xref: /haiku/src/apps/mediaplayer/support/StackBlurFilter.cpp (revision c8ccdf52032feea5e7853631dee57cf7f6302c5d)
1 /*
2  * Copyright 2008-2010 Stephan Aßmus <superstippi@gmx.de>
3  * Distributed under the terms of the MIT license.
4  */
5 //----------------------------------------------------------------------------
6 // Anti-Grain Geometry - Version 2.4
7 // Copyright (C) 2002-2005 Maxim Shemanarev (http://www.antigrain.com)
8 //
9 // Permission to copy, use, modify, sell and distribute this software
10 // is granted provided this copyright notice appears in all copies.
11 // This software is provided "as is" without express or implied
12 // warranty, and with no claim as to its suitability for any purpose.
13 //
14 //----------------------------------------------------------------------------
15 // Contact: mcseem@antigrain.com
16 //          mcseemagg@yahoo.com
17 //          http://www.antigrain.com
18 //----------------------------------------------------------------------------
19 //
20 // The Stack Blur Algorithm was invented by Mario Klingemann,
21 // mario@quasimondo.com and described here:
22 // http://incubator.quasimondo.com/processing/fast_blur_deluxe.php
23 // (search phrase "Stackblur: Fast But Goodlooking").
24 // The major improvement is that there's no more division table
25 // that was very expensive to create for large blur radii. Insted,
26 // for 8-bit per channel and radius not exceeding 254 the division is
27 // replaced by multiplication and shift.
28 //
29 //----------------------------------------------------------------------------
30 
31 
32 #include "StackBlurFilter.h"
33 
34 #include <stdio.h>
35 
36 #include <Bitmap.h>
37 
38 #include <agg_array.h>
39 
40 
41 template<class T> struct stack_blur_tables
42 {
43 	static uint16 const g_stack_blur8_mul[255];
44 	static uint8  const g_stack_blur8_shr[255];
45 };
46 
47 
48 template<class T>
49 uint16 const stack_blur_tables<T>::g_stack_blur8_mul[255] =
50 {
51 	512,512,456,512,328,456,335,512,405,328,271,456,388,335,292,512,
52 	454,405,364,328,298,271,496,456,420,388,360,335,312,292,273,512,
53 	482,454,428,405,383,364,345,328,312,298,284,271,259,496,475,456,
54 	437,420,404,388,374,360,347,335,323,312,302,292,282,273,265,512,
55 	497,482,468,454,441,428,417,405,394,383,373,364,354,345,337,328,
56 	320,312,305,298,291,284,278,271,265,259,507,496,485,475,465,456,
57 	446,437,428,420,412,404,396,388,381,374,367,360,354,347,341,335,
58 	329,323,318,312,307,302,297,292,287,282,278,273,269,265,261,512,
59 	505,497,489,482,475,468,461,454,447,441,435,428,422,417,411,405,
60 	399,394,389,383,378,373,368,364,359,354,350,345,341,337,332,328,
61 	324,320,316,312,309,305,301,298,294,291,287,284,281,278,274,271,
62 	268,265,262,259,257,507,501,496,491,485,480,475,470,465,460,456,
63 	451,446,442,437,433,428,424,420,416,412,408,404,400,396,392,388,
64 	385,381,377,374,370,367,363,360,357,354,350,347,344,341,338,335,
