X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fvp8dsp.c;h=34e2da0c345cdc02c78a9f3e9c2d3c4da2ddeaa8;hb=2df7f7714a12a59d31058aba15fb1e348e36b0ab;hp=4fb602522f930801ac8b027a5d933d763c664db3;hpb=0178d14fe53762e432f64353d36cc4c757fb867f;p=ffmpeg diff --git a/libavcodec/vp8dsp.c b/libavcodec/vp8dsp.c index 4fb602522f9..34e2da0c345 100644 --- a/libavcodec/vp8dsp.c +++ b/libavcodec/vp8dsp.c @@ -1,31 +1,35 @@ -/** - * VP8 compatible video decoder - * +/* * Copyright (C) 2010 David Conrad * Copyright (C) 2010 Ronald S. Bultje * - * This file is part of FFmpeg. + * This file is part of Libav. * - * FFmpeg is free software; you can redistribute it and/or + * Libav 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 2.1 of the License, or (at your option) any later version. * - * FFmpeg is distributed in the hope that it will be useful, + * Libav 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 FFmpeg; if not, write to the Free Software + * License along with Libav; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ +/** + * @file + * VP8 compatible video decoder + */ + #include "dsputil.h" #include "vp8dsp.h" +#include "libavutil/common.h" // TODO: Maybe add dequant -static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16]) +static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16]) { int i, t0, t1, t2, t3; @@ -46,28 +50,48 @@ static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16]) t1 = dc[i*4+1] + dc[i*4+2]; t2 = dc[i*4+1] - dc[i*4+2]; t3 = dc[i*4+0] - dc[i*4+3] + 3; // rounding - - *block[i][0] = (t0 + t1) >> 3; - *block[i][1] = (t3 + t2) >> 3; - *block[i][2] = (t0 - t1) >> 3; - *block[i][3] = (t3 - t2) >> 3; + dc[i*4+0] = 0; + dc[i*4+1] = 0; + dc[i*4+2] = 0; + dc[i*4+3] = 0; + + block[i][0][0] = (t0 + t1) >> 3; + block[i][1][0] = (t3 + t2) >> 3; + block[i][2][0] = (t0 - t1) >> 3; + block[i][3][0] = (t3 - t2) >> 3; } } +static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16]) +{ + int i, val = (dc[0] + 3) >> 3; + dc[0] = 0; + + for (i = 0; i < 4; i++) { + block[i][0][0] = val; + block[i][1][0] = val; + block[i][2][0] = val; + block[i][3][0] = val; + } +} #define MUL_20091(a) ((((a)*20091) >> 16) + (a)) #define MUL_35468(a) (((a)*35468) >> 16) -static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], int stride) +static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride) { int i, t0, t1, t2, t3; - DCTELEM tmp[16]; + int16_t tmp[16]; for (i = 0; i < 4; i++) { t0 = block[0*4+i] + block[2*4+i]; t1 = block[0*4+i] - block[2*4+i]; t2 = MUL_35468(block[1*4+i]) - MUL_20091(block[3*4+i]); t3 = MUL_20091(block[1*4+i]) + MUL_35468(block[3*4+i]); + block[0*4+i] = 0; + block[1*4+i] = 0; + block[2*4+i] = 0; + block[3*4+i] = 0; tmp[i*4+0] = t0 + t3; tmp[i*4+1] = t1 + t2; @@ -89,9 +113,10 @@ static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], int stride) } } -static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], int stride) +static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride) { int i, dc = (block[0] + 4) >> 3; + block[0] = 0; for (i = 0; i < 4; i++) { dst[0] = av_clip_uint8(dst[0] + dc); @@ -102,6 +127,21 @@ static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], int stride) } } +static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride) +{ + vp8_idct_dc_add_c(dst+stride*0+0, block[0], stride); + vp8_idct_dc_add_c(dst+stride*0+4, block[1], stride); + vp8_idct_dc_add_c(dst+stride*4+0, block[2], stride); + vp8_idct_dc_add_c(dst+stride*4+4, block[3], stride); +} + +static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride) +{ + vp8_idct_dc_add_c(dst+ 0, block[0], stride); + vp8_idct_dc_add_c(dst+ 4, block[1], stride); + vp8_idct_dc_add_c(dst+ 8, block[2], stride); + vp8_idct_dc_add_c(dst+12, block[3], stride); +} // because I like only having two parameters to pass functions... #define LOAD_PIXELS\ @@ -114,17 +154,20 @@ static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], int stride) int av_unused q2 = p[ 2*stride];\ int av_unused q3 = p[ 3*stride]; -static av_always_inline void filter_common(uint8_t *p, int stride, int is4tap) +#define clip_int8(n) (cm[n+0x80]-0x80) + +static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, int is4tap) { LOAD_PIXELS int a, f1, f2; + const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; a = 3*(q0 - p0); if (is4tap) - a += av_clip_int8(p1 - q1); + a += clip_int8(p1 - q1); - a = av_clip_int8(a); + a = clip_int8(a); // We deviate from the spec here with c(a+3) >> 3 // since that's what libvpx does. @@ -133,18 +176,18 @@ static av_always_inline void filter_common(uint8_t *p, int stride, int is4tap) // Despite what the spec says, we do need to clamp here to // be bitexact with libvpx. - p[-1*stride] = av_clip_uint8(p0 + f2); - p[ 0*stride] = av_clip_uint8(q0 - f1); + p[-1*stride] = cm[p0 + f2]; + p[ 0*stride] = cm[q0 - f1]; // only used for _inner on blocks without high edge variance if (!is4tap) { a = (f1+1)>>1; - p[-2*stride] = av_clip_uint8(p1 + a); - p[ 1*stride] = av_clip_uint8(q1 - a); + p[-2*stride] = cm[p1 + a]; + p[ 1*stride] = cm[q1 - a]; } } -static av_always_inline int simple_limit(uint8_t *p, int stride, int flim) +static av_always_inline int simple_limit(uint8_t *p, ptrdiff_t stride, int flim) { LOAD_PIXELS return 2*FFABS(p0-q0) + (FFABS(p1-q1) >> 1) <= flim; @@ -154,44 +197,45 @@ static av_always_inline int simple_limit(uint8_t *p, int stride, int flim) * E - limit at the macroblock edge * I - limit for interior difference */ -static av_always_inline int normal_limit(uint8_t *p, int stride, int E, int I) +static av_always_inline int normal_limit(uint8_t *p, ptrdiff_t stride, int E, int I) { LOAD_PIXELS - return simple_limit(p, stride, 2*E+I) + return simple_limit(p, stride, E) && FFABS(p3-p2) <= I && FFABS(p2-p1) <= I && FFABS(p1-p0) <= I && FFABS(q3-q2) <= I && FFABS(q2-q1) <= I && FFABS(q1-q0) <= I; } // high edge variance -static av_always_inline int hev(uint8_t *p, int stride, int thresh) +static av_always_inline int hev(uint8_t *p, ptrdiff_t stride, int thresh) { LOAD_PIXELS return FFABS(p1-p0) > thresh || FFABS(q1-q0) > thresh; } -static av_always_inline void filter_mbedge(uint8_t *p, int stride) +static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride) { int a0, a1, a2, w; + const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; LOAD_PIXELS - w = av_clip_int8(p1-q1); - w = av_clip_int8(w + 3*(q0-p0)); + w = clip_int8(p1-q1); + w = clip_int8(w + 3*(q0-p0)); a0 = (27*w + 63) >> 7; a1 = (18*w + 63) >> 7; a2 = ( 9*w + 63) >> 7; - p[-3*stride] = av_clip_uint8(p2 + a2); - p[-2*stride] = av_clip_uint8(p1 + a1); - p[-1*stride] = av_clip_uint8(p0 + a0); - p[ 0*stride] = av_clip_uint8(q0 - a0); - p[ 1*stride] = av_clip_uint8(q1 - a1); - p[ 2*stride] = av_clip_uint8(q2 - a2); + p[-3*stride] = cm[p2 + a2]; + p[-2*stride] = cm[p1 + a1]; + p[-1*stride] = cm[p0 + a0]; + p[ 0*stride] = cm[q0 - a0]; + p[ 1*stride] = cm[q1 - a1]; + p[ 2*stride] = cm[q2 - a2]; } -#define LOOP_FILTER(dir, size, stridea, strideb) \ -static void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, int stride,\ +#define LOOP_FILTER(dir, size, stridea, strideb, maybe_inline) \ +static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, ptrdiff_t stride,\ int flim_E, int flim_I, int hev_thresh)\ {\ int i;\ @@ -205,24 +249,43 @@ static void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, int stride,\ }\ }\ \ -static void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, int stride,\ +static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, ptrdiff_t stride,\ int flim_E, int flim_I, int hev_thresh)\ {\ - int i, hv;\ + int i;\ \ for (i = 0; i < size; i++)\ if (normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\ - hv = hev(dst+i*stridea, strideb, hev_thresh);\ - filter_common(dst+i*stridea, strideb, hv);\ + int hv = hev(dst+i*stridea, strideb, hev_thresh);\ + if (hv) \ + filter_common(dst+i*stridea, strideb, 1);\ + else \ + filter_common(dst+i*stridea, strideb, 0);\ }\ } -LOOP_FILTER(v, 16, 1, stride) -LOOP_FILTER(h, 16, stride, 1) -LOOP_FILTER(v, 8, 1, stride) -LOOP_FILTER(h, 8, stride, 1) +LOOP_FILTER(v, 16, 1, stride,) +LOOP_FILTER(h, 16, stride, 1,) + +#define UV_LOOP_FILTER(dir, stridea, strideb) \ +LOOP_FILTER(dir, 8, stridea, strideb, av_always_inline) \ +static void vp8_ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\ + int fE, int fI, int hev_thresh)\ +{\ + vp8_ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);\ + vp8_ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);\ +}\ +static void vp8_ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\ + int fE, int fI, int hev_thresh)\ +{\ + vp8_ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);\ + vp8_ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);\ +} + +UV_LOOP_FILTER(v, 1, stride) +UV_LOOP_FILTER(h, stride, 1) -static void vp8_v_loop_filter_simple_c(uint8_t *dst, int stride, int flim) +static void vp8_v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim) { int i; @@ -231,7 +294,7 @@ static void vp8_v_loop_filter_simple_c(uint8_t *dst, int stride, int flim) filter_common(dst+i, stride, 1); } -static void vp8_h_loop_filter_simple_c(uint8_t *dst, int stride, int flim) +static void vp8_h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim) { int i; @@ -251,7 +314,7 @@ static const uint8_t subpel_filters[7][6] = { }; #define PUT_PIXELS(WIDTH) \ -static void put_vp8_pixels ## WIDTH ##_c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int x, int y) { \ +static void put_vp8_pixels ## WIDTH ##_c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int x, int y) { \ int i; \ for (i = 0; i < h; i++, dst+= dststride, src+= srcstride) { \ memcpy(dst, src, WIDTH); \ @@ -263,93 +326,96 @@ PUT_PIXELS(8) PUT_PIXELS(4) #define FILTER_6TAP(src, F, stride) \ - av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \ - F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7) + cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \ + F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7] #define FILTER_4TAP(src, F, stride) \ - av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \ - F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7) + cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \ + F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7] -#define VP8_EPEL_H(SIZE, FILTER, FILTERNAME) \ -static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \ +#define VP8_EPEL_H(SIZE, TAPS) \ +static void put_vp8_epel ## SIZE ## _h ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ { \ const uint8_t *filter = subpel_filters[mx-1]; \ + const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \ int x, y; \ \ for (y = 0; y < h; y++) { \ for (x = 0; x < SIZE; x++) \ - dst[x] = FILTER(src, filter, 1); \ + dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \ dst += dststride; \ src += srcstride; \ } \ } -#define VP8_EPEL_V(SIZE, FILTER, FILTERNAME) \ -static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \ +#define VP8_EPEL_V(SIZE, TAPS) \ +static void put_vp8_epel ## SIZE ## _v ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ { \ const uint8_t *filter = subpel_filters[my-1]; \ + const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \ int x, y; \ \ for (y = 0; y < h; y++) { \ for (x = 0; x < SIZE; x++) \ - dst[x] = FILTER(src, filter, srcstride); \ + dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \ dst += dststride; \ src += srcstride; \ } \ } -#define VP8_EPEL_HV(SIZE, FILTERX, FILTERY, FILTERNAME) \ -static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \ +#define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \ +static void put_vp8_epel ## SIZE ## _h ## HTAPS ## v ## VTAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ { \ const uint8_t *filter = subpel_filters[mx-1]; \ + const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \ int x, y; \ - uint8_t tmp_array[(2*SIZE+5)*SIZE]; \ + uint8_t tmp_array[(2*SIZE+VTAPS-1)*SIZE]; \ uint8_t *tmp = tmp_array; \ - src -= 2*srcstride; \ + src -= (2-(VTAPS==4))*srcstride; \ \ - for (y = 0; y < h+5; y++) { \ + for (y = 0; y < h+VTAPS-1; y++) { \ for (x = 0; x < SIZE; x++) \ - tmp[x] = FILTERX(src, filter, 1); \ + tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \ tmp += SIZE; \ src += srcstride; \ } \ \ - tmp = tmp_array + 2*SIZE; \ + tmp = tmp_array + (2-(VTAPS==4))*SIZE; \ filter = subpel_filters[my-1]; \ \ for (y = 0; y < h; y++) { \ for (x = 0; x < SIZE; x++) \ - dst[x] = FILTERY(tmp, filter, SIZE); \ + dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \ dst += dststride; \ tmp += SIZE; \ } \ } -VP8_EPEL_H(16, FILTER_4TAP, h4) -VP8_EPEL_H(8, FILTER_4TAP, h4) -VP8_EPEL_H(4, FILTER_4TAP, h4) -VP8_EPEL_H(16, FILTER_6TAP, h6) -VP8_EPEL_H(8, FILTER_6TAP, h6) -VP8_EPEL_H(4, FILTER_6TAP, h6) -VP8_EPEL_V(16, FILTER_4TAP, v4) -VP8_EPEL_V(8, FILTER_4TAP, v4) -VP8_EPEL_V(4, FILTER_4TAP, v4) -VP8_EPEL_V(16, FILTER_6TAP, v6) -VP8_EPEL_V(8, FILTER_6TAP, v6) -VP8_EPEL_V(4, FILTER_6TAP, v6) -VP8_EPEL_HV(16, FILTER_4TAP, FILTER_4TAP, h4v4) -VP8_EPEL_HV(8, FILTER_4TAP, FILTER_4TAP, h4v4) -VP8_EPEL_HV(4, FILTER_4TAP, FILTER_4TAP, h4v4) -VP8_EPEL_HV(16, FILTER_4TAP, FILTER_6TAP, h4v6) -VP8_EPEL_HV(8, FILTER_4TAP, FILTER_6TAP, h4v6) -VP8_EPEL_HV(4, FILTER_4TAP, FILTER_6TAP, h4v6) -VP8_EPEL_HV(16, FILTER_6TAP, FILTER_4TAP, h6v4) -VP8_EPEL_HV(8, FILTER_6TAP, FILTER_4TAP, h6v4) -VP8_EPEL_HV(4, FILTER_6TAP, FILTER_4TAP, h6v4) -VP8_EPEL_HV(16, FILTER_6TAP, FILTER_6TAP, h6v6) -VP8_EPEL_HV(8, FILTER_6TAP, FILTER_6TAP, h6v6) -VP8_EPEL_HV(4, FILTER_6TAP, FILTER_6TAP, h6v6) +VP8_EPEL_H(16, 