Skip to content

F
ffmpeg.wasm-core
Commit a84616b7 authored by Luca Barbato's avatar Luca Barbato
Browse files
Options

mpegvideo: K&R formatting cosmetics

parent 93f30547
Hide whitespace changes
Inline Side-by-side
Showing with 516 additions and 449 deletions
libavcodec/mpegvideo_motion.c
View file @ a84616b7
...@@ -39,83 +39,98 @@ static void gmc1_motion(MpegEncContext *s, ...@@ -39,83 +39,98 @@ static void gmc1_motion(MpegEncContext *s,
uint8_t *ptr; uint8_t *ptr;
int src_x, src_y, motion_x, motion_y; int src_x, src_y, motion_x, motion_y;
ptrdiff_t offset, linesize, uvlinesize; ptrdiff_t offset, linesize, uvlinesize;
int emu=0; int emu = 0;
motion_x= s->sprite_offset[0][0]; motion_x = s->sprite_offset[0][0];
motion_y= s->sprite_offset[0][1]; motion_y = s->sprite_offset[0][1];
src_x = s->mb_x * 16 + (motion_x >> (s->sprite_warping_accuracy+1)); src_x = s->mb_x * 16 + (motion_x >> (s->sprite_warping_accuracy + 1));
src_y = s->mb_y * 16 + (motion_y >> (s->sprite_warping_accuracy+1)); src_y = s->mb_y * 16 + (motion_y >> (s->sprite_warping_accuracy + 1));
motion_x<<=(3-s->sprite_warping_accuracy); motion_x <<= (3 - s->sprite_warping_accuracy);
motion_y<<=(3-s->sprite_warping_accuracy); motion_y <<= (3 - s->sprite_warping_accuracy);
src_x = av_clip(src_x, -16, s->width); src_x = av_clip(src_x, -16, s->width);
if (src_x == s->width) if (src_x == s->width)
motion_x =0; motion_x = 0;
src_y = av_clip(src_y, -16, s->height); src_y = av_clip(src_y, -16, s->height);
if (src_y == s->height) if (src_y == s->height)
motion_y =0; motion_y = 0;
linesize = s->linesize; linesize = s->linesize;
uvlinesize = s->uvlinesize; uvlinesize = s->uvlinesize;
ptr = ref_picture[0] + (src_y * linesize) + src_x; ptr = ref_picture[0] + src_y * linesize + src_x;
if(s->flags&CODEC_FLAG_EMU_EDGE){ if (s->flags & CODEC_FLAG_EMU_EDGE) {
if( (unsigned)src_x >= FFMAX(s->h_edge_pos - 17, 0) if ((unsigned)src_x >= FFMAX(s->h_edge_pos - 17, 0) ||
|| (unsigned)src_y >= FFMAX(s->v_edge_pos - 17, 0)){ (unsigned)src_y >= FFMAX(s->v_edge_pos - 17, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, linesize, 17, 17, src_x, src_y, s->h_edge_pos, s->v_edge_pos); s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr,
ptr= s->edge_emu_buffer; linesize,
17, 17,
src_x, src_y,
s->h_edge_pos, s->v_edge_pos);
ptr = s->edge_emu_buffer;
} }
} }
if((motion_x|motion_y)&7){ if ((motion_x | motion_y) & 7) {
s->dsp.gmc1(dest_y , ptr , linesize, 16, motion_x&15, motion_y&15, 128 - s->no_rounding); s->dsp.gmc1(dest_y, ptr, linesize, 16,
s->dsp.gmc1(dest_y+8, ptr+8, linesize, 16, motion_x&15, motion_y&15, 128 - s->no_rounding); motion_x & 15, motion_y & 15, 128 - s->no_rounding);
}else{ s->dsp.gmc1(dest_y + 8, ptr + 8, linesize, 16,
motion_x & 15, motion_y & 15, 128 - s->no_rounding);
} else {
int dxy; int dxy;
dxy= ((motion_x>>3)&1) | ((motion_y>>2)&2); dxy = ((motion_x >> 3) & 1) | ((motion_y >> 2) & 2);
if (s->no_rounding){ if (s->no_rounding) {
s->hdsp.put_no_rnd_pixels_tab[0][dxy](dest_y, ptr, linesize, 16); s->hdsp.put_no_rnd_pixels_tab[0][dxy](dest_y, ptr, linesize, 16);
}else{ } else {
s->hdsp.put_pixels_tab [0][dxy](dest_y, ptr, linesize, 16); s->hdsp.put_pixels_tab[0][dxy](dest_y, ptr, linesize, 16);
} }
} }
if(CONFIG_GRAY && s->flags&CODEC_FLAG_GRAY) return; if (CONFIG_GRAY && s->flags & CODEC_FLAG_GRAY)
return;
motion_x= s->sprite_offset[1][0];