65 	332,329,326,323,320,318,315,312,310,307,304,302,299,297,294,292,
66 	289,287,285,282,280,278,275,273,271,269,267,265,263,261,259
67 };
68 
69 
70 template<class T>
71 uint8 const stack_blur_tables<T>::g_stack_blur8_shr[255] =
72 {
73 	  9, 11, 12, 13, 13, 14, 14, 15, 15, 15, 15, 16, 16, 16, 16, 17,
74 	 17, 17, 17, 17, 17, 17, 18, 18, 18, 18, 18, 18, 18, 18, 18, 19,
75 	 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 19, 20, 20, 20,
76 	 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 21,
77 	 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 21,
78 	 21, 21, 21, 21, 21, 21, 21, 21, 21, 21, 22, 22, 22, 22, 22, 22,
79 	 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22,
80 	 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 22, 23,
81 	 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23,
82 	 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23,
83 	 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23, 23,
84 	 23, 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
85 	 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
86 	 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
87 	 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24,
88 	 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24
89 };
90 
91 
92 #if __GNUC__ < 4
93 static inline float
roundf(float v)94 roundf(float v)
95 {
96 	return floorf(v + 0.5);
97 }
98 #endif
99 
100 
StackBlurFilter()101 StackBlurFilter::StackBlurFilter()
102 {
103 }
104 
105 
~StackBlurFilter()106 StackBlurFilter::~StackBlurFilter()
107 {
108 }
109 
110 
111 void
Filter(BBitmap * bitmap,double radius)112 StackBlurFilter::Filter(BBitmap* bitmap, double radius)
113 {
114 	if (radius < 1.0)
115 		return;
116 
117 	int32 width = bitmap->Bounds().IntegerWidth() + 1;
118 	int32 height = bitmap->Bounds().IntegerHeight() + 1;
119 	uint32 bpr = bitmap->BytesPerRow();
120 	uint8* bits = (uint8*)bitmap->Bits();
121 
122 	unsigned r = (unsigned)roundf(radius);
123 
124 	if (bitmap->ColorSpace() == B_RGBA32
125 		|| bitmap->ColorSpace() == B_RGB32) {
126 
127 		_Filter32(bits, width, height, bpr, r, r);
128 
129 	} else if (bitmap->ColorSpace() == B_GRAY8) {
130 
131 		_Filter8(bits, width, height, bpr, r, r);
132 
133 	} else
134 		printf("StackBlurFilter::Filter() - unsupported color space\n");
135 }
136 
137 
138 // #pragma mark -
139 
140 
141 struct rgba {
142 	uint8	r;
143 	uint8	g;
144 	uint8	b;
145 	uint8	a;
146 };
147 
148 
149 void
_Filter32(uint8 * buffer,unsigned width,unsigned height,int32 bpr,unsigned rx,unsigned ry) const150 StackBlurFilter::_Filter32(uint8* buffer, unsigned width, unsigned height,
151 	int32 bpr, unsigned rx, unsigned ry) const
152 {
153 	typedef rgba color_type;
154 
155 	unsigned x, y, xp, yp, i;
156 	unsigned stack_ptr;
157 	unsigned stack_start;
158 
159 	const uint8* src_pix_ptr;
160 		  uint8* dst_pix_ptr;
161 	color_type*  stack_pix_ptr;
162 
