/*
* Copyright (C) 2004 the ffmpeg project
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
- * version 2 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
- * License along with this library; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file vp3dsp.c
- * Standard C DSP-oriented functions cribbed from the original VP3
+ * Standard C DSP-oriented functions cribbed from the original VP3
* source code.
*/
#include "common.h"
#include "avcodec.h"
#include "dsputil.h"
-#include "vp3data.h"
#define IdctAdjustBeforeShift 8
#define xC1S7 64277
#define xC6S2 25080
#define xC7S1 12785
-void vp3_dsp_init_c(void)
-{
- /* nop */
-}
+#define M(a,b) (((a) * (b))>>16)
-void vp3_idct_c(int16_t *input_data, int16_t *dequant_matrix,
- int coeff_count, int16_t *output_data)
+static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
{
- int32_t dequantized_data[64];
- int32_t *ip = dequantized_data;
- int16_t *op = output_data;
+ int16_t *ip = input;
+ uint8_t *cm = cropTbl + MAX_NEG_CROP;
- int32_t A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_;
- int32_t _Ed, _Gd, _Add, _Bdd, _Fd, _Hd;
- int32_t t1, t2;
+ int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
+ int Ed, Gd, Add, Bdd, Fd, Hd;
- int i, j;
-
- /* de-zigzag and dequantize */
- for (i = 0; i < coeff_count; i++) {
- j = dezigzag_index[i];
- dequantized_data[j] = dequant_matrix[i] * input_data[i];
- }
+ int i;
/* Inverse DCT on the rows now */
for (i = 0; i < 8; i++) {
/* Check for non-zero values */
if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
- t1 = (int32_t)(xC1S7 * ip[1]);
- t2 = (int32_t)(xC7S1 * ip[7]);
- t1 >>= 16;
- t2 >>= 16;
- A_ = t1 + t2;
-
- t1 = (int32_t)(xC7S1 * ip[1]);
- t2 = (int32_t)(xC1S7 * ip[7]);
- t1 >>= 16;
- t2 >>= 16;
- B_ = t1 - t2;
-
- t1 = (int32_t)(xC3S5 * ip[3]);
- t2 = (int32_t)(xC5S3 * ip[5]);
- t1 >>= 16;
- t2 >>= 16;
- C_ = t1 + t2;
-
- t1 = (int32_t)(xC3S5 * ip[5]);
- t2 = (int32_t)(xC5S3 * ip[3]);
- t1 >>= 16;
- t2 >>= 16;
- D_ = t1 - t2;
-
-
- t1 = (int32_t)(xC4S4 * (A_ - C_));
- t1 >>= 16;
- _Ad = t1;
-
- t1 = (int32_t)(xC4S4 * (B_ - D_));
- t1 >>= 16;
- _Bd = t1;
-
-
- _Cd = A_ + C_;
- _Dd = B_ + D_;
-
- t1 = (int32_t)(xC4S4 * (ip[0] + ip[4]));
- t1 >>= 16;
- E_ = t1;
+ A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
+ B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
+ C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
+ D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
- t1 = (int32_t)(xC4S4 * (ip[0] - ip[4]));
- t1 >>= 16;
- F_ = t1;
+ Ad = M(xC4S4, (A - C));
+ Bd = M(xC4S4, (B - D));
- t1 = (int32_t)(xC2S6 * ip[2]);
- t2 = (int32_t)(xC6S2 * ip[6]);
- t1 >>= 16;
- t2 >>= 16;
- G_ = t1 + t2;
+ Cd = A + C;
+ Dd = B + D;
- t1 = (int32_t)(xC6S2 * ip[2]);
- t2 = (int32_t)(xC2S6 * ip[6]);
- t1 >>= 16;
- t2 >>= 16;
- H_ = t1 - t2;
+ E = M(xC4S4, (ip[0] + ip[4]));
+ F = M(xC4S4, (ip[0] - ip[4]));
