f, 1.0f);
/* compute u on the curve segment.*/
u = i_st * (1 - w) + i_en * w;
r_curr = r1 + (r2 - r1) * w;
r_ext = or1 + (or2 - or1) * w;
coverage = r_curr/r_ext;
}
/* we found a new intersection.*/
/*stochastic fade from minimum width*/
if(lcg_state && coverage != 1.0f) {
if(lcg_step(lcg_state) > coverage)
return;
}
#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, curveAddr) & visibility)
#endif
{
/* record intersection */
isect->prim = curveAddr;
isect->segment = segment;
isect->object = object;
isect->u = u;
isect->v = 0.0f;
/*isect->v = 1.0f - coverage; */
isect->t = t;
}
tree++;
level = tree & -tree;
}
else {
/* split the curve into two curves and process */
level = level >> 1;
}
}
}
__device_inline void bvh_curve_intersect(KernelGlobals *kg, Intersection *isect,
float3 P, float3 idir, uint visibility, int object, int curveAddr, int segment, uint *lcg_state, float difl, float extmax)
{
/* curve Intersection check */
int flags = kernel_data.curve_kernel_data.curveflags;
int prim = kernel_tex_fetch(__prim_index, curveAddr);
float4 v00 = kernel_tex_fetch(__curves, prim);
int cnum = __float_as_int(v00.x);
int k0 = cnum + segment;
int k1 = k0 + 1;
float4 P1 = kernel_tex_fetch(__curve_keys, k0);
float4 P2 = kernel_tex_fetch(__curve_keys, k1);
float or1 = P1.w;
float or2 = P2.w;
float3 p1 = float4_to_float3(P1);
float3 p2 = float4_to_float3(P2);
/*minimum width extension*/
float r1 = or1;
float r2 = or2;
if(difl != 0.0f) {
float pixelsize = min(len(p1 - P) * difl, extmax);
r1 = or1 < pixelsize ? pixelsize : or1;
pixelsize = min(len(p2 - P) * difl, extmax);
r2 = or2 < pixelsize ? pixelsize : or2;
}
/* --- */
float mr = max(r1,r2);
float3 dif = P - p1;
float3 dir = 1.0f/idir;
float l = len(p2 - p1);
float sp_r = mr + 0.5f * l;
float3 sphere_dif = P - ((p1 + p2) * 0.5f);
float sphere_b = dot(dir,sphere_dif);
sphere_dif = sphere_dif - sphere_b * dir;
sphere_b = dot(dir,sphere_dif);
float sdisc = sphere_b * sphere_b - len_squared(sphere_dif) + sp_r * sp_r;
if(sdisc < 0.0f)
return;
/* obtain parameters and test midpoint distance for suitable modes*/
float3 tg = (p2 - p1) / l;
float gd = (r2 - r1) / l;
float dirz = dot(dir,tg);
float difz = dot(dif,tg);
float a = 1.0f - (dirz*dirz*(1 + gd*gd));
float halfb = dot(dir,dif) - dirz*(difz + gd*(difz*gd + r1));
float tcentre = -halfb/a;
float zcentre = difz + (dirz * tcentre);
if((tcentre > isect->t) && !(flags & CURVE_KN_ACCURATE))
return;
if((zcentre < 0 || zcentre > l) && !(flags & CURVE_KN_ACCURATE) && !(flags & CURVE_KN_INTERSECTCORRECTION))
return;
/* test minimum separation*/
float3 cprod = cross(tg, dir);
float3 cprod2 = cross(tg, dif);
float cprodsq = len_squared(cprod);
float cprod2sq = len_squared(cprod2);
float distscaled = dot(cprod,dif);
if(cprodsq == 0)
distscaled = cprod2sq;
else
distscaled = (distscaled*distscaled)/cprodsq;
if(distscaled > mr*mr)
return;
/* calculate true intersection*/
float3 tdif = P - p1 + tcentre * dir;
float tdifz = dot(tdif,tg);
float tb = 2*(dot(dir,tdif) - dirz*(tdifz + gd*(tdifz*gd + r1)));
float tc = dot(tdif,tdif) - tdifz * tdifz * (1 + gd*gd) - r1*r1 - 2*r1*tdifz*gd;
float td = tb*tb - 4*a*tc;
if (td < 0.0f)
return;
float rootd = 0.0f;
