*
*/
+#include "libavutil/common.h"
+#include "h264chroma.h"
#include "vc1dsp.h"
}
}
+static void vc1_v_s_overlap_c(int16_t *top, int16_t *bottom)
+{
+ int i;
+ int a, b, c, d;
+ int d1, d2;
+ int rnd1 = 4, rnd2 = 3;
+ for(i = 0; i < 8; i++) {
+ a = top[48];
+ b = top[56];
+ c = bottom[0];
+ d = bottom[8];
+ d1 = a - d;
+ d2 = a - d + b - c;
+
+ top[48] = ((a << 3) - d1 + rnd1) >> 3;
+ top[56] = ((b << 3) - d2 + rnd2) >> 3;
+ bottom[0] = ((c << 3) + d2 + rnd1) >> 3;
+ bottom[8] = ((d << 3) + d1 + rnd2) >> 3;
+
+ bottom++;
+ top++;
+ rnd2 = 7 - rnd2;
+ rnd1 = 7 - rnd1;
+ }
+}
+
+static void vc1_h_s_overlap_c(int16_t *left, int16_t *right)
+{
+ int i;
+ int a, b, c, d;
+ int d1, d2;
+ int rnd1 = 4, rnd2 = 3;
+ for(i = 0; i < 8; i++) {
+ a = left[6];
+ b = left[7];
+ c = right[0];
+ d = right[1];
+ d1 = a - d;
+ d2 = a - d + b - c;
+
+ left[6] = ((a << 3) - d1 + rnd1) >> 3;
+ left[7] = ((b << 3) - d2 + rnd2) >> 3;
+ right[0] = ((c << 3) + d2 + rnd1) >> 3;
+ right[1] = ((d << 3) + d1 + rnd2) >> 3;
+
+ right += 8;
+ left += 8;
+ rnd2 = 7 - rnd2;
+ rnd1 = 7 - rnd1;
+ }
+}
+
/**
* VC-1 in-loop deblocking filter for one line
* @param src source block type
* @see 8.6
*/
static av_always_inline int vc1_filter_line(uint8_t* src, int stride, int pq){
- uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
-
int a0 = (2*(src[-2*stride] - src[ 1*stride]) - 5*(src[-1*stride] - src[ 0*stride]) + 4) >> 3;
int a0_sign = a0 >> 31; /* Store sign */
a0 = (a0 ^ a0_sign) - a0_sign; /* a0 = FFABS(a0); */
else{
d = FFMIN(d, clip);
d = (d ^ d_sign) - d_sign; /* Restore sign */
- src[-1*stride] = cm[src[-1*stride] - d];
- src[ 0*stride] = cm[src[ 0*stride] + d];
+ src[-1*stride] = av_clip_uint8(src[-1*stride] - d);
+ src[ 0*stride] = av_clip_uint8(src[ 0*stride] + d);
}
return 1;
}
/** Do inverse transform on 8x8 block
*/
-static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
int dc = block[0];
- const uint8_t *cm;
dc = (3 * dc + 1) >> 1;
dc = (3 * dc + 16) >> 5;
- cm = ff_cropTbl + MAX_NEG_CROP + dc;
for(i = 0; i < 8; i++){
- dest[0] = cm[dest[0]];
- dest[1] = cm[dest[1]];
- dest[2] = cm[dest[2]];
- dest[3] = cm[dest[3]];
- dest[4] = cm[dest[4]];
- dest[5] = cm[dest[5]];
- dest[6] = cm[dest[6]];
- dest[7] = cm[dest[7]];
+ dest[0] = av_clip_uint8(dest[0] + dc);
+ dest[1] = av_clip_uint8(dest[1] + dc);
+ dest[2] = av_clip_uint8(dest[2] + dc);
+ dest[3] = av_clip_uint8(dest[3] + dc);
+ dest[4] = av_clip_uint8(dest[4] + dc);
+ dest[5] = av_clip_uint8(dest[5] + dc);
+ dest[6] = av_clip_uint8(dest[6] + dc);
+ dest[7] = av_clip_uint8(dest[7] + dc);
dest += linesize;
}
}
-static av_always_inline void vc1_inv_trans_8x8_c(DCTELEM block[64], int shl, int sub)
+static void vc1_inv_trans_8x8_c(int16_t block[64])
{
int i;
register int t1,t2,t3,t4,t5,t6,t7,t8;
- DCTELEM *src, *dst, temp[64];
+ int16_t *src, *dst, temp[64];
src = block;
dst = temp;
t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56];
t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56];