+ G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
+ H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
- _Ed = E_ - G_;
- _Gd = E_ + G_;
+ Ed = E - G;
+ Gd = E + G;
- _Add = F_ + _Ad;
- _Bdd = _Bd - H_;
+ Add = F + Ad;
+ Bdd = Bd - H;
- _Fd = F_ - _Ad;
- _Hd = _Bd + H_;
+ Fd = F - Ad;
+ Hd = Bd + H;
/* Final sequence of operations over-write original inputs. */
- ip[0] = (int16_t)((_Gd + _Cd ) >> 0);
- ip[7] = (int16_t)((_Gd - _Cd ) >> 0);
+ ip[0] = Gd + Cd ;
+ ip[7] = Gd - Cd ;
- ip[1] = (int16_t)((_Add + _Hd ) >> 0);
- ip[2] = (int16_t)((_Add - _Hd ) >> 0);
+ ip[1] = Add + Hd;
+ ip[2] = Add - Hd;
- ip[3] = (int16_t)((_Ed + _Dd ) >> 0);
- ip[4] = (int16_t)((_Ed - _Dd ) >> 0);
-
- ip[5] = (int16_t)((_Fd + _Bdd ) >> 0);
- ip[6] = (int16_t)((_Fd - _Bdd ) >> 0);
+ ip[3] = Ed + Dd ;
+ ip[4] = Ed - Dd ;
+ ip[5] = Fd + Bdd;
+ ip[6] = Fd - Bdd;
}
ip += 8; /* next row */
}
- ip = dequantized_data;
+ ip = input;
for ( i = 0; i < 8; i++) {
/* Check for non-zero values (bitwise or faster than ||) */
- if ( ip[0 * 8] | ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
+ if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
- t1 = (int32_t)(xC1S7 * ip[1*8]);
- t2 = (int32_t)(xC7S1 * ip[7*8]);
- t1 >>= 16;
- t2 >>= 16;
- A_ = t1 + t2;
-
- t1 = (int32_t)(xC7S1 * ip[1*8]);
- t2 = (int32_t)(xC1S7 * ip[7*8]);
- t1 >>= 16;
- t2 >>= 16;
- B_ = t1 - t2;
-
- t1 = (int32_t)(xC3S5 * ip[3*8]);
- t2 = (int32_t)(xC5S3 * ip[5*8]);
- t1 >>= 16;
- t2 >>= 16;
- C_ = t1 + t2;
-
- t1 = (int32_t)(xC3S5 * ip[5*8]);
- t2 = (int32_t)(xC5S3 * ip[3*8]);
- t1 >>= 16;
- t2 >>= 16;
- D_ = t1 - t2;
+ A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
+ B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
+ C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
+ D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
+ Ad = M(xC4S4, (A - C));
+ Bd = M(xC4S4, (B - D));
- t1 = (int32_t)(xC4S4 * (A_ - C_));
- t1 >>= 16;
- _Ad = t1;
+ Cd = A + C;
+ Dd = B + D;
- t1 = (int32_t)(xC4S4 * (B_ - D_));
- t1 >>= 16;
- _Bd = t1;
+ E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
+ F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;
+ if(type==1){ //HACK
+ E += 16*128;
+ F += 16*128;
+ }
- _Cd = A_ + C_;
- _Dd = B_ + D_;
+ G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
+ H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
- t1 = (int32_t)(xC4S4 * (ip[0*8] + ip[4*8]));
- t1 >>= 16;
- E_ = t1;
+ Ed = E - G;
+ Gd = E + G;
- t1 = (int32_t)(xC4S4 * (ip[0*8] - ip[4*8]));
- t1 >>= 16;
- F_ = t1;
+ Add = F + Ad;
+ Bdd = Bd - H;
- t1 = (int32_t)(xC2S6 * ip[2*8]);
- t2 = (int32_t)(xC6S2 * ip[6*8]);
- t1 >>= 16;
- t2 >>= 16;
- G_ = t1 + t2;
+ Fd = F - Ad;
+ Hd = Bd + H;
- t1 = (int32_t)(xC6S2 * ip[2*8]);
- t2 = (int32_t)(xC2S6 * ip[6*8]);
- t1 >>= 16;
- t2 >>= 16;
- H_ = t1 - t2;
+ /* Final sequence of operations over-write original inputs. */
+ if(type==0){
+ ip[0*8] = (Gd + Cd ) >> 4;
