-/**
- * 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])
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(DCTELEM block[4][4][16], DCTELEM 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, DCTELEM block[16], ptrdiff_t stride)
{
int i, t0, t1, t2, t3;
- uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
DCTELEM tmp[16];
for (i = 0; i < 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;
t2 = MUL_35468(tmp[1*4+i]) - MUL_20091(tmp[3*4+i]);
t3 = MUL_20091(tmp[1*4+i]) + MUL_35468(tmp[3*4+i]);
- dst[0] = cm[dst[0] + ((t0 + t3 + 4) >> 3)];
- dst[1] = cm[dst[1] + ((t1 + t2 + 4) >> 3)];
- dst[2] = cm[dst[2] + ((t1 - t2 + 4) >> 3)];
- dst[3] = cm[dst[3] + ((t0 - t3 + 4) >> 3)];
+ dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3));
+ dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3));
+ dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3));
+ dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3));
dst += 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, DCTELEM block[16], ptrdiff_t stride)
{
int i, dc = (block[0] + 4) >> 3;
- uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + dc;
+ block[0] = 0;
for (i = 0; i < 4; i++) {
- dst[0] = cm[dst[0]];
- dst[1] = cm[dst[1]];
- dst[2] = cm[dst[2]];
- dst[3] = cm[dst[3]];
+ dst[0] = av_clip_uint8(dst[0] + dc);
+ dst[1] = av_clip_uint8(dst[1] + dc);
+ dst[2] = av_clip_uint8(dst[2] + dc);
+ dst[3] = av_clip_uint8(dst[3] + dc);
dst += stride;
}
}
+static void vp8_idct_dc_add4uv_c(uint8_t *dst, DCTELEM 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, DCTELEM 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\
#define clip_int8(n) (cm[n+0x80]-0x80)
-static av_always_inline void filter_common(uint8_t *p, int stride, int is4tap)
+static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, int is4tap)
{
LOAD_PIXELS
int a, f1, f2;
}
}
-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;
* 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, E)
}
// 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;
uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
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;\
}\
}\
\
-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;\
}\
}
-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,)
-static void vp8_v_loop_filter_simple_c(uint8_t *dst, int stride, int flim)
+#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, ptrdiff_t stride, int flim)
{
int i;
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;
};
#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); \
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]; \
uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
\
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]; \
uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
\
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]; \
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; \
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; \
} \
} \
\
-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; \
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;
VP8_BILINEAR_MC_FUNC(1, 8);
VP8_BILINEAR_MC_FUNC(2, 4);
- if (HAVE_MMX)
+ if (ARCH_X86)
ff_vp8dsp_init_x86(dsp);
if (HAVE_ALTIVEC)
ff_vp8dsp_init_altivec(dsp);
+ if (ARCH_ARM)
+ ff_vp8dsp_init_arm(dsp);
}