summaryrefslogtreecommitdiffstats
path: root/postproc/postprocess.c
blob: 1ccea03dfb52ddaf8f6fe4b285db3624ebe672ef (plain)
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
/*
    Copyright (C) 2001-2002 Michael Niedermayer (michaelni@gmx.at)

    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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
*/

/*
			C	MMX	MMX2	3DNow
isVertDC		Ec	Ec
isVertMinMaxOk		Ec	Ec
doVertLowPass		E		e	e
doVertDefFilter		Ec	Ec	e	e
isHorizDC		Ec	Ec
isHorizMinMaxOk		a	E
doHorizLowPass		E		e	e
doHorizDefFilter	Ec	Ec	e	e
deRing			E		e	e*
Vertical RKAlgo1	E		a	a
Horizontal RKAlgo1			a	a
Vertical X1#		a		E	E
Horizontal X1#		a		E	E
LinIpolDeinterlace	e		E	E*
CubicIpolDeinterlace	a		e	e*
LinBlendDeinterlace	e		E	E*
MedianDeinterlace#	 	Ec	Ec
TempDeNoiser#		E		e	e

* i dont have a 3dnow CPU -> its untested, but noone said it doesnt work so it seems to work
# more or less selfinvented filters so the exactness isnt too meaningfull
E = Exact implementation
e = allmost exact implementation (slightly different rounding,...)
a = alternative / approximate impl
c = checked against the other implementations (-vo md5)
*/

/*
TODO:
reduce the time wasted on the mem transfer
unroll stuff if instructions depend too much on the prior one
move YScale thing to the end instead of fixing QP
write a faster and higher quality deblocking filter :)
make the mainloop more flexible (variable number of blocks at once
	(the if/else stuff per block is slowing things down)
compare the quality & speed of all filters
split this huge file
optimize c versions
try to unroll inner for(x=0 ... loop to avoid these damn if(x ... checks
...
*/

//Changelog: use the CVS log

#include "../config.h"
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#ifdef HAVE_MALLOC_H
#include <malloc.h>
#endif
//#undef HAVE_MMX2
//#define HAVE_3DNOW
//#undef HAVE_MMX
//#undef ARCH_X86
//#define DEBUG_BRIGHTNESS
#ifndef PIC
#include "../libvo/fastmemcpy.h"
#endif
#include "postprocess.h"
#include "postprocess_internal.h"
#include "../mangle.h"

#define MIN(a,b) ((a) > (b) ? (b) : (a))
#define MAX(a,b) ((a) < (b) ? (b) : (a))
#define ABS(a) ((a) > 0 ? (a) : (-(a)))
#define SIGN(a) ((a) > 0 ? 1 : -1)

#define GET_MODE_BUFFER_SIZE 500
#define OPTIONS_ARRAY_SIZE 10
#define BLOCK_SIZE 8
#define TEMP_STRIDE 8
//#define NUM_BLOCKS_AT_ONCE 16 //not used yet

#ifdef ARCH_X86
static uint64_t __attribute__((aligned(8))) w05=		0x0005000500050005LL;
static uint64_t __attribute__((aligned(8))) w20=		0x0020002000200020LL;
static uint64_t __attribute__((aligned(8))) b00= 		0x0000000000000000LL;
static uint64_t __attribute__((aligned(8))) b01= 		0x0101010101010101LL;
static uint64_t __attribute__((aligned(8))) b02= 		0x0202020202020202LL;
static uint64_t __attribute__((aligned(8))) b08= 		0x0808080808080808LL;
static uint64_t __attribute__((aligned(8))) b80= 		0x8080808080808080LL;
#endif

static int verbose= 0;

static const int deringThreshold= 20;