float correction = 0.0f;
if(flags & CURVE_KN_ACCURATE) {
rootd = sqrtf(td);
correction = ((-tb - rootd)/(2*a));
}
float t = tcentre + correction;
if(t < isect->t) {
if(flags & CURVE_KN_INTERSECTCORRECTION) {
rootd = sqrtf(td);
correction = ((-tb - rootd)/(2*a));
t = tcentre + correction;
}
float z = zcentre + (dirz * correction);
bool backface = false;
if(flags & CURVE_KN_BACKFACING && (t < 0.0f || z < 0 || z > l)) {
backface = true;
correction = ((-tb + rootd)/(2*a));
t = tcentre + correction;
z = zcentre + (dirz * correction);
}
/*stochastic fade from minimum width*/
float adjradius = or1 + z * (or2 - or1) / l;
adjradius = adjradius / (r1 + z * gd);
if(lcg_state && adjradius != 1.0f) {
if(lcg_step(lcg_state) > adjradius)
return;
}
/* --- */
if(t > 0.0f && t < isect->t && z >= 0 && z <= l) {
if (flags & CURVE_KN_ENCLOSEFILTER) {
float enc_ratio = kernel_data.curve_kernel_data.encasing_ratio;
if((dot(P - p1, tg) > -r1 * enc_ratio) && (dot(P - p2, tg) < r2 * enc_ratio)) {
float a2 = 1.0f - (dirz*dirz*(1 + gd*gd*enc_ratio*enc_ratio));
float c2 = dot(dif,dif) - difz * difz * (1 + gd*gd*enc_ratio*enc_ratio) - r1*r1*enc_ratio*enc_ratio - 2*r1*difz*gd*enc_ratio;
if(a2*c2 < 0.0f)
return;
}
}
#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, curveAddr) & visibility)
#endif
{
/* record intersection */
isect->prim = curveAddr;
isect->segment = segment;
isect->object = object;
isect->u = z/l;
isect->v = td/(4*a*a);
/*isect->v = 1.0f - adjradius;*/
isect->t = t;
if(backface)
isect->u = -isect->u;
}
}
}
}
#endif
#ifdef __SUBSURFACE__
/* 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. */
__device_inline void bvh_triangle_intersect_subsurface(KernelGlobals *kg, Intersection *isect,
float3 P, float3 idir, int object, int triAddr, float tmax, int *num_hits, float subsurface_random)
{
/* compute and check intersection t-value */
float4 v00 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+0);
float4 v11 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+1);
float3 dir = 1.0f/idir;
float Oz = v00.w - P.x*v00.x - P.y*v00.y - P.z*v00.z;
float invDz = 1.0f/(dir.x*v00.x + dir.y*v00.y + dir.z*v00.z);
float t = Oz * invDz;
if(t > 0.0f && t < tmax) {
/* compute and check barycentric u */
float Ox = v11.w + P.x*v11.x + P.y*v11.y + P.z*v11.z;
float Dx = dir.x*v11.x + dir.y*v11.y + dir.z*v11.z;
float u = Ox + t*Dx;
if(u >= 0.0f) {
/* compute and check barycentric v */
float4 v22 = kernel_tex_fetch(__tri_woop, triAddr*TRI_NODE_SIZE+2);
float Oy = v22.w + P.x*v22.x + P.y*v22.y + P.z*v22.z;
float Dy = dir.x*v22.x + dir.y*v22.y + dir.z*v22.z;
float v = Oy + t*Dy;
if(v >= 0.0f && u + v <= 1.0f) {
(*num_hits)++;
if(subsurface_random * (*num_hits) <= 1.0f) {
/* record intersection */
isect->prim = triAddr;
isect->object = object;
isect->u = u;
isect->v = v;
isect->t = t;
}
}
}
}
}
#endif
/* BVH intersection function variations */
#define BVH_INSTANCING 1
#define BVH_MOTION 2
#define BVH_HAIR 4
#define BVH_HAIR_MINIMUM_WIDTH 8
#define BVH_SUBSURFACE 16
#define BVH_FUNCTION_NAME bvh_intersect
#define BVH_FUNCTION_FEATURES 0
#include "kernel_bvh_traversal.h"
#if defined(__INSTANCING__)
#define BVH_FUNCTION_NAME bvh_intersect_instancing