163 	unsigned sum_r;
164 	unsigned sum_g;
165 	unsigned sum_b;
166 	unsigned sum_a;
167 	unsigned sum_in_r;
168 	unsigned sum_in_g;
169 	unsigned sum_in_b;
170 	unsigned sum_in_a;
171 	unsigned sum_out_r;
172 	unsigned sum_out_g;
173 	unsigned sum_out_b;
174 	unsigned sum_out_a;
175 
176 	unsigned w   = width;
177 	unsigned h   = height;
178 	unsigned wm  = w - 1;
179 	unsigned hm  = h - 1;
180 
181 	unsigned div;
182 	unsigned mul_sum;
183 	unsigned shr_sum;
184 
185 	agg::pod_vector<color_type> stack;
186 
187 	if (rx > 0) {
188 		if (rx > 254)
189 			rx = 254;
190 		div = rx * 2 + 1;
191 		mul_sum = stack_blur_tables<int>::g_stack_blur8_mul[rx];
192 		shr_sum = stack_blur_tables<int>::g_stack_blur8_shr[rx];
193 		stack.allocate(div);
194 
195 		for (y = 0; y < h; y++) {
196 			sum_r =
197 			sum_g =
198 			sum_b =
199 			sum_a =
200 			sum_in_r =
201 			sum_in_g =
202 			sum_in_b =
203 			sum_in_a =
204 			sum_out_r =
205 			sum_out_g =
206 			sum_out_b =
207 			sum_out_a = 0;
208 
209 			src_pix_ptr = buffer + bpr * y;
210 			for (i = 0; i <= rx; i++) {
211 				stack_pix_ptr	= &stack[i];
212 				stack_pix_ptr->r = src_pix_ptr[2];
213 				stack_pix_ptr->g = src_pix_ptr[1];
214 				stack_pix_ptr->b = src_pix_ptr[0];
215 				stack_pix_ptr->a = src_pix_ptr[3];
216 				sum_r		   += src_pix_ptr[2] * (i + 1);
217 				sum_g		   += src_pix_ptr[1] * (i + 1);
218 				sum_b		   += src_pix_ptr[0] * (i + 1);
219 				sum_a		   += src_pix_ptr[3] * (i + 1);
220 				sum_out_r	   += src_pix_ptr[2];
221 				sum_out_g	   += src_pix_ptr[1];
222 				sum_out_b	   += src_pix_ptr[0];
223 				sum_out_a	   += src_pix_ptr[3];
224 			}
225 			for (i = 1; i <= rx; i++) {
226 				if (i <= wm)
227 					src_pix_ptr += 4;
228 				stack_pix_ptr = &stack[i + rx];
229 				stack_pix_ptr->r = src_pix_ptr[2];
230 				stack_pix_ptr->g = src_pix_ptr[1];
231 				stack_pix_ptr->b = src_pix_ptr[0];
232 				stack_pix_ptr->a = src_pix_ptr[3];
233 				sum_r		   += src_pix_ptr[2] * (rx + 1 - i);
234 				sum_g		   += src_pix_ptr[1] * (rx + 1 - i);
235 				sum_b		   += src_pix_ptr[0] * (rx + 1 - i);
236 				sum_a		   += src_pix_ptr[3] * (rx + 1 - i);
237 				sum_in_r		+= src_pix_ptr[2];
238 				sum_in_g		+= src_pix_ptr[1];
239 				sum_in_b		+= src_pix_ptr[0];
240 				sum_in_a		+= src_pix_ptr[3];
241 			}
242 
243 			stack_ptr = rx;
244 			xp = rx;
245 			if (xp > wm)
246 				xp = wm;
247 			src_pix_ptr = buffer + xp * 4 + y * bpr;
248 			dst_pix_ptr = buffer + y * bpr;
249 			for (x = 0; x < w; x++) {
250 				dst_pix_ptr[2] = (sum_r * mul_sum) >> shr_sum;
251 				dst_pix_ptr[1] = (sum_g * mul_sum) >> shr_sum;
252 				dst_pix_ptr[0] = (sum_b * mul_sum) >> shr_sum;
253 				dst_pix_ptr[3] = (sum_a * mul_sum) >> shr_sum;
254 				dst_pix_ptr += 4;
255 
256 				sum_r -= sum_out_r;
257 				sum_g -= sum_out_g;
258 				sum_b -= sum_out_b;
259 				sum_a -= sum_out_a;
260 
261 				stack_start = stack_ptr + div - rx;