static struct PPFilter filters[]=
{
	{"hb", "hdeblock", 		1, 1, 3, H_DEBLOCK},
	{"vb", "vdeblock", 		1, 2, 4, V_DEBLOCK},
/*	{"hr", "rkhdeblock", 		1, 1, 3, H_RK1_FILTER},
	{"vr", "rkvdeblock", 		1, 2, 4, V_RK1_FILTER},*/
	{"h1", "x1hdeblock", 		1, 1, 3, H_X1_FILTER},
	{"v1", "x1vdeblock", 		1, 2, 4, V_X1_FILTER},
	{"dr", "dering", 		1, 5, 6, DERING},
	{"al", "autolevels", 		0, 1, 2, LEVEL_FIX},
	{"lb", "linblenddeint", 	1, 1, 4, LINEAR_BLEND_DEINT_FILTER},
	{"li", "linipoldeint", 		1, 1, 4, LINEAR_IPOL_DEINT_FILTER},
	{"ci", "cubicipoldeint",	1, 1, 4, CUBIC_IPOL_DEINT_FILTER},
	{"md", "mediandeint", 		1, 1, 4, MEDIAN_DEINT_FILTER},
	{"fd", "ffmpegdeint", 		1, 1, 4, FFMPEG_DEINT_FILTER},
	{"tn", "tmpnoise", 		1, 7, 8, TEMP_NOISE_FILTER},
	{"fq", "forcequant", 		1, 0, 0, FORCE_QUANT},
	{NULL, NULL,0,0,0,0} //End Marker
};

static char *replaceTable[]=
{
	"default", 	"hdeblock:a,vdeblock:a,dering:a,autolevels,tmpnoise:a:150:200:400",
	"de", 		"hdeblock:a,vdeblock:a,dering:a,autolevels,tmpnoise:a:150:200:400",
	"fast", 	"x1hdeblock:a,x1vdeblock:a,dering:a,autolevels,tmpnoise:a:150:200:400",
	"fa", 		"x1hdeblock:a,x1vdeblock:a,dering:a,autolevels,tmpnoise:a:150:200:400",
	NULL //End Marker
};

#ifdef ARCH_X86
static inline void unusedVariableWarningFixer()
{
	if(w05 + w20 + b00 + b01 + b02 + b08 + b80 == 0) b00=0;
}
#endif


#ifdef ARCH_X86
static inline void prefetchnta(void *p)
{
	asm volatile(	"prefetchnta (%0)\n\t"
		: : "r" (p)
	);
}

static inline void prefetcht0(void *p)
{
	asm volatile(	"prefetcht0 (%0)\n\t"
		: : "r" (p)
	);
}

static inline void prefetcht1(void *p)
{
	asm volatile(	"prefetcht1 (%0)\n\t"
		: : "r" (p)
	);
}

static inline void prefetcht2(void *p)
{
	asm volatile(	"prefetcht2 (%0)\n\t"
		: : "r" (p)
	);
}
#endif

// The horizontal Functions exist only in C cuz the MMX code is faster with vertical filters and transposing

/**
 * Check if the given 8x8 Block is mostly "flat"
 */
static inline int isHorizDC(uint8_t src[], int stride, PPContext *c)
{
	int numEq= 0;
	int y;
	const int dcOffset= ((c->QP*c->ppMode.baseDcDiff)>>8) + 1;
	const int dcThreshold= dcOffset*2 + 1;
	for(y=0; y<BLOCK_SIZE; y++)
	{
		if(((unsigned)(src[0] - src[1] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[1] - src[2] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[2] - src[3] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[3] - src[4] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[4] - src[5] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[5] - src[6] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[6] - src[7] + dcOffset)) < dcThreshold) numEq++;
		src+= stride;
	}
	return numEq > c->ppMode.flatnessThreshold;
}