#define BVH_FUNCTION_FEATURES BVH_INSTANCING
#include "kernel_bvh_traversal.h"
#endif
#if defined(__HAIR__)
#define BVH_FUNCTION_NAME bvh_intersect_hair
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH
#include "kernel_bvh_traversal.h"
#endif
#if defined(__OBJECT_MOTION__)
#define BVH_FUNCTION_NAME bvh_intersect_motion
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_MOTION
#include "kernel_bvh_traversal.h"
#endif
#if defined(__HAIR__) && defined(__OBJECT_MOTION__)
#define BVH_FUNCTION_NAME bvh_intersect_hair_motion
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH|BVH_MOTION
#include "kernel_bvh_traversal.h"
#endif
#if defined(__SUBSURFACE__)
#define BVH_FUNCTION_NAME bvh_intersect_subsurface
#define BVH_FUNCTION_FEATURES BVH_SUBSURFACE
#include "kernel_bvh_traversal.h"
#endif
#if defined(__SUBSURFACE__) && defined(__INSTANCING__)
#define BVH_FUNCTION_NAME bvh_intersect_subsurface_instancing
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_SUBSURFACE
#include "kernel_bvh_traversal.h"
#endif
#if defined(__SUBSURFACE__) && defined(__HAIR__)
#define BVH_FUNCTION_NAME bvh_intersect_subsurface_hair
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_SUBSURFACE|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH
#include "kernel_bvh_traversal.h"
#endif
#if defined(__SUBSURFACE__) && defined(__OBJECT_MOTION__)
#define BVH_FUNCTION_NAME bvh_intersect_subsurface_motion
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_SUBSURFACE|BVH_MOTION
#include "kernel_bvh_traversal.h"
#endif
#if defined(__SUBSURFACE__) && defined(__HAIR__) && defined(__OBJECT_MOTION__)
#define BVH_FUNCTION_NAME bvh_intersect_subsurface_hair_motion
#define BVH_FUNCTION_FEATURES BVH_INSTANCING|BVH_SUBSURFACE|BVH_HAIR|BVH_HAIR_MINIMUM_WIDTH|BVH_MOTION
#include "kernel_bvh_traversal.h"
#endif
#ifdef __HAIR__
__device_inline bool scene_intersect(KernelGlobals *kg, const Ray *ray, const uint visibility, Intersection *isect, uint *lcg_state, float difl, float extmax)
#else
__device_inline bool scene_intersect(KernelGlobals *kg, const Ray *ray, const uint visibility, Intersection *isect)
#endif
{
#ifdef __OBJECT_MOTION__
if(kernel_data.bvh.have_motion) {
#ifdef __HAIR__
if(kernel_data.bvh.have_curves)
return bvh_intersect_hair_motion(kg, ray, isect, visibility, lcg_state, difl, extmax);
#endif /* __HAIR__ */
return bvh_intersect_motion(kg, ray, isect, visibility);
}
#endif /* __OBJECT_MOTION__ */
#ifdef __HAIR__
if(kernel_data.bvh.have_curves)
return bvh_intersect_hair(kg, ray, isect, visibility, lcg_state, difl, extmax);
#endif /* __HAIR__ */
#ifdef __KERNEL_CPU__
#ifdef __INSTANCING__
if(kernel_data.bvh.have_instancing)
return bvh_intersect_instancing(kg, ray, isect, visibility);
#endif /* __INSTANCING__ */
return bvh_intersect(kg, ray, isect, visibility);
#else /* __KERNEL_CPU__ */
#ifdef __INSTANCING__
return bvh_intersect_instancing(kg, ray, isect, visibility);
#else
return bvh_intersect(kg, ray, isect, visibility);
#endif /* __INSTANCING__ */
#endif /* __KERNEL_CPU__ */
}
#ifdef __SUBSURFACE__
__device_inline int scene_intersect_subsurface(KernelGlobals *kg, const Ray *ray, Intersection *isect, int subsurface_object, float subsurface_random)
{