- dst[ 0] = (((t5 + t1 ) >> 7) - sub) << shl;
- dst[ 8] = (((t6 + t2 ) >> 7) - sub) << shl;
- dst[16] = (((t7 + t3 ) >> 7) - sub) << shl;
- dst[24] = (((t8 + t4 ) >> 7) - sub) << shl;
- dst[32] = (((t8 - t4 + 1) >> 7) - sub) << shl;
- dst[40] = (((t7 - t3 + 1) >> 7) - sub) << shl;
- dst[48] = (((t6 - t2 + 1) >> 7) - sub) << shl;
- dst[56] = (((t5 - t1 + 1) >> 7) - sub) << shl;
+ dst[ 0] = (t5 + t1) >> 7;
+ dst[ 8] = (t6 + t2) >> 7;
+ dst[16] = (t7 + t3) >> 7;
+ dst[24] = (t8 + t4) >> 7;
+ dst[32] = (t8 - t4 + 1) >> 7;
+ dst[40] = (t7 - t3 + 1) >> 7;
+ dst[48] = (t6 - t2 + 1) >> 7;
+ dst[56] = (t5 - t1 + 1) >> 7;
src++;
dst++;
}
}
-static void vc1_inv_trans_8x8_add_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
- vc1_inv_trans_8x8_c(block, 0, 0);
- ff_add_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_signed_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
- vc1_inv_trans_8x8_c(block, 0, 0);
- ff_put_signed_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_signed_rangered_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
- vc1_inv_trans_8x8_c(block, 1, 0);
- ff_put_signed_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
- vc1_inv_trans_8x8_c(block, 0, 0);
- ff_put_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_rangered_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
- vc1_inv_trans_8x8_c(block, 1, 64);
- ff_put_pixels_clamped_c(block, dest, linesize);
-}
-
/** Do inverse transform on 8x4 part of block
*/
-static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
int dc = block[0];
- const uint8_t *cm;
dc = ( 3 * dc + 1) >> 1;
dc = (17 * dc + 64) >> 7;
- cm = ff_cropTbl + MAX_NEG_CROP + dc;
for(i = 0; i < 4; i++){
- dest[0] = cm[dest[0]];
- dest[1] = cm[dest[1]];
- dest[2] = cm[dest[2]];
- dest[3] = cm[dest[3]];
- dest[4] = cm[dest[4]];
- dest[5] = cm[dest[5]];
- dest[6] = cm[dest[6]];
- dest[7] = cm[dest[7]];
+ dest[0] = av_clip_uint8(dest[0] + dc);
+ dest[1] = av_clip_uint8(dest[1] + dc);
+ dest[2] = av_clip_uint8(dest[2] + dc);
+ dest[3] = av_clip_uint8(dest[3] + dc);
+ dest[4] = av_clip_uint8(dest[4] + dc);
+ dest[5] = av_clip_uint8(dest[5] + dc);
+ dest[6] = av_clip_uint8(dest[6] + dc);
+ dest[7] = av_clip_uint8(dest[7] + dc);
dest += linesize;
}
}
-static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
register int t1,t2,t3,t4,t5,t6,t7,t8;
- DCTELEM *src, *dst;
- const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+ int16_t *src, *dst;
src = block;
dst = block;
t3 = 22 * src[ 8] + 10 * src[24];
t4 = 22 * src[24] - 10 * src[ 8];
- dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)];
- dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)];
- dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)];
- dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)];
+ dest[0*linesize] = av_clip_uint8(dest[0*linesize] + ((t1 + t3) >> 7));
+ dest[1*linesize] = av_clip_uint8(dest[1*linesize] + ((t2 - t4) >> 7));
+ dest[2*linesize] = av_clip_uint8(dest[2*linesize] + ((t2 + t4) >> 7));