262 				if(stack_start >= div) stack_start -= div;
263 				stack_pix_ptr = &stack[stack_start];
264 
265 				sum_out_r -= stack_pix_ptr->r;
266 				sum_out_g -= stack_pix_ptr->g;
267 				sum_out_b -= stack_pix_ptr->b;
268 				sum_out_a -= stack_pix_ptr->a;
269 
270 				if (xp < wm) {
271 					src_pix_ptr += 4;
272 					++xp;
273 				}
274 
275 				stack_pix_ptr->r = src_pix_ptr[2];
276 				stack_pix_ptr->g = src_pix_ptr[1];
277 				stack_pix_ptr->b = src_pix_ptr[0];
278 				stack_pix_ptr->a = src_pix_ptr[3];
279 
280 				sum_in_r += src_pix_ptr[2];
281 				sum_in_g += src_pix_ptr[1];
282 				sum_in_b += src_pix_ptr[0];
283 				sum_in_a += src_pix_ptr[3];
284 				sum_r	+= sum_in_r;
285 				sum_g	+= sum_in_g;
286 				sum_b	+= sum_in_b;
287 				sum_a	+= sum_in_a;
288 
289 				++stack_ptr;
290 				if (stack_ptr >= div)
291 					stack_ptr = 0;
292 				stack_pix_ptr = &stack[stack_ptr];
293 
294 				sum_out_r += stack_pix_ptr->r;
295 				sum_out_g += stack_pix_ptr->g;
296 				sum_out_b += stack_pix_ptr->b;
297 				sum_out_a += stack_pix_ptr->a;
298 				sum_in_r  -= stack_pix_ptr->r;
299 				sum_in_g  -= stack_pix_ptr->g;
300 				sum_in_b  -= stack_pix_ptr->b;
301 				sum_in_a  -= stack_pix_ptr->a;
302 			}
303 		}
304 	}
305 
306 	if (ry > 0) {
307 		if (ry > 254)
308 			ry = 254;
309 		div = ry * 2 + 1;
310 		mul_sum = stack_blur_tables<int>::g_stack_blur8_mul[ry];
311 		shr_sum = stack_blur_tables<int>::g_stack_blur8_shr[ry];
312 		stack.allocate(div);
313 
314 		int stride = bpr;
315 		for(x = 0; x < w; x++) {
316 			sum_r =
317 			sum_g =
318 			sum_b =
319 			sum_a =
320 			sum_in_r =
321 			sum_in_g =
322 			sum_in_b =
323 			sum_in_a =
324 			sum_out_r =
325 			sum_out_g =
326 			sum_out_b =
327 			sum_out_a = 0;
328 
329 			src_pix_ptr = buffer + x * 4;
330 			for (i = 0; i <= ry; i++) {
331 				stack_pix_ptr	= &stack[i];
332 				stack_pix_ptr->r = src_pix_ptr[2];
333 				stack_pix_ptr->g = src_pix_ptr[1];
334 				stack_pix_ptr->b = src_pix_ptr[0];
335 				stack_pix_ptr->a = src_pix_ptr[3];
336 				sum_r		   += src_pix_ptr[2] * (i + 1);
337 				sum_g		   += src_pix_ptr[1] * (i + 1);
338 				sum_b		   += src_pix_ptr[0] * (i + 1);
339 				sum_a		   += src_pix_ptr[3] * (i + 1);
340 				sum_out_r	   += src_pix_ptr[2];
341 				sum_out_g	   += src_pix_ptr[1];
342 				sum_out_b	   += src_pix_ptr[0];
343 				sum_out_a	   += src_pix_ptr[3];
344 			}
345 			for (i = 1; i <= ry; i++) {
346 				if (i <= hm)
347 					src_pix_ptr += stride;
348 				stack_pix_ptr = &stack[i + ry];
349 				stack_pix_ptr->r = src_pix_ptr[2];
350 				stack_pix_ptr->g = src_pix_ptr[1];
351 				stack_pix_ptr->b = src_pix_ptr[0];
352 				stack_pix_ptr->a = src_pix_ptr[3];
353 				sum_r		   += src_pix_ptr[2] * (ry + 1 - i);
354 				sum_g		   += src_pix_ptr[1] * (ry + 1 - i);
355 				sum_b		   += src_pix_ptr[0] * (ry + 1 - i);
356 				sum_a		   += src_pix_ptr[3] * (ry + 1 - i);
357 				sum_in_r		+= src_pix_ptr[2];