/**
 * Check if the middle 8x8 Block in the given 8x16 block is flat
 */
static inline int isVertDC_C(uint8_t src[], int stride, PPContext *c){
	int numEq= 0;
	int y;
	const int dcOffset= ((c->QP*c->ppMode.baseDcDiff)>>8) + 1;
	const int dcThreshold= dcOffset*2 + 1;
	src+= stride*4; // src points to begin of the 8x8 Block
	for(y=0; y<BLOCK_SIZE-1; y++)
	{
		if(((unsigned)(src[0] - src[0+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[1] - src[1+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[2] - src[2+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[3] - src[3+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[4] - src[4+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[5] - src[5+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[6] - src[6+stride] + dcOffset)) < dcThreshold) numEq++;
		if(((unsigned)(src[7] - src[7+stride] + dcOffset)) < dcThreshold) numEq++;
		src+= stride;
	}
	return numEq > c->ppMode.flatnessThreshold;
}

static inline int isHorizMinMaxOk(uint8_t src[], int stride, int QP)
{
	if(abs(src[0] - src[7]) > 2*QP) return 0;

	return 1;
}

static inline void doHorizDefFilter(uint8_t dst[], int stride, int QP)
{
	int y;
	for(y=0; y<BLOCK_SIZE; y++)
	{
		const int middleEnergy= 5*(dst[4] - dst[5]) + 2*(dst[2] - dst[5]);

		if(ABS(middleEnergy) < 8*QP)
		{
			const int q=(dst[3] - dst[4])/2;
			const int leftEnergy=  5*(dst[2] - dst[1]) + 2*(dst[0] - dst[3]);
			const int rightEnergy= 5*(dst[6] - dst[5]) + 2*(dst[4] - dst[7]);

			int d= ABS(middleEnergy) - MIN( ABS(leftEnergy), ABS(rightEnergy) );
			d= MAX(d, 0);

			d= (5*d + 32) >> 6;
			d*= SIGN(-middleEnergy);

			if(q>0)
			{
				d= d<0 ? 0 : d;
				d= d>q ? q : d;
			}
			else
			{
				d= d>0 ? 0 : d;
				d= d<q ? q : d;
			}

        		dst[3]-= d;
	        	dst[4]+= d;
		}
		dst+= stride;
	}
}

/**
 * Do a horizontal low pass filter on the 10x8 block (dst points to middle 8x8 Block)
 * using the 9-Tap Filter (1,1,2,2,4,2,2,1,1)/16 (C version)
 */
static inline void doHorizLowPass(uint8_t dst[], int stride, int QP)
{

	int y;
	for(y=0; y<BLOCK_SIZE; y++)
	{
		const int first= ABS(dst[-1] - dst[0]) < QP ? dst[-1] : dst[0];
		const int last= ABS(dst[8] - dst[7]) < QP ? dst[8] : dst[7];

		int sums[9];
		sums[0] = first + dst[0];
		sums[1] = dst[0] + dst[1];
		sums[2] = dst[1] + dst[2];
		sums[3] = dst[2] + dst[3];
		sums[4] = dst[3] + dst[4];
		sums[5] = dst[4] + dst[5];
		sums[6] = dst[5] + dst[6];
		sums[7] = dst[6] + dst[7];
		sums[8] = dst[7] + last;

		dst[0]= ((sums[0]<<2) + ((first + sums[2])<<1) + sums[4] + 8)>>4;
		dst[1]= ((dst[1]<<2) + ((first + sums[0] + sums[3])<<1) + sums[5] + 8)>>4;
		dst[2]= ((dst[2]<<2) + ((first + sums[1] + sums[4])<<1) + sums[6] + 8)>>4;
		dst[3]= ((dst[3]<<2) + ((sums[2] + sums[5])<<1) + sums[0] + sums[7] + 8)>>4;
		dst[4]= ((dst[4]<<2) + ((sums[3] + sums[6])<<1) + sums[1] + sums[8] + 8)>>4;
		dst[5]= ((dst[5]<<2) + ((last + sums[7] + sums[4])<<1) + sums[2] + 8)>>4;
		dst[6]= (((last + dst[6])<<2) + ((dst[7] + sums[5])<<1) + sums[3] + 8)>>4;
		dst[7]= ((sums[8]<<2) + ((last + sums[6])<<1) + sums[4] + 8)>>4;