+ dest[3*linesize] = av_clip_uint8(dest[3*linesize] + ((t1 - t3) >> 7));
src ++;
dest++;
/** Do inverse transform on 4x8 parts of block
*/
-static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
int dc = block[0];
- const uint8_t *cm;
dc = (17 * dc + 4) >> 3;
dc = (12 * dc + 64) >> 7;
- cm = ff_cropTbl + MAX_NEG_CROP + dc;
for(i = 0; i < 8; i++){
- dest[0] = cm[dest[0]];
- dest[1] = cm[dest[1]];
- dest[2] = cm[dest[2]];
- dest[3] = cm[dest[3]];
+ dest[0] = av_clip_uint8(dest[0] + dc);
+ dest[1] = av_clip_uint8(dest[1] + dc);
+ dest[2] = av_clip_uint8(dest[2] + dc);
+ dest[3] = av_clip_uint8(dest[3] + dc);
dest += linesize;
}
}
-static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
register int t1,t2,t3,t4,t5,t6,t7,t8;
- DCTELEM *src, *dst;
- const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+ int16_t *src, *dst;
src = block;
dst = block;
t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56];
t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56];
- dest[0*linesize] = cm[dest[0*linesize] + ((t5 + t1) >> 7)];
- dest[1*linesize] = cm[dest[1*linesize] + ((t6 + t2) >> 7)];
- dest[2*linesize] = cm[dest[2*linesize] + ((t7 + t3) >> 7)];
- dest[3*linesize] = cm[dest[3*linesize] + ((t8 + t4) >> 7)];
- dest[4*linesize] = cm[dest[4*linesize] + ((t8 - t4 + 1) >> 7)];
- dest[5*linesize] = cm[dest[5*linesize] + ((t7 - t3 + 1) >> 7)];
- dest[6*linesize] = cm[dest[6*linesize] + ((t6 - t2 + 1) >> 7)];
- dest[7*linesize] = cm[dest[7*linesize] + ((t5 - t1 + 1) >> 7)];
+ dest[0*linesize] = av_clip_uint8(dest[0*linesize] + ((t5 + t1) >> 7));
+ dest[1*linesize] = av_clip_uint8(dest[1*linesize] + ((t6 + t2) >> 7));
+ dest[2*linesize] = av_clip_uint8(dest[2*linesize] + ((t7 + t3) >> 7));
+ dest[3*linesize] = av_clip_uint8(dest[3*linesize] + ((t8 + t4) >> 7));
+ dest[4*linesize] = av_clip_uint8(dest[4*linesize] + ((t8 - t4 + 1) >> 7));
+ dest[5*linesize] = av_clip_uint8(dest[5*linesize] + ((t7 - t3 + 1) >> 7));
+ dest[6*linesize] = av_clip_uint8(dest[6*linesize] + ((t6 - t2 + 1) >> 7));
+ dest[7*linesize] = av_clip_uint8(dest[7*linesize] + ((t5 - t1 + 1) >> 7));
src ++;
dest++;
/** Do inverse transform on 4x4 part of block
*/
-static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
int dc = block[0];
- const uint8_t *cm;
dc = (17 * dc + 4) >> 3;
dc = (17 * dc + 64) >> 7;
- cm = ff_cropTbl + MAX_NEG_CROP + dc;
for(i = 0; i < 4; i++){
- dest[0] = cm[dest[0]];
- dest[1] = cm[dest[1]];
- dest[2] = cm[dest[2]];
- dest[3] = cm[dest[3]];
+ dest[0] = av_clip_uint8(dest[0] + dc);
+ dest[1] = av_clip_uint8(dest[1] + dc);
+ dest[2] = av_clip_uint8(dest[2] + dc);
+ dest[3] = av_clip_uint8(dest[3] + dc);
dest += linesize;
}
}
-static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, int16_t *block)
{
int i;
register int t1,t2,t3,t4;
- DCTELEM *src, *dst;
- const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
+ int16_t *src, *dst;
src = block;
dst = block;
t3 = 22 * src[ 8] + 10 * src[24];
t4 = 22 * src[24] - 10 * src[ 8];
- dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)];
- dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)];
- dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)];
- dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)];
+ dest[0*linesize] = av_clip_uint8(dest[0*linesize] + ((t1 + t3) >> 7));
+ dest[1*linesize] = av_clip_uint8(dest[1*linesize] + ((t2 - t4) >> 7));
+ dest[2*linesize] = av_clip_uint8(dest[2*linesize] + ((t2 + t4) >> 7));
+ dest[3*linesize] = av_clip_uint8(dest[3*linesize] + ((t1 - t3) >> 7));
src ++;
dest++;
return 0; /* should not occur */ \
}
-VC1_MSPEL_FILTER_16B(ver, uint8_t);
-VC1_MSPEL_FILTER_16B(hor, int16_t);
+VC1_MSPEL_FILTER_16B(ver, uint8_t)
+VC1_MSPEL_FILTER_16B(hor, int16_t)
/** Filter used to interpolate fractional pel values
/** Function used to do motion compensation with bicubic interpolation
*/
#define VC1_MSPEL_MC(OP, OPNAME)\
-static void OPNAME ## vc1_mspel_mc(uint8_t *dst, const uint8_t *src, int stride, int hmode, int vmode, int rnd)\
+static av_always_inline void OPNAME ## vc1_mspel_mc(uint8_t *dst, const uint8_t *src, int stride, int hmode, int vmode, int rnd)\
{\
int i, j;\
\
/* pixel functions - really are entry points to vc1_mspel_mc */
#define PUT_VC1_MSPEL(a, b)\
-static void put_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
- put_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
-}\
-static void avg_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
- avg_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
+static void put_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, \
+ const uint8_t *src, \
+ ptrdiff_t stride, int rnd) \
+{ \
+ put_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
+} \
+static void avg_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, \
+ const uint8_t *src, \
+ ptrdiff_t stride, int rnd) \
+{ \
+ avg_vc1_mspel_mc(dst, src, stride, a, b, rnd); \
}
PUT_VC1_MSPEL(1, 0)
}
}
+static void put_no_rnd_vc1_chroma_mc4_c(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y){
+ const int A=(8-x)*(8-y);
+ const int B=( x)*(8-y);
+ const int C=(8-x)*( y);
+ const int D=( x)*( y);
+ int i;
+
+ assert(x<8 && y<8 && x>=0 && y>=0);
+
+ for(i=0; i<h; i++)
+ {
+ dst[0] = (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6;
+ dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6;
+ dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6;
+ dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6;
+ dst+= stride;
+ src+= stride;
+ }
+}
+
#define avg2(a,b) ((a+b+1)>>1)
static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
const int A=(8-x)*(8-y);
}
}
+#if CONFIG_WMV3IMAGE_DECODER || CONFIG_VC1IMAGE_DECODER
+
+static void sprite_h_c(uint8_t *dst, const uint8_t *src, int offset, int advance, int count)
+{
+ while (count--) {
+ int a = src[(offset >> 16) ];
+ int b = src[(offset >> 16) + 1];
+ *dst++ = a + ((b - a) * (offset&0xFFFF) >> 16);
+ offset += advance;
+ }
+}
+
+static av_always_inline void sprite_v_template(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset1,