358 				sum_in_g		+= src_pix_ptr[1];
359 				sum_in_b		+= src_pix_ptr[0];
360 				sum_in_a		+= src_pix_ptr[3];
361 			}
362 
363 			stack_ptr = ry;
364 			yp = ry;
365 			if (yp > hm)
366 				yp = hm;
367 			src_pix_ptr = buffer + x * 4 + yp * bpr;
368 			dst_pix_ptr = buffer + x * 4;
369 			for (y = 0; y < h; y++) {
370 				dst_pix_ptr[2] = (sum_r * mul_sum) >> shr_sum;
371 				dst_pix_ptr[1] = (sum_g * mul_sum) >> shr_sum;
372 				dst_pix_ptr[0] = (sum_b * mul_sum) >> shr_sum;
373 				dst_pix_ptr[3] = (sum_a * mul_sum) >> shr_sum;
374 				dst_pix_ptr += stride;
375 
376 				sum_r -= sum_out_r;
377 				sum_g -= sum_out_g;
378 				sum_b -= sum_out_b;
379 				sum_a -= sum_out_a;
380 
381 				stack_start = stack_ptr + div - ry;
382 				if (stack_start >= div)
383 					stack_start -= div;
384 
385 				stack_pix_ptr = &stack[stack_start];
386 				sum_out_r -= stack_pix_ptr->r;
387 				sum_out_g -= stack_pix_ptr->g;
388 				sum_out_b -= stack_pix_ptr->b;
389 				sum_out_a -= stack_pix_ptr->a;
390 
391 				if (yp < hm) {
392 					src_pix_ptr += stride;
393 					++yp;
394 				}
395 
396 				stack_pix_ptr->r = src_pix_ptr[2];
397 				stack_pix_ptr->g = src_pix_ptr[1];
398 				stack_pix_ptr->b = src_pix_ptr[0];
399 				stack_pix_ptr->a = src_pix_ptr[3];
400 
401 				sum_in_r += src_pix_ptr[2];
402 				sum_in_g += src_pix_ptr[1];
403 				sum_in_b += src_pix_ptr[0];
404 				sum_in_a += src_pix_ptr[3];
405 				sum_r	+= sum_in_r;
406 				sum_g	+= sum_in_g;
407 				sum_b	+= sum_in_b;
408 				sum_a	+= sum_in_a;
409 
410 				++stack_ptr;
411 				if (stack_ptr >= div)
412 					stack_ptr = 0;
413 				stack_pix_ptr = &stack[stack_ptr];
414 
415 				sum_out_r += stack_pix_ptr->r;
416 				sum_out_g += stack_pix_ptr->g;
417 				sum_out_b += stack_pix_ptr->b;
418 				sum_out_a += stack_pix_ptr->a;
419 				sum_in_r  -= stack_pix_ptr->r;
420 				sum_in_g  -= stack_pix_ptr->g;
421 				sum_in_b  -= stack_pix_ptr->b;
422 				sum_in_a  -= stack_pix_ptr->a;
423 			}
424 		}
425 	}
426 }
427 
428 
429 void
_Filter8(uint8 * buffer,unsigned width,unsigned height,int32 bpr,unsigned rx,unsigned ry) const430 StackBlurFilter::_Filter8(uint8* buffer, unsigned width, unsigned height,
431 	int32 bpr, unsigned rx, unsigned ry) const
432 {
433 	unsigned x, y, xp, yp, i;
434 	unsigned stack_ptr;
435 	unsigned stack_start;
436 
437 	const uint8* src_pix_ptr;
438 		  uint8* dst_pix_ptr;
439 	unsigned pix;
440 	unsigned stack_pix;
441 	unsigned sum;
442 	unsigned sum_in;
443 	unsigned sum_out;
444 
445 	unsigned w   = width;
446 	unsigned h   = height;
447 	unsigned wm  = w - 1;
448 	unsigned hm  = h - 1;
449 
450 	unsigned div;
451 	unsigned mul_sum;
452 	unsigned shr_sum;
453 
454 	agg::pod_vector<uint8> stack;
455 
456 	if(rx > 0)
457 	{
458 		if(rx > 254) rx = 254;
459 		div = rx * 2 + 1;
460 		mul_sum = stack_blur_tables<int>::g_stack_blur8_mul[rx];
461 		shr_sum = stack_blur_tables<int>::g_stack_blur8_shr[rx];
462 		stack.allocate(div);
