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
|
/* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* Copyright 2014, Blender Foundation.
*/
#include "MEM_guardedalloc.h"
#include "COM_SunBeamsOperation.h"
namespace blender::compositor {
SunBeamsOperation::SunBeamsOperation()
{
this->addInputSocket(DataType::Color);
this->addOutputSocket(DataType::Color);
this->setResolutionInputSocketIndex(0);
this->flags.complex = true;
}
void SunBeamsOperation::initExecution()
{
/* convert to pixels */
this->m_source_px[0] = this->m_data.source[0] * this->getWidth();
this->m_source_px[1] = this->m_data.source[1] * this->getHeight();
this->m_ray_length_px = this->m_data.ray_length * MAX2(this->getWidth(), this->getHeight());
}
/**
* Defines a line accumulator for a specific sector,
* given by the four matrix entries that rotate from buffer space into the sector
*
* (x,y) is used to designate buffer space coordinates
* (u,v) is used to designate sector space coordinates
*
* For a target point (x,y) the sector should be chosen such that
* ``u >= v >= 0``
* This removes the need to handle all sorts of special cases.
*
* Template parameters:
* fxu : buffer increment in x for sector u+1
* fxv : buffer increment in x for sector v+1
* fyu : buffer increment in y for sector u+1
* fyv : buffer increment in y for sector v+1
*/
template<int fxu, int fxv, int fyu, int fyv> struct BufferLineAccumulator {
/* utility functions implementing the matrix transform to/from sector space */
static inline void buffer_to_sector(const float source[2], int x, int y, int &u, int &v)
{
int x0 = (int)source[0];
int y0 = (int)source[1];
x -= x0;
y -= y0;
u = x * fxu + y * fyu;
v = x * fxv + y * fyv;
}
static inline void buffer_to_sector(const float source[2], float x, float y, float &u, float &v)
{
int x0 = (int)source[0];
int y0 = (int)source[1];
x -= (float)x0;
y -= (float)y0;
u = x * fxu + y * fyu;
v = x * fxv + y * fyv;
}
static inline void sector_to_buffer(const float source[2], int u, int v, int &x, int &y)
{
int x0 = (int)source[0];
int y0 = (int)source[1];
x = x0 + u * fxu + v * fxv;
y = y0 + u * fyu + v * fyv;
}
static inline void sector_to_buffer(const float source[2], float u, float v, float &x, float &y)
{
int x0 = (int)source[0];
int y0 = (int)source[1];
x = (float)x0 + u * fxu + v * fxv;
y = (float)y0 + u * fyu + v * fyv;
}
/**
* Set up the initial buffer pointer and calculate necessary variables for looping.
*
* Note that sector space is centered around the "source" point while the loop starts
* at dist_min from the target pt. This way the loop can be canceled as soon as it runs
* out of the buffer rect, because no pixels further along the line can contribute.
*
* \param x, y: Start location in the buffer
* \param num: Total steps in the loop
* \param v, dv: Vertical offset in sector space, for line offset perpendicular to the loop axis
*/
static float *init_buffer_iterator(MemoryBuffer *input,
const float source[2],
const float co[2],
float dist_min,
float dist_max,
int &x,
int &y,
int &num,
float &v,
float &dv,
float &falloff_factor)
{
float pu, pv;
buffer_to_sector(source, co[0], co[1], pu, pv);
/* line angle */
float tan_phi = pv / pu;
float dr = sqrtf(tan_phi * tan_phi + 1.0f);
float cos_phi = 1.0f / dr;
/* clamp u range to avoid influence of pixels "behind" the source */
float umin = max_ff(pu - cos_phi * dist_min, 0.0f);
float umax = max_ff(pu - cos_phi * dist_max, 0.0f);
v = umin * tan_phi;
dv = tan_phi;
int start = (int)floorf(umax);
int end = (int)ceilf(umin);
num = end - start;
sector_to_buffer(source, end, (int)ceilf(v), x, y);
falloff_factor = dist_max > dist_min ? dr / (float)(dist_max - dist_min) : 0.0f;
float *iter = input->getBuffer() + COM_NUM_CHANNELS_COLOR * (x + input->getWidth() * y);
return iter;
}
/**
* Perform the actual accumulation along a ray segment from source to pt.
* Only pixels within dist_min..dist_max contribute.
*
* The loop runs backwards(!) over the primary sector space axis u, i.e. increasing distance to
* pt. After each step it decrements v by dv < 1, adding a buffer shift when necessary.
*/
static void eval(MemoryBuffer *input,
float output[4],
const float co[2],
const float source[2],
float dist_min,
float dist_max)
{
const rcti &rect = input->get_rect();
int buffer_width = input->getWidth();
int x, y, num;
float v, dv;
float falloff_factor;
float border[4];
zero_v4(output);
if ((int)(co[0] - source[0]) == 0 && (int)(co[1] - source[1]) == 0) {
copy_v4_v4(output,
input->getBuffer() + COM_NUM_CHANNELS_COLOR *
((int)source[0] + input->getWidth() * (int)source[1]));
return;
}
/* Initialize the iteration variables. */
float *buffer = init_buffer_iterator(
input, source, co, dist_min, dist_max, x, y, num, v, dv, falloff_factor);
zero_v3(border);
border[3] = 1.0f;
/* v_local keeps track of when to decrement v (see below) */
float v_local = v - floorf(v);
for (int i = 0; i < num; i++) {
float weight = 1.0f - (float)i * falloff_factor;
weight *= weight;
/* range check, use last valid color when running beyond the image border */
if (x >= rect.xmin && x < rect.xmax && y >= rect.ymin && y < rect.ymax) {
madd_v4_v4fl(output, buffer, buffer[3] * weight);
/* use as border color in case subsequent pixels are out of bounds */
copy_v4_v4(border, buffer);
}
else {
madd_v4_v4fl(output, border, border[3] * weight);
}
/* TODO implement proper filtering here, see
* https://en.wikipedia.org/wiki/Lanczos_resampling
* https://en.wikipedia.org/wiki/Sinc_function
*
* using lanczos with x = distance from the line segment,
* normalized to a == 0.5f, could give a good result
*
* for now just divide equally at the end ...