4) +VP8_EPEL_H(8, 4) +VP8_EPEL_H(4, 4) +VP8_EPEL_H(16, 6) +VP8_EPEL_H(8, 6) +VP8_EPEL_H(4, 6) +VP8_EPEL_V(16, 4) +VP8_EPEL_V(8, 4) +VP8_EPEL_V(4, 4) +VP8_EPEL_V(16, 6) +VP8_EPEL_V(8, 6) +VP8_EPEL_V(4, 6) +VP8_EPEL_HV(16, 4, 4) +VP8_EPEL_HV(8, 4, 4) +VP8_EPEL_HV(4, 4, 4) +VP8_EPEL_HV(16, 4, 6) +VP8_EPEL_HV(8, 4, 6) +VP8_EPEL_HV(4, 4, 6) +VP8_EPEL_HV(16, 6, 4) +VP8_EPEL_HV(8, 6, 4) +VP8_EPEL_HV(4, 6, 4) +VP8_EPEL_HV(16, 6, 6) +VP8_EPEL_HV(8, 6, 6) +VP8_EPEL_HV(4, 6, 6) #define VP8_BILINEAR(SIZE) \ -static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, int stride, uint8_t *src, int s2, int h, int mx, int my) \ +static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \ { \ int a = 8-mx, b = mx; \ int x, y; \ @@ -361,7 +427,7 @@ static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, int stride, uint8_t * src += stride; \ } \ } \ -static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, int stride, uint8_t *src, int s2, int h, int mx, int my) \ +static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \ { \ int c = 8-my, d = my; \ int x, y; \ @@ -374,7 +440,7 @@ static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, int stride, uint8_t * } \ } \ \ -static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, int stride, uint8_t *src, int s2, int h, int mx, int my) \ +static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \ { \ int a = 8-mx, b = mx; \ int c = 8-my, d = my; \ @@ -427,19 +493,22 @@ VP8_BILINEAR(4) av_cold void ff_vp8dsp_init(VP8DSPContext *dsp) { - dsp->vp8_luma_dc_wht = vp8_luma_dc_wht_c; - dsp->vp8_idct_add = vp8_idct_add_c; - dsp->vp8_idct_dc_add = vp8_idct_dc_add_c; - - dsp->vp8_v_loop_filter16 = vp8_v_loop_filter16_c; - dsp->vp8_h_loop_filter16 = vp8_h_loop_filter16_c; - dsp->vp8_v_loop_filter8 = vp8_v_loop_filter8_c; - dsp->vp8_h_loop_filter8 = vp8_h_loop_filter8_c; - - dsp->vp8_v_loop_filter16_inner = vp8_v_loop_filter16_inner_c; - dsp->vp8_h_loop_filter16_inner = vp8_h_loop_filter16_inner_c; - dsp->vp8_v_loop_filter8_inner = vp8_v_loop_filter8_inner_c; - dsp->vp8_h_loop_filter8_inner = vp8_h_loop_filter8_inner_c; + dsp->vp8_luma_dc_wht = vp8_luma_dc_wht_c; + dsp->vp8_luma_dc_wht_dc = vp8_luma_dc_wht_dc_c; + dsp->vp8_idct_add = vp8_idct_add_c; + dsp->vp8_idct_dc_add = vp8_idct_dc_add_c; + dsp->vp8_idct_dc_add4y = vp8_idct_dc_add4y_c; + dsp->vp8_idct_dc_add4uv = vp8_idct_dc_add4uv_c; + + dsp->vp8_v_loop_filter16y = vp8_v_loop_filter16_c; + dsp->vp8_h_loop_filter16y = vp8_h_loop_filter16_c; + dsp->vp8_v_loop_filter8uv = vp8_v_loop_filter8uv_c; + dsp->vp8_h_loop_filter8uv = vp8_h_loop_filter8uv_c; + + dsp->vp8_v_loop_filter16y_inner = vp8_v_loop_filter16_inner_c; + dsp->vp8_h_loop_filter16y_inner = vp8_h_loop_filter16_inner_c; + dsp->vp8_v_loop_filter8uv_inner = vp8_v_loop_filter8uv_inner_c; + dsp->vp8_h_loop_filter8uv_inner = vp8_h_loop_filter8uv_inner_c; dsp->vp8_v_loop_filter_simple = vp8_v_loop_filter_simple_c; dsp->vp8_h_loop_filter_simple = vp8_h_loop_filter_simple_c; @@ -452,6 +521,10 @@ av_cold void ff_vp8dsp_init(VP8DSPContext *dsp) VP8_BILINEAR_MC_FUNC(1, 8); VP8_BILINEAR_MC_FUNC(2, 4); + if (ARCH_ARM) + ff_vp8dsp_init_arm(dsp); + if (ARCH_PPC) + ff_vp8dsp_init_ppc(dsp); if (ARCH_X86) ff_vp8dsp_init_x86(dsp); }