		dst+= stride;
	}
}

/**
 * Experimental Filter 1 (Horizontal)
 * will not damage linear gradients
 * Flat blocks should look like they where passed through the (1,1,2,2,4,2,2,1,1) 9-Tap filter
 * can only smooth blocks at the expected locations (it cant smooth them if they did move)
 * MMX2 version does correct clipping C version doesnt
 * not identical with the vertical one
 */
static inline void horizX1Filter(uint8_t *src, int stride, int QP)
{
	int y;
	static uint64_t *lut= NULL;
	if(lut==NULL)
	{
		int i;
		lut= (uint64_t*)memalign(8, 256*8);
		for(i=0; i<256; i++)
		{
			int v= i < 128 ? 2*i : 2*(i-256);
/*
//Simulate 112242211 9-Tap filter
			uint64_t a= (v/16) & 0xFF;
			uint64_t b= (v/8) & 0xFF;
			uint64_t c= (v/4) & 0xFF;
			uint64_t d= (3*v/8) & 0xFF;
*/
//Simulate piecewise linear interpolation
			uint64_t a= (v/16) & 0xFF;
			uint64_t b= (v*3/16) & 0xFF;
			uint64_t c= (v*5/16) & 0xFF;
			uint64_t d= (7*v/16) & 0xFF;
			uint64_t A= (0x100 - a)&0xFF;
			uint64_t B= (0x100 - b)&0xFF;
			uint64_t C= (0x100 - c)&0xFF;
			uint64_t D= (0x100 - c)&0xFF;

			lut[i]   = (a<<56) | (b<<48) | (c<<40) | (d<<32) |
				(D<<24) | (C<<16) | (B<<8) | (A);
			//lut[i] = (v<<32) | (v<<24);
		}
	}

	for(y=0; y<BLOCK_SIZE; y++)
	{
		int a= src[1] - src[2];
		int b= src[3] - src[4];
		int c= src[5] - src[6];

		int d= MAX(ABS(b) - (ABS(a) + ABS(c))/2, 0);

		if(d < QP)
		{
			int v = d * SIGN(-b);

			src[1] +=v/8;
			src[2] +=v/4;
			src[3] +=3*v/8;
			src[4] -=3*v/8;
			src[5] -=v/4;
			src[6] -=v/8;

		}
		src+=stride;
	}
}


//Note: we have C, MMX, MMX2, 3DNOW version there is no 3DNOW+MMX2 one
//Plain C versions
#if !defined (HAVE_MMX) || defined (RUNTIME_CPUDETECT)
#define COMPILE_C
#endif

#ifdef ARCH_X86

#if (defined (HAVE_MMX) && !defined (HAVE_3DNOW) && !defined (HAVE_MMX2)) || defined (RUNTIME_CPUDETECT)
#define COMPILE_MMX
#endif

#if defined (HAVE_MMX2) || defined (RUNTIME_CPUDETECT)
#define COMPILE_MMX2
#endif

#if (defined (HAVE_3DNOW) && !defined (HAVE_MMX2)) || defined (RUNTIME_CPUDETECT)
#define COMPILE_3DNOW
#endif
#endif //ARCH_X86

#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#undef ARCH_X86

#ifdef COMPILE_C
#undef HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#undef ARCH_X86
#define RENAME(a) a ## _C
#include "postprocess_template.c"
#endif

//MMX versions
#ifdef COMPILE_MMX
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#undef HAVE_3DNOW
#define ARCH_X86
#define RENAME(a) a ## _MMX
#include "postprocess_template.c"
#endif

//MMX2 versions
#ifdef COMPILE_MMX2
#undef RENAME
#define HAVE_MMX
#define HAVE_MMX2
#undef HAVE_3DNOW
#define ARCH_X86
#define RENAME(a) a ## _MMX2
#include "postprocess_template.c"
#endif