+ ip[7*8] = (Gd - Cd ) >> 4;
+ ip[1*8] = (Add + Hd ) >> 4;
+ ip[2*8] = (Add - Hd ) >> 4;
- _Ed = E_ - G_;
- _Gd = E_ + G_;
+ ip[3*8] = (Ed + Dd ) >> 4;
+ ip[4*8] = (Ed - Dd ) >> 4;
- _Add = F_ + _Ad;
- _Bdd = _Bd - H_;
+ ip[5*8] = (Fd + Bdd ) >> 4;
+ ip[6*8] = (Fd - Bdd ) >> 4;
+ }else if(type==1){
+ dst[0*stride] = cm[(Gd + Cd ) >> 4];
+ dst[7*stride] = cm[(Gd - Cd ) >> 4];
- _Fd = F_ - _Ad;
- _Hd = _Bd + H_;
+ dst[1*stride] = cm[(Add + Hd ) >> 4];
+ dst[2*stride] = cm[(Add - Hd ) >> 4];
- _Gd += IdctAdjustBeforeShift;
- _Add += IdctAdjustBeforeShift;
- _Ed += IdctAdjustBeforeShift;
- _Fd += IdctAdjustBeforeShift;
+ dst[3*stride] = cm[(Ed + Dd ) >> 4];
+ dst[4*stride] = cm[(Ed - Dd ) >> 4];
- /* Final sequence of operations over-write original inputs. */
- op[0*8] = (int16_t)((_Gd + _Cd ) >> 4);
- op[7*8] = (int16_t)((_Gd - _Cd ) >> 4);
+ dst[5*stride] = cm[(Fd + Bdd ) >> 4];
+ dst[6*stride] = cm[(Fd - Bdd ) >> 4];
+ }else{
+ dst[0*stride] = cm[dst[0*stride] + ((Gd + Cd ) >> 4)];
+ dst[7*stride] = cm[dst[7*stride] + ((Gd - Cd ) >> 4)];
- op[1*8] = (int16_t)((_Add + _Hd ) >> 4);
- op[2*8] = (int16_t)((_Add - _Hd ) >> 4);
+ dst[1*stride] = cm[dst[1*stride] + ((Add + Hd ) >> 4)];
+ dst[2*stride] = cm[dst[2*stride] + ((Add - Hd ) >> 4)];
- op[3*8] = (int16_t)((_Ed + _Dd ) >> 4);
- op[4*8] = (int16_t)((_Ed - _Dd ) >> 4);
+ dst[3*stride] = cm[dst[3*stride] + ((Ed + Dd ) >> 4)];
+ dst[4*stride] = cm[dst[4*stride] + ((Ed - Dd ) >> 4)];
- op[5*8] = (int16_t)((_Fd + _Bdd ) >> 4);
- op[6*8] = (int16_t)((_Fd - _Bdd ) >> 4);
+ dst[5*stride] = cm[dst[5*stride] + ((Fd + Bdd ) >> 4)];
+ dst[6*stride] = cm[dst[6*stride] + ((Fd - Bdd ) >> 4)];
+ }
} else {
-
- op[0*8] = 0;
- op[7*8] = 0;
- op[1*8] = 0;
- op[2*8] = 0;
- op[3*8] = 0;
- op[4*8] = 0;
- op[5*8] = 0;
- op[6*8] = 0;
+ if(type==0){
+ ip[0*8] =
+ ip[1*8] =
+ ip[2*8] =
+ ip[3*8] =
+ ip[4*8] =
+ ip[5*8] =
+ ip[6*8] =
+ ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
+ }else if(type==1){
+ dst[0*stride]=
+ dst[1*stride]=
+ dst[2*stride]=
+ dst[3*stride]=
+ dst[4*stride]=
+ dst[5*stride]=
+ dst[6*stride]=
+ dst[7*stride]= 128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
+ }else{
+ if(ip[0*8]){
+ int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
+ dst[0*stride] = cm[dst[0*stride] + v];
+ dst[1*stride] = cm[dst[1*stride] + v];
+ dst[2*stride] = cm[dst[2*stride] + v];
+ dst[3*stride] = cm[dst[3*stride] + v];
+ dst[4*stride] = cm[dst[4*stride] + v];
+ dst[5*stride] = cm[dst[5*stride] + v];
+ dst[6*stride] = cm[dst[6*stride] + v];
+ dst[7*stride] = cm[dst[7*stride] + v];
+ }
+ }
}
ip++; /* next column */
- op++;
+ dst++;
}
}
+
+void ff_vp3_idct_c(DCTELEM *block/* align 16*/){
+ idct(NULL, 0, block, 0);
+}
+
+void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+ idct(dest, line_size, block, 1);
+}
+
+void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+ idct(dest, line_size, block, 2);
+}