#ifdef __OBJECT_MOTION__
if(kernel_data.bvh.have_motion) {
#ifdef __HAIR__
if(kernel_data.bvh.have_curves)
return bvh_intersect_subsurface_hair_motion(kg, ray, isect, subsurface_object, subsurface_random);
#endif /* __HAIR__ */
return bvh_intersect_subsurface_motion(kg, ray, isect, subsurface_object, subsurface_random);
}
#endif /* __OBJECT_MOTION__ */
#ifdef __HAIR__
if(kernel_data.bvh.have_curves)
return bvh_intersect_subsurface_hair(kg, ray, isect, subsurface_object, subsurface_random);
#endif /* __HAIR__ */
#ifdef __KERNEL_CPU__
#ifdef __INSTANCING__
if(kernel_data.bvh.have_instancing)
return bvh_intersect_subsurface_instancing(kg, ray, isect, subsurface_object, subsurface_random);
#endif /* __INSTANCING__ */
return bvh_intersect_subsurface(kg, ray, isect, subsurface_object, subsurface_random);
#else /* __KERNEL_CPU__ */
#ifdef __INSTANCING__
return bvh_intersect_subsurface_instancing(kg, ray, isect, subsurface_object, subsurface_random);
#else
return bvh_intersect_subsurface(kg, ray, isect, subsurface_object, subsurface_random);
#endif /* __INSTANCING__ */
#endif /* __KERNEL_CPU__ */
}
#endif
/* Ray offset to avoid self intersection */
__device_inline float3 ray_offset(float3 P, float3 Ng)
{
#ifdef __INTERSECTION_REFINE__
const float epsilon_f = 1e-5f;
/* ideally this should match epsilon_f, but instancing/mblur
* precision makes it problematic */
const float epsilon_test = 1.0f;
const int epsilon_i = 32;
float3 res;
/* x component */
if(fabsf(P.x) < epsilon_test) {
res.x = P.x + Ng.x*epsilon_f;
}
else {
uint ix = __float_as_uint(P.x);
ix += ((ix ^ __float_as_uint(Ng.x)) >> 31)? -epsilon_i: epsilon_i;
res.x = __uint_as_float(ix);
}
/* y component */
if(fabsf(P.y) < epsilon_test) {
res.y = P.y + Ng.y*epsilon_f;
}
else {
uint iy = __float_as_uint(P.y);
iy += ((iy ^ __float_as_uint(Ng.y)) >> 31)? -epsilon_i: epsilon_i;
res.y = __uint_as_float(iy);
}
/* z component */
if(fabsf(P.z) < epsilon_test) {
res.z = P.z + Ng.z*epsilon_f;
}
else {
uint iz = __float_as_uint(P.z);
iz += ((iz ^ __float_as_uint(Ng.z)) >> 31)? -epsilon_i: epsilon_i;
res.z = __uint_as_float(iz);
}
return res;
#else
const float epsilon_f = 1e-4f;
return P + epsilon_f*Ng;
#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. */
__device_inline float3 bvh_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 != ~0) {
#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;
float4 v00 = kernel_tex_fetch(__tri_woop, isect->prim*TRI_NODE_SIZE+0);
float Oz = v00.w - P.x*v00.x - P.y*v00.y - P.z*v00.z;
float invDz = 1.0f/(D.x*v00.x + D.y*v00.y + D.z*v00.z);
float rt = Oz * invDz;
P = P + D*rt;
if(isect->object != ~0) {
#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
}
#ifdef __HAIR__
__device_inline float3 curvetangent(float t, float3 p0, float3 p1, float3 p2, float3 p3)
{
float fc = 0.71f;
float data[4];
float t2 = t * t;
data[0] = -3.0f * fc * t2 + 4.0f * fc * t - fc;
data[1] = 3.0f * (2.0f - fc) * t2 + 2.0f * (fc - 3.0f) * t;
data[2] = 3.0f * (fc - 2.0f) * t2 + 2.0f * (3.0f - 2.0f * fc) * t + fc;
data[3] = 3.0f * fc * t2 - 2.0f * fc * t;
return data[0] * p0 + data[1] * p1 + data[2] * p2 + data[3] * p3;
}
__device_inline float3 curvepoint(float t, float3 p0, float3 p1, float3 p2, float3 p3)
{
