FFmpeg
 All Data Structures Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
vp3dsp.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2004 the ffmpeg project
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 /**
22  * @file
23  * Standard C DSP-oriented functions cribbed from the original VP3
24  * source code.
25  */
26 
27 #include "libavutil/attributes.h"
28 #include "libavutil/common.h"
29 #include "avcodec.h"
30 #include "dsputil.h"
31 #include "rnd_avg.h"
32 #include "vp3dsp.h"
33 
34 #define IdctAdjustBeforeShift 8
35 #define xC1S7 64277
36 #define xC2S6 60547
37 #define xC3S5 54491
38 #define xC4S4 46341
39 #define xC5S3 36410
40 #define xC6S2 25080
41 #define xC7S1 12785
42 
43 #define M(a,b) (((a) * (b))>>16)
44 
45 static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
46 {
47  int16_t *ip = input;
48 
49  int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
50  int Ed, Gd, Add, Bdd, Fd, Hd;
51 
52  int i;
53 
54  /* Inverse DCT on the rows now */
55  for (i = 0; i < 8; i++) {
56  /* Check for non-zero values */
57  if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
58  A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
59  B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
60  C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
61  D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
62 
63  Ad = M(xC4S4, (A - C));
64  Bd = M(xC4S4, (B - D));
65 
66  Cd = A + C;
67  Dd = B + D;
68 
69  E = M(xC4S4, (ip[0] + ip[4]));
70  F = M(xC4S4, (ip[0] - ip[4]));
71 
72  G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
73  H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
74 
75  Ed = E - G;
76  Gd = E + G;
77 
78  Add = F + Ad;
79  Bdd = Bd - H;
80 
81  Fd = F - Ad;
82  Hd = Bd + H;
83 
84  /* Final sequence of operations over-write original inputs. */
85  ip[0] = Gd + Cd ;
86  ip[7] = Gd - Cd ;
87 
88  ip[1] = Add + Hd;
89  ip[2] = Add - Hd;
90 
91  ip[3] = Ed + Dd ;
92  ip[4] = Ed - Dd ;
93 
94  ip[5] = Fd + Bdd;
95  ip[6] = Fd - Bdd;
96  }
97 
98  ip += 8; /* next row */
99  }
100 
101  ip = input;
102 
103  for ( i = 0; i < 8; i++) {
104  /* Check for non-zero values (bitwise or faster than ||) */
105  if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
106  ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
107 
108  A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
109  B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
110  C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
111  D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
112 
113  Ad = M(xC4S4, (A - C));
114  Bd = M(xC4S4, (B - D));
115 
116  Cd = A + C;
117  Dd = B + D;
118 
119  E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
120  F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;
121 
122  if(type==1){ //HACK
123  E += 16*128;
124  F += 16*128;
125  }
126 
127  G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
128  H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
129 
130  Ed = E - G;
131  Gd = E + G;
132 
133  Add = F + Ad;
134  Bdd = Bd - H;
135 
136  Fd = F - Ad;
137  Hd = Bd + H;
138 
139  /* Final sequence of operations over-write original inputs. */
140  if(type==0){
141  ip[0*8] = (Gd + Cd ) >> 4;
142  ip[7*8] = (Gd - Cd ) >> 4;
143 
144  ip[1*8] = (Add + Hd ) >> 4;
145  ip[2*8] = (Add - Hd ) >> 4;
146 
147  ip[3*8] = (Ed + Dd ) >> 4;
148  ip[4*8] = (Ed - Dd ) >> 4;
149 
150  ip[5*8] = (Fd + Bdd ) >> 4;
151  ip[6*8] = (Fd - Bdd ) >> 4;
152  }else if(type==1){
153  dst[0*stride] = av_clip_uint8((Gd + Cd ) >> 4);
154  dst[7*stride] = av_clip_uint8((Gd - Cd ) >> 4);
155 
156  dst[1*stride] = av_clip_uint8((Add + Hd ) >> 4);
157  dst[2*stride] = av_clip_uint8((Add - Hd ) >> 4);
158 
159  dst[3*stride] = av_clip_uint8((Ed + Dd ) >> 4);
160  dst[4*stride] = av_clip_uint8((Ed - Dd ) >> 4);
161 
162  dst[5*stride] = av_clip_uint8((Fd + Bdd ) >> 4);
163  dst[6*stride] = av_clip_uint8((Fd - Bdd ) >> 4);
164  }else{
165  dst[0*stride] = av_clip_uint8(dst[0*stride] + ((Gd + Cd ) >> 4));
166  dst[7*stride] = av_clip_uint8(dst[7*stride] + ((Gd - Cd ) >> 4));
167 
168  dst[1*stride] = av_clip_uint8(dst[1*stride] + ((Add + Hd ) >> 4));
169  dst[2*stride] = av_clip_uint8(dst[2*stride] + ((Add - Hd ) >> 4));
170 