//3DNOW versions
#ifdef COMPILE_3DNOW
#undef RENAME
#define HAVE_MMX
#undef HAVE_MMX2
#define HAVE_3DNOW
#define ARCH_X86
#define RENAME(a) a ## _3DNow
#include "postprocess_template.c"
#endif

// minor note: the HAVE_xyz is messed up after that line so dont use it

static inline void postProcess(uint8_t src[], int srcStride, uint8_t dst[], int dstStride, int width, int height,
	QP_STORE_T QPs[], int QPStride, int isColor, pp_mode_t *vm, pp_context_t *vc)
{
	PPContext *c= (PPContext *)vc;
	PPMode *ppMode= (PPMode *)vm;
	c->ppMode= *ppMode; //FIXME

	// useing ifs here as they are faster than function pointers allthough the
	// difference wouldnt be messureable here but its much better because
	// someone might exchange the cpu whithout restarting mplayer ;)
#ifdef RUNTIME_CPUDETECT
#ifdef ARCH_X86
	// ordered per speed fasterst first
	if(c->cpuCaps & PP_CPU_CAPS_MMX2)
		postProcess_MMX2(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
	else if(c->cpuCaps & PP_CPU_CAPS_3DNOW)
		postProcess_3DNow(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
	else if(c->cpuCaps & PP_CPU_CAPS_MMX)
		postProcess_MMX(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
	else
		postProcess_C(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#else
		postProcess_C(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#endif
#else //RUNTIME_CPUDETECT
#ifdef HAVE_MMX2
		postProcess_MMX2(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#elif defined (HAVE_3DNOW)
		postProcess_3DNow(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#elif defined (HAVE_MMX)
		postProcess_MMX(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#else
		postProcess_C(src, srcStride, dst, dstStride, width, height, QPs, QPStride, isColor, c);
#endif
#endif //!RUNTIME_CPUDETECT
}

//static void postProcess(uint8_t src[], int srcStride, uint8_t dst[], int dstStride, int width, int height,
//	QP_STORE_T QPs[], int QPStride, int isColor, struct PPMode *ppMode);

/* -pp Command line Help
*/
char *pp_help=
"<filterName>[:<option>[:<option>...]][[,|/][-]<filterName>[:<option>...]]...\n"
"long form example:\n"
"vdeblock:autoq/hdeblock:autoq/linblenddeint	default,-vdeblock\n"
"short form example:\n"
"vb:a/hb:a/lb					de,-vb\n"
"more examples:\n"
"tn:64:128:256\n"
"Filters			Options\n"
"short	long name	short	long option	Description\n"
"*	*		a	autoq		cpu power dependant enabler\n"
"			c	chrom		chrominance filtring enabled\n"
"			y	nochrom		chrominance filtring disabled\n"
"hb	hdeblock	(2 Threshold)		horizontal deblocking filter\n"
"	1. difference factor: default=64, higher -> more deblocking\n"
"	2. flatness threshold: default=40, lower -> more deblocking\n"
"			the h & v deblocking filters share these\n"
"			so u cant set different thresholds for h / v\n"
"vb	vdeblock	(2 Threshold)		vertical deblocking filter\n"
"h1	x1hdeblock				Experimental h deblock filter 1\n"
"v1	x1vdeblock				Experimental v deblock filter 1\n"
"dr	dering					Deringing filter\n"
"al	autolevels				automatic brightness / contrast\n"
"			f	fullyrange	stretch luminance to (0..255)\n"
"lb	linblenddeint				linear blend deinterlacer\n"
"li	linipoldeint				linear interpolating deinterlace\n"
"ci	cubicipoldeint				cubic interpolating deinterlacer\n"
"md	mediandeint				median deinterlacer\n"
"fd	ffmpegdeint				ffmpeg deinterlacer\n"
"de	default					hb:a,vb:a,dr:a,al\n"
"fa	fast					h1:a,v1:a,dr:a,al\n"
"tn	tmpnoise	(3 Thresholds)		Temporal Noise Reducer\n"
"			1. <= 2. <= 3.		larger -> stronger filtering\n"
"fq	forceQuant	<quantizer>		Force quantizer\n"
;