+ int two_sprites, const uint8_t *src2a, const uint8_t *src2b, int offset2,
+ int alpha, int scaled, int width)
+{
+ int a1, b1, a2, b2;
+ while (width--) {
+ a1 = *src1a++;
+ if (scaled) {
+ b1 = *src1b++;
+ a1 = a1 + ((b1 - a1) * offset1 >> 16);
+ }
+ if (two_sprites) {
+ a2 = *src2a++;
+ if (scaled > 1) {
+ b2 = *src2b++;
+ a2 = a2 + ((b2 - a2) * offset2 >> 16);
+ }
+ a1 = a1 + ((a2 - a1) * alpha >> 16);
+ }
+ *dst++ = a1;
+ }
+}
+
+static void sprite_v_single_c(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset, int width)
+{
+ sprite_v_template(dst, src1a, src1b, offset, 0, NULL, NULL, 0, 0, 1, width);
+}
+
+static void sprite_v_double_noscale_c(uint8_t *dst, const uint8_t *src1a, const uint8_t *src2a, int alpha, int width)
+{
+ sprite_v_template(dst, src1a, NULL, 0, 1, src2a, NULL, 0, alpha, 0, width);
+}
+
+static void sprite_v_double_onescale_c(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset1,
+ const uint8_t *src2a, int alpha, int width)
+{
+ sprite_v_template(dst, src1a, src1b, offset1, 1, src2a, NULL, 0, alpha, 1, width);
+}
+
+static void sprite_v_double_twoscale_c(uint8_t *dst, const uint8_t *src1a, const uint8_t *src1b, int offset1,
+ const uint8_t *src2a, const uint8_t *src2b, int offset2,
+ int alpha, int width)
+{
+ sprite_v_template(dst, src1a, src1b, offset1, 1, src2a, src2b, offset2, alpha, 2, width);
+}
+
+#endif
+
av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
- dsp->vc1_inv_trans_8x8_add = vc1_inv_trans_8x8_add_c;
- dsp->vc1_inv_trans_8x8_put_signed[0] = vc1_inv_trans_8x8_put_signed_c;
- dsp->vc1_inv_trans_8x8_put_signed[1] = vc1_inv_trans_8x8_put_signed_rangered_c;
- dsp->vc1_inv_trans_8x8_put[0] = vc1_inv_trans_8x8_put_c;
- dsp->vc1_inv_trans_8x8_put[1] = vc1_inv_trans_8x8_put_rangered_c;
+ dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_c;
dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c;
dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c;
dsp->vc1_inv_trans_4x4 = vc1_inv_trans_4x4_c;
dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
dsp->vc1_h_overlap = vc1_h_overlap_c;
dsp->vc1_v_overlap = vc1_v_overlap_c;
+ dsp->vc1_h_s_overlap = vc1_h_s_overlap_c;
+ dsp->vc1_v_s_overlap = vc1_v_s_overlap_c;
dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
dsp->put_no_rnd_vc1_chroma_pixels_tab[0]= put_no_rnd_vc1_chroma_mc8_c;
dsp->avg_no_rnd_vc1_chroma_pixels_tab[0]= avg_no_rnd_vc1_chroma_mc8_c;
-
- if (HAVE_ALTIVEC)
- ff_vc1dsp_init_altivec(dsp);
- if (HAVE_MMX)
- ff_vc1dsp_init_mmx(dsp);
+ dsp->put_no_rnd_vc1_chroma_pixels_tab[1] = put_no_rnd_vc1_chroma_mc4_c;
+
+#if CONFIG_WMV3IMAGE_DECODER || CONFIG_VC1IMAGE_DECODER
+ dsp->sprite_h = sprite_h_c;
+ dsp->sprite_v_single = sprite_v_single_c;
+ dsp->sprite_v_double_noscale = sprite_v_double_noscale_c;
+ dsp->sprite_v_double_onescale = sprite_v_double_onescale_c;
+ dsp->sprite_v_double_twoscale = sprite_v_double_twoscale_c;
+#endif
+
+ if (ARCH_X86)
+ ff_vc1dsp_init_x86(dsp);
+ if (ARCH_PPC)
+ ff_vc1dsp_init_ppc(dsp);
}