171  dst[3*stride] = av_clip_uint8(dst[3*stride] + ((Ed + Dd ) >> 4));
172  dst[4*stride] = av_clip_uint8(dst[4*stride] + ((Ed - Dd ) >> 4));
173 
174  dst[5*stride] = av_clip_uint8(dst[5*stride] + ((Fd + Bdd ) >> 4));
175  dst[6*stride] = av_clip_uint8(dst[6*stride] + ((Fd - Bdd ) >> 4));
176  }
177 
178  } else {
179  if(type==0){
180  ip[0*8] =
181  ip[1*8] =
182  ip[2*8] =
183  ip[3*8] =
184  ip[4*8] =
185  ip[5*8] =
186  ip[6*8] =
187  ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
188  }else if(type==1){
189  dst[0*stride]=
190  dst[1*stride]=
191  dst[2*stride]=
192  dst[3*stride]=
193  dst[4*stride]=
194  dst[5*stride]=
195  dst[6*stride]=
196  dst[7*stride]= av_clip_uint8(128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20));
197  }else{
198  if(ip[0*8]){
199  int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
200  dst[0*stride] = av_clip_uint8(dst[0*stride] + v);
201  dst[1*stride] = av_clip_uint8(dst[1*stride] + v);
202  dst[2*stride] = av_clip_uint8(dst[2*stride] + v);
203  dst[3*stride] = av_clip_uint8(dst[3*stride] + v);
204  dst[4*stride] = av_clip_uint8(dst[4*stride] + v);
205  dst[5*stride] = av_clip_uint8(dst[5*stride] + v);
206  dst[6*stride] = av_clip_uint8(dst[6*stride] + v);
207  dst[7*stride] = av_clip_uint8(dst[7*stride] + v);
208  }
209  }
210  }
211 
212  ip++; /* next column */
213  dst++;
214  }
215 }
216 
217 static void vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size,
218  int16_t *block/*align 16*/)
219 {
220  idct(dest, line_size, block, 1);
221  memset(block, 0, sizeof(*block) * 64);
222 }
223 
224 static void vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size,
225  int16_t *block/*align 16*/)
226 {
227  idct(dest, line_size, block, 2);
228  memset(block, 0, sizeof(*block) * 64);
229 }
230 
231 static void vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size,
232  int16_t *block/*align 16*/)
233 {
234  int i, dc = (block[0] + 15) >> 5;
235 
236  for(i = 0; i < 8; i++){
237  dest[0] = av_clip_uint8(dest[0] + dc);
238  dest[1] = av_clip_uint8(dest[1] + dc);
239  dest[2] = av_clip_uint8(dest[2] + dc);
240  dest[3] = av_clip_uint8(dest[3] + dc);
241  dest[4] = av_clip_uint8(dest[4] + dc);
242  dest[5] = av_clip_uint8(dest[5] + dc);
243  dest[6] = av_clip_uint8(dest[6] + dc);
244  dest[7] = av_clip_uint8(dest[7] + dc);
245  dest += line_size;
246  }
247  block[0] = 0;
248 }
249 
250 static void vp3_v_loop_filter_c(uint8_t *first_pixel, int stride,
251  int *bounding_values)
252 {
253  unsigned char *end;
254  int filter_value;
255  const int nstride= -stride;
256 
257  for (end= first_pixel + 8; first_pixel < end; first_pixel++) {
258  filter_value =
259  (first_pixel[2 * nstride] - first_pixel[ stride])
260  +3*(first_pixel[0 ] - first_pixel[nstride]);
261  filter_value = bounding_values[(filter_value + 4) >> 3];
262  first_pixel[nstride] = av_clip_uint8(first_pixel[nstride] + filter_value);
263  first_pixel[0] = av_clip_uint8(first_pixel[0] - filter_value);
264  }
265 }
266 
267 static void vp3_h_loop_filter_c(uint8_t *first_pixel, int stride,
268  int *bounding_values)
269 {
270  unsigned char *end;
271  int filter_value;
272 
273  for (end= first_pixel + 8*stride; first_pixel != end; first_pixel += stride) {
274  filter_value =
275  (first_pixel[-2] - first_pixel[ 1])
276  +3*(first_pixel[ 0] - first_pixel[-1]);
277  filter_value = bounding_values[(filter_value + 4) >> 3];
278  first_pixel[-1] = av_clip_uint8(first_pixel[-1] + filter_value);
279  first_pixel[ 0] = av_clip_uint8(first_pixel[ 0] - filter_value);
280  }
281 }
282 
283 static void put_no_rnd_pixels_l2(uint8_t *dst, const uint8_t *src1,
284  const uint8_t *src2, ptrdiff_t stride, int h)
285 {
286  int i;
287 
288  for (i = 0; i < h; i++) {
289  uint32_t a, b;
290 
291  a = AV_RN32(&src1[i * stride]);
292  b = AV_RN32(&src2[i * stride]);
293  AV_WN32A(&dst[i * stride], no_rnd_avg32(a, b));
294  a = AV_RN32(&src1[i * stride + 4]);
295  b = AV_RN32(&src2[i * stride + 4]);
296  AV_WN32A(&dst[i * stride + 4], no_rnd_avg32(a, b));
297  }
298 }
299 
301 {
303 
309 
311 
312  if (ARCH_ARM)
313  ff_vp3dsp_init_arm(c, flags);
314  if (ARCH_BFIN)
315  ff_vp3dsp_init_bfin(c, flags);
316  if (ARCH_PPC)
317  ff_vp3dsp_init_ppc(c, flags);
318  if (ARCH_X86)
319  ff_vp3dsp_init_x86(c, flags);
320 }