float data[4];
float fc = 0.71f;
float t2 = t * t;
float t3 = t2 * t;
data[0] = -fc * t3 + 2.0f * fc * t2 - fc * t;
data[1] = (2.0f - fc) * t3 + (fc - 3.0f) * t2 + 1.0f;
data[2] = (fc - 2.0f) * t3 + (3.0f - 2.0f * fc) * t2 + fc * t;
data[3] = fc * t3 - fc * t2;
return data[0] * p0 + data[1] * p1 + data[2] * p2 + data[3] * p3;
}
__device_inline float3 bvh_curve_refine(KernelGlobals *kg, ShaderData *sd, const Intersection *isect, const Ray *ray, float t)
{
int flag = kernel_data.curve_kernel_data.curveflags;
float3 P = ray->P;
float3 D = ray->D;
if(isect->object != ~0) {
#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);
}
int prim = kernel_tex_fetch(__prim_index, isect->prim);
float4 v00 = kernel_tex_fetch(__curves, prim);
int k0 = __float_as_int(v00.x) + isect->segment;
int k1 = k0 + 1;
float4 P1 = kernel_tex_fetch(__curve_keys, k0);
float4 P2 = kernel_tex_fetch(__curve_keys, k1);
float l = 1.0f;
float3 tg = normalize_len(float4_to_float3(P2 - P1),&l);
float r1 = P1.w;
float r2 = P2.w;
float gd = ((r2 - r1)/l);
P = P + D*t;
if(flag & CURVE_KN_INTERPOLATE) {
int ka = max(k0 - 1,__float_as_int(v00.x));
int kb = min(k1 + 1,__float_as_int(v00.x) + __float_as_int(v00.y) - 1);
float4 P0 = kernel_tex_fetch(__curve_keys, ka);
float4 P3 = kernel_tex_fetch(__curve_keys, kb);
float3 p[4];
p[0] = float4_to_float3(P0);
p[1] = float4_to_float3(P1);
p[2] = float4_to_float3(P2);
p[3] = float4_to_float3(P3);
tg = normalize(curvetangent(isect->u,p[0],p[1],p[2],p[3]));
float3 p_curr = curvepoint(isect->u,p[0],p[1],p[2],p[3]);
#ifdef __UV__
sd->u = isect->u;
sd->v = 0.0f;
#endif
if(kernel_data.curve_kernel_data.curveflags & CURVE_KN_RIBBONS)
sd->Ng = normalize(-(D - tg * (dot(tg,D))));
else {
sd->Ng = normalize(P - p_curr);
sd->Ng = sd->Ng - gd * tg;
sd->Ng = normalize(sd->Ng);
}
sd->N = sd->Ng;
}
else {
float3 dif = P - float4_to_float3(P1);
#ifdef __UV__
sd->u = dot(dif,tg)/l;
sd->v = 0.0f;
#endif
if (flag & CURVE_KN_TRUETANGENTGNORMAL) {
sd->Ng = -(D - tg * (dot(tg,D) * kernel_data.curve_kernel_data.normalmix));
sd->Ng = normalize(sd->Ng);
if (flag & CURVE_KN_NORMALCORRECTION) {
sd->Ng = sd->Ng - gd * tg;
sd->Ng = normalize(sd->Ng);
}
}
else {
sd->Ng = (dif - tg * sd->u * l) / (P1.w + sd->u * l * gd);
if (gd != 0.0f) {
sd->Ng = sd->Ng - gd * tg ;
sd->Ng = normalize(sd->Ng);
}
}
sd->N = sd->Ng;
if (flag & CURVE_KN_TANGENTGNORMAL && !(flag & CURVE_KN_TRUETANGENTGNORMAL)) {
sd->N = -(D - tg * (dot(tg,D) * kernel_data.curve_kernel_data.normalmix));
sd->N = normalize(sd->N);
if (flag & CURVE_KN_NORMALCORRECTION) {
sd->N = sd->N - gd * tg;
sd->N = normalize(sd->N);
}
}
if (!(flag & CURVE_KN_TANGENTGNORMAL) && flag & CURVE_KN_TRUETANGENTGNORMAL) {
sd->N = (dif - tg * sd->u * l) / (P1.w + sd->u * l * gd);
if (gd != 0.0f) {
sd->N = sd->N - gd * tg ;
sd->N = normalize(sd->N);
}
}
}
#ifdef __DPDU__
/* dPdu/dPdv */
sd->dPdu = tg;
sd->dPdv = cross(tg,sd->Ng);
#endif
/*add fading parameter for minimum pixel width with transparency bsdf*/
/*sd->curve_transparency = isect->v;*/
/*sd->curve_radius = sd->u * gd * l + r1;*/
if(isect->object != ~0) {
#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;
}
#endif
CCL_NAMESPACE_END
|