pp_mode_t *pp_get_mode_by_name_and_quality(char *name, int quality)
{
	char temp[GET_MODE_BUFFER_SIZE];
	char *p= temp;
	char *filterDelimiters= ",/";
	char *optionDelimiters= ":";
	struct PPMode *ppMode;
	char *filterToken;

	ppMode= memalign(8, sizeof(PPMode));
	
	ppMode->lumMode= 0;
	ppMode->chromMode= 0;
	ppMode->maxTmpNoise[0]= 700;
	ppMode->maxTmpNoise[1]= 1500;
	ppMode->maxTmpNoise[2]= 3000;
	ppMode->maxAllowedY= 234;
	ppMode->minAllowedY= 16;
	ppMode->baseDcDiff= 256/4;
	ppMode->flatnessThreshold= 56-16;
	ppMode->maxClippedThreshold= 0.01;
	ppMode->error=0;

	strncpy(temp, name, GET_MODE_BUFFER_SIZE);

	if(verbose>1) printf("pp: %s\n", name);

	for(;;){
		char *filterName;
		int q= 1000000; //PP_QUALITY_MAX;
		int chrom=-1;
		char *option;
		char *options[OPTIONS_ARRAY_SIZE];
		int i;
		int filterNameOk=0;
		int numOfUnknownOptions=0;
		int enable=1; //does the user want us to enabled or disabled the filter

		filterToken= strtok(p, filterDelimiters);
		if(filterToken == NULL) break;
		p+= strlen(filterToken) + 1; // p points to next filterToken
		filterName= strtok(filterToken, optionDelimiters);
		if(verbose>1) printf("pp: %s::%s\n", filterToken, filterName);

		if(*filterName == '-')
		{
			enable=0;
			filterName++;
		}

		for(;;){ //for all options
			option= strtok(NULL, optionDelimiters);
			if(option == NULL) break;

			if(verbose>1) printf("pp: option: %s\n", option);
			if(!strcmp("autoq", option) || !strcmp("a", option)) q= quality;
			else if(!strcmp("nochrom", option) || !strcmp("y", option)) chrom=0;
			else if(!strcmp("chrom", option) || !strcmp("c", option)) chrom=1;
			else
			{
				options[numOfUnknownOptions] = option;
				numOfUnknownOptions++;
			}
			if(numOfUnknownOptions >= OPTIONS_ARRAY_SIZE-1) break;
		}
		options[numOfUnknownOptions] = NULL;

		/* replace stuff from the replace Table */
		for(i=0; replaceTable[2*i]!=NULL; i++)
		{
			if(!strcmp(replaceTable[2*i], filterName))
			{
				int newlen= strlen(replaceTable[2*i + 1]);
				int plen;
				int spaceLeft;

				if(p==NULL) p= temp, *p=0; 	//last filter
				else p--, *p=',';		//not last filter

				plen= strlen(p);
				spaceLeft= p - temp + plen;
				if(spaceLeft + newlen  >= GET_MODE_BUFFER_SIZE)
				{
					ppMode->error++;
					break;
				}
				memmove(p + newlen, p, plen+1);
				memcpy(p, replaceTable[2*i + 1], newlen);
				filterNameOk=1;
			}
		}

		for(i=0; filters[i].shortName!=NULL; i++)
		{
//			printf("Compareing %s, %s, %s\n", filters[i].shortName,filters[i].longName, filterName);
			if(   !strcmp(filters[i].longName, filterName)
			   || !strcmp(filters[i].shortName, filterName))
			{
				ppMode->lumMode &= ~filters[i].mask;
				ppMode->chromMode &= ~filters[i].mask;