463 
464 		for(y = 0; y < h; y++)
465 		{
466 			sum = sum_in = sum_out = 0;
467 
468 			src_pix_ptr = buffer + y * bpr;
469 			pix = *src_pix_ptr;
470 			for(i = 0; i <= rx; i++)
471 			{
472 				stack[i] = pix;
473 				sum	 += pix * (i + 1);
474 				sum_out += pix;
475 			}
476 			for(i = 1; i <= rx; i++)
477 			{
478 				if(i <= wm) src_pix_ptr += 1;
479 				pix = *src_pix_ptr;
480 				stack[i + rx] = pix;
481 				sum	+= pix * (rx + 1 - i);
482 				sum_in += pix;
483 			}
484 
485 			stack_ptr = rx;
486 			xp = rx;
487 			if(xp > wm) xp = wm;
488 			src_pix_ptr = buffer + xp + y * bpr;
489 			dst_pix_ptr = buffer + y * bpr;
490 			for(x = 0; x < w; x++)
491 			{
492 				*dst_pix_ptr = (sum * mul_sum) >> shr_sum;
493 				dst_pix_ptr += 1;
494 
495 				sum -= sum_out;
496 
497 				stack_start = stack_ptr + div - rx;
498 				if(stack_start >= div) stack_start -= div;
499 				sum_out -= stack[stack_start];
500 
501 				if(xp < wm)
502 				{
503 					src_pix_ptr += 1;
504 					pix = *src_pix_ptr;
505 					++xp;
506 				}
507 
508 				stack[stack_start] = pix;
509 
510 				sum_in += pix;
511 				sum	+= sum_in;
512 
513 				++stack_ptr;
514 				if(stack_ptr >= div) stack_ptr = 0;
515 				stack_pix = stack[stack_ptr];
516 
517 				sum_out += stack_pix;
518 				sum_in  -= stack_pix;
519 			}
520 		}
521 	}
522 
523 	if(ry > 0)
524 	{
525 		if(ry > 254) ry = 254;
526 		div = ry * 2 + 1;
527 		mul_sum = stack_blur_tables<int>::g_stack_blur8_mul[ry];
528 		shr_sum = stack_blur_tables<int>::g_stack_blur8_shr[ry];
529 		stack.allocate(div);
530 
531 		int stride = bpr;
532 		for(x = 0; x < w; x++)
533 		{
534 			sum = sum_in = sum_out = 0;
535 
536 			src_pix_ptr = buffer + x;
537 			pix = *src_pix_ptr;
538 			for(i = 0; i <= ry; i++)
539 			{
540 				stack[i] = pix;
541 				sum	 += pix * (i + 1);
542 				sum_out += pix;
543 			}
544 			for(i = 1; i <= ry; i++)
545 			{
546 				if(i <= hm) src_pix_ptr += stride;
547 				pix = *src_pix_ptr;
548 				stack[i + ry] = pix;
549 				sum	+= pix * (ry + 1 - i);
550 				sum_in += pix;
551 			}
552 
553 			stack_ptr = ry;
554 			yp = ry;
555 			if(yp > hm) yp = hm;
556 			src_pix_ptr = buffer + x + yp * bpr;
557 			dst_pix_ptr = buffer + x;
558 			for(y = 0; y < h; y++)
559 			{
560 				*dst_pix_ptr = (sum * mul_sum) >> shr_sum;
561 				dst_pix_ptr += stride;
562 
563 				sum -= sum_out;
564 
565 				stack_start = stack_ptr + div - ry;
566 				if(stack_start >= div) stack_start -= div;
567 				sum_out -= stack[stack_start];
568 
569 				if(yp < hm)
570 				{
571 					src_pix_ptr += stride;
572 					pix = *src_pix_ptr;
573 					++yp;
574 				}
575 
576 				stack[stack_start] = pix;
577 
578 				sum_in += pix;
579 				sum	+= sum_in;
580 
581 				++stack_ptr;
582 				if(stack_ptr >= div) stack_ptr = 0;
583 				stack_pix = stack[stack_ptr];
584 
585 				sum_out += stack_pix;
586 				sum_in  -= stack_pix;
587 			}
588 		}
589 	}
590 }
591 
592