*/
/* decrement u */
x -= fxu;
y -= fyu;
buffer -= (fxu + fyu * buffer_width) * COM_NUM_CHANNELS_COLOR;
/* decrement v (in steps of dv < 1) */
v_local -= dv;
if (v_local < 0.0f) {
v_local += 1.0f;
x -= fxv;
y -= fyv;
buffer -= (fxv + fyv * buffer_width) * COM_NUM_CHANNELS_COLOR;
}
}
/* normalize */
if (num > 0) {
mul_v4_fl(output, 1.0f / (float)num);
}
}
};
/**
* Dispatch function which selects an appropriate accumulator based on the sector of the target
* point, relative to the source.
*
* The BufferLineAccumulator defines the actual loop over the buffer, with an efficient inner loop
* due to using compile time constants instead of a local matrix variable defining the sector
* space.
*/
static void accumulate_line(MemoryBuffer *input,
float output[4],
const float co[2],
const float source[2],
float dist_min,
float dist_max)
{
/* coordinates relative to source */
float pt_ofs[2] = {co[0] - source[0], co[1] - source[1]};
/* The source sectors are defined like so:
*
* \ 3 | 2 /
* \ | /
* 4 \ | / 1
* \|/
* -----------
* /|\
* 5 / | \ 8
* / | \
* / 6 | 7 \
*
* The template arguments encode the transformation into "sector space",
* by means of rotation/mirroring matrix elements.
*/
if (fabsf(pt_ofs[1]) > fabsf(pt_ofs[0])) {
if (pt_ofs[0] > 0.0f) {
if (pt_ofs[1] > 0.0f) {
/* 2 */
BufferLineAccumulator<0, 1, 1, 0>::eval(input, output, co, source, dist_min, dist_max);
}
else {
/* 7 */
BufferLineAccumulator<0, 1, -1, 0>::eval(input, output, co, source, dist_min, dist_max);
}
}
else {
if (pt_ofs[1] > 0.0f) {
/* 3 */
BufferLineAccumulator<0, -1, 1, 0>::eval(input, output, co, source, dist_min, dist_max);
}
else {
/* 6 */
BufferLineAccumulator<0, -1, -1, 0>::eval(input, output, co, source, dist_min, dist_max);
}
}
}
else {
if (pt_ofs[0] > 0.0f) {
if (pt_ofs[1] > 0.0f) {
/* 1 */
BufferLineAccumulator<1, 0, 0, 1>::eval(input, output, co, source, dist_min, dist_max);
}
else {
/* 8 */
BufferLineAccumulator<1, 0, 0, -1>::eval(input, output, co, source, dist_min, dist_max);
}
}
else {
if (pt_ofs[1] > 0.0f) {
/* 4 */
BufferLineAccumulator<-1, 0, 0, 1>::eval(input, output, co, source, dist_min, dist_max);
}
else {
/* 5 */
BufferLineAccumulator<-1, 0, 0, -1>::eval(input, output, co, source, dist_min, dist_max);
}
}
}
}
void *SunBeamsOperation::initializeTileData(rcti * /*rect*/)
{
void *buffer = getInputOperation(0)->initializeTileData(nullptr);
return buffer;
}
void SunBeamsOperation::executePixel(float output[4], int x, int y, void *data)
{
const float co[2] = {(float)x, (float)y};
accumulate_line(
(MemoryBuffer *)data, output, co, this->m_source_px, 0.0f, this->m_ray_length_px);
}
static void calc_ray_shift(rcti *rect, float x, float y, const float source[2], float ray_length)
{
float co[2] = {(float)x, (float)y};
float dir[2], dist;
/* move (x,y) vector toward the source by ray_length distance */
sub_v2_v2v2(dir, co, source);
dist = normalize_v2(dir);
mul_v2_fl(dir, min_ff(dist, ray_length));
sub_v2_v2(co, dir);
int ico[2] = {(int)co[0], (int)co[1]};
BLI_rcti_do_minmax_v(rect, ico);
}
bool SunBeamsOperation::determineDependingAreaOfInterest(rcti *input,
ReadBufferOperation *readOperation,
rcti *output)
{
/* Enlarges the rect by moving each corner toward the source.
* This is the maximum distance that pixels can influence each other
* and gives a rect that contains all possible accumulated pixels.
*/
rcti rect = *input;
calc_ray_shift(&rect, input->xmin, input->ymin, this->m_source_px, this->m_ray_length_px);
calc_ray_shift(&rect, input->xmin, input->ymax, this->m_source_px, this->m_ray_length_px);
calc_ray_shift(&rect, input->xmax, input->ymin, this->m_source_px, this->m_ray_length_px);
calc_ray_shift(&rect, input->xmax, input->ymax, this->m_source_px, this->m_ray_length_px);
return NodeOperation::determineDependingAreaOfInterest(&rect, readOperation, output);
}
} // namespace blender::compositor
|