				filterNameOk=1;
				if(!enable) break; // user wants to disable it

				if(q >= filters[i].minLumQuality)
					ppMode->lumMode|= filters[i].mask;
				if(chrom==1 || (chrom==-1 && filters[i].chromDefault))
					if(q >= filters[i].minChromQuality)
						ppMode->chromMode|= filters[i].mask;

				if(filters[i].mask == LEVEL_FIX)
				{
					int o;
					ppMode->minAllowedY= 16;
					ppMode->maxAllowedY= 234;
					for(o=0; options[o]!=NULL; o++)
					{
						if(  !strcmp(options[o],"fullyrange")
						   ||!strcmp(options[o],"f"))
						{
							ppMode->minAllowedY= 0;
							ppMode->maxAllowedY= 255;
							numOfUnknownOptions--;
						}
					}
				}
				else if(filters[i].mask == TEMP_NOISE_FILTER)
				{
					int o;
					int numOfNoises=0;

					for(o=0; options[o]!=NULL; o++)
					{
						char *tail;
						ppMode->maxTmpNoise[numOfNoises]=
							strtol(options[o], &tail, 0);
						if(tail!=options[o])
						{
							numOfNoises++;
							numOfUnknownOptions--;
							if(numOfNoises >= 3) break;
						}
					}
				}
				else if(filters[i].mask == V_DEBLOCK || filters[i].mask == H_DEBLOCK)
				{
					int o;

					for(o=0; options[o]!=NULL && o<2; o++)
					{
						char *tail;
						int val= strtol(options[o], &tail, 0);
						if(tail==options[o]) break;

						numOfUnknownOptions--;
						if(o==0) ppMode->baseDcDiff= val;
						else ppMode->flatnessThreshold= val;
					}
				}
				else if(filters[i].mask == FORCE_QUANT)
				{
					int o;
					ppMode->forcedQuant= 15;

					for(o=0; options[o]!=NULL && o<1; o++)
					{
						char *tail;
						int val= strtol(options[o], &tail, 0);
						if(tail==options[o]) break;

						numOfUnknownOptions--;
						ppMode->forcedQuant= val;
					}
				}
			}
		}
		if(!filterNameOk) ppMode->error++;
		ppMode->error += numOfUnknownOptions;
	}

	if(verbose>1) printf("pp: lumMode=%X, chromMode=%X\n", ppMode->lumMode, ppMode->chromMode);
	if(ppMode->error)
	{
		fprintf(stderr, "%d errors in postprocess string \"%s\"\n", ppMode->error, name);
		free(ppMode);
		return NULL;
	}
	return ppMode;
}

void pp_free_mode(pp_mode_t *mode){
    if(mode) free(mode);
}

pp_context_t *pp_get_context(int width, int height, int cpuCaps){
	PPContext *c= memalign(32, sizeof(PPContext));
	int i;
	int mbWidth = (width+15)>>4;
	int mbHeight= (height+15)>>4;
	
	c->cpuCaps= cpuCaps;

	c->tempBlocks= (uint8_t*)memalign(8, 2*16*8);
	c->yHistogram= (uint64_t*)memalign(8, 256*sizeof(uint64_t));
	for(i=0; i<256; i++)
		c->yHistogram[i]= width*height/64*15/256;

	for(i=0; i<3; i++)
	{
		//Note:the +17*1024 is just there so i dont have to worry about r/w over te end
		c->tempBlured[i]= (uint8_t*)memalign(8, ((width+7)&(~7))*2*((height+7)&(~7)) + 17*1024); //FIXME dstStride instead of width
		c->tempBluredPast[i]= (uint32_t*)memalign(8, 256*((height+7)&(~7))/2 + 17*1024);

		memset(c->tempBlured[i], 0, ((width+7)&(~7))*2*((height+7)&(~7)) + 17*1024);
		memset(c->tempBluredPast[i], 0, 256*((height+7)&(~7))/2 + 17*1024);
	}
	
	c->tempDst= (uint8_t*)memalign(8, 1024*24);
	c->tempSrc= (uint8_t*)memalign(8, 1024*24);
	c->tempDstBlock= (uint8_t*)memalign(8, 1024*24);
	c->tempSrcBlock= (uint8_t*)memalign(8, 1024*24);
	c->deintTemp= (uint8_t*)memalign(8, width+16);
	c->nonBQPTable= (QP_STORE_T*)memalign(8, mbWidth*mbHeight*sizeof(QP_STORE_T));
	memset(c->nonBQPTable, 0, mbWidth*mbHeight*sizeof(QP_STORE_T));

	c->frameNum=-1;

	return c;
}

void pp_free_context(void *vc){
	PPContext *c = (PPContext*)vc;
	int i;
	
	for(i=0; i<3; i++) free(c->tempBlured[i]);
	for(i=0; i<3; i++) free(c->tempBluredPast[i]);
	
	free(c->tempBlocks);
	free(c->yHistogram);
	free(c->tempDst);
	free(c->tempSrc);
	free(c->tempDstBlock);
	free(c->tempSrcBlock);
	free(c->deintTemp);
	free(c->nonBQPTable);
	
	free(c);
}

void  pp_postprocess(uint8_t * src[3], int srcStride[3],
                 uint8_t * dst[3], int dstStride[3],
                 int width, int height,
                 QP_STORE_T *QP_store,  int QPStride,
		 pp_mode_t *vm,  void *vc, int pict_type)
{
	int mbWidth = (width+15)>>4;
	int mbHeight= (height+15)>>4;
	QP_STORE_T quantArray[2048/8];
	PPMode *mode = (PPMode*)vm;
	PPContext *c = (PPContext*)vc;

	if(QP_store==NULL || (mode->lumMode & FORCE_QUANT)) 
	{
		int i;
		QP_store= quantArray;
		QPStride= 0;
		if(mode->lumMode & FORCE_QUANT)
			for(i=0; i<2048/8; i++) quantArray[i]= mode->forcedQuant;
		else
			for(i=0; i<2048/8; i++) quantArray[i]= 1;
	}
if(0){
int x,y;
for(y=0; y<mbHeight; y++){
	for(x=0; x<mbWidth; x++){
		printf("%2d ", QP_store[x + y*QPStride]);
	}
	printf("\n");
}
	printf("\n");
}
//printf("pict_type:%d\n", pict_type);

	if(pict_type!=3)
	{
		int x,y;
		for(y=0; y<mbHeight; y++){
			for(x=0; x<mbWidth; x++){
				int qscale= QP_store[x + y*QPStride];
				if(qscale&~31)
				    qscale=31;
				c->nonBQPTable[y*mbWidth + x]= qscale;
			}
		}
	}

	if(verbose>2)
	{
		printf("using npp filters 0x%X/0x%X\n", mode->lumMode, mode->chromMode);
	}

	postProcess(src[0], srcStride[0], dst[0], dstStride[0],
		width, height, QP_store, QPStride, 0, mode, c);

	width  = (width +1)>>1;
	height = (height+1)>>1;

	if(mode->chromMode)
	{
		postProcess(src[1], srcStride[1], dst[1], dstStride[1],
			width, height, QP_store, QPStride, 1, mode, c);
		postProcess(src[2], srcStride[2], dst[2], dstStride[2],
			width, height, QP_store, QPStride, 2, mode, c);
	}
	else if(srcStride[1] == dstStride[1] && srcStride[2] == dstStride[2])
	{
		memcpy(dst[1], src[1], srcStride[1]*height);
		memcpy(dst[2], src[2], srcStride[2]*height);
	}
	else
	{
		int y;
		for(y=0; y<height; y++)
		{
			memcpy(&(dst[1][y*dstStride[1]]), &(src[1][y*srcStride[1]]), width);
			memcpy(&(dst[2][y*dstStride[2]]), &(src[2][y*srcStride[2]]), width);
		}
	}
}