motion_y= s->sprite_offset[1][1]; motion_x = s->sprite_offset[1][0];
src_x = s->mb_x * 8 + (motion_x >> (s->sprite_warping_accuracy+1)); motion_y = s->sprite_offset[1][1];
src_y = s->mb_y * 8 + (motion_y >> (s->sprite_warping_accuracy+1)); src_x = s->mb_x * 8 + (motion_x >> (s->sprite_warping_accuracy + 1));
motion_x<<=(3-s->sprite_warping_accuracy); src_y = s->mb_y * 8 + (motion_y >> (s->sprite_warping_accuracy + 1));
motion_y<<=(3-s->sprite_warping_accuracy); motion_x <<= (3 - s->sprite_warping_accuracy);
src_x = av_clip(src_x, -8, s->width>>1); motion_y <<= (3 - s->sprite_warping_accuracy);
if (src_x == s->width>>1) src_x = av_clip(src_x, -8, s->width >> 1);
motion_x =0; if (src_x == s->width >> 1)
src_y = av_clip(src_y, -8, s->height>>1); motion_x = 0;
if (src_y == s->height>>1) src_y = av_clip(src_y, -8, s->height >> 1);
motion_y =0; if (src_y == s->height >> 1)
motion_y = 0;
offset = (src_y * uvlinesize) + src_x; offset = (src_y * uvlinesize) + src_x;
ptr = ref_picture[1] + offset; ptr = ref_picture[1] + offset;
if(s->flags&CODEC_FLAG_EMU_EDGE){ if (s->flags & CODEC_FLAG_EMU_EDGE) {
if( (unsigned)src_x >= FFMAX((s->h_edge_pos>>1) - 9, 0) if ((unsigned)src_x >= FFMAX((s->h_edge_pos >> 1) - 9, 0) ||
|| (unsigned)src_y >= FFMAX((s->v_edge_pos>>1) - 9, 0)){ (unsigned)src_y >= FFMAX((s->v_edge_pos >> 1) - 9, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1); s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr,
ptr= s->edge_emu_buffer; uvlinesize,
emu=1; 9, 9,
src_x, src_y,
s->h_edge_pos >> 1, s->v_edge_pos >> 1);
ptr = s->edge_emu_buffer;
emu = 1;
} }
} }
s->dsp.gmc1(dest_cb, ptr, uvlinesize, 8, motion_x&15, motion_y&15, 128 - s->no_rounding); s->dsp.gmc1(dest_cb, ptr, uvlinesize, 8,
motion_x & 15, motion_y & 15, 128 - s->no_rounding);
ptr = ref_picture[2] + offset; ptr = ref_picture[2] + offset;
if(emu){ if (emu) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, uvlinesize, 9, 9, src_x, src_y, s->h_edge_pos>>1, s->v_edge_pos>>1); s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr,
ptr= s->edge_emu_buffer; uvlinesize,
9, 9,
src_x, src_y,
s->h_edge_pos >> 1, s->v_edge_pos >> 1);
ptr = s->edge_emu_buffer;
} }
s->dsp.gmc1(dest_cr, ptr, uvlinesize, 8, motion_x&15, motion_y&15, 128 - s->no_rounding); s->dsp.gmc1(dest_cr, ptr, uvlinesize, 8,
motion_x & 15, motion_y & 15, 128 - s->no_rounding);
return;
} }
static void gmc_motion(MpegEncContext *s, static void gmc_motion(MpegEncContext *s,
...@@ -124,70 +139,72 @@ static void gmc_motion(MpegEncContext *s, ...@@ -124,70 +139,72 @@ static void gmc_motion(MpegEncContext *s,
{ {
uint8_t *ptr; uint8_t *ptr;
int linesize, uvlinesize; int linesize, uvlinesize;
const int a= s->sprite_warping_accuracy; const int a = s->sprite_warping_accuracy;
int ox, oy; int ox, oy;
linesize = s->linesize; linesize = s->linesize;
uvlinesize = s->uvlinesize; uvlinesize = s->uvlinesize;
ptr = ref_picture[0]; ptr = ref_picture[0];
ox= s->sprite_offset[0][0] + s->sprite_delta[0][0]*s->mb_x*16 + s->sprite_delta[0][1]*s->mb_y*16; ox = s->sprite_offset[0][0] + s->sprite_delta[0][0] * s->mb_x * 16 +
oy= s->sprite_offset[0][1] + s->sprite_delta[1][0]*s->mb_x*16 + s->sprite_delta[1][1]*s->mb_y*16; s->sprite_delta[0][1] * s->mb_y * 16;
oy = s->sprite_offset[0][1] + s->sprite_delta[1][0] * s->mb_x * 16 +
s->sprite_delta[1][1] * s->mb_y * 16;
s->dsp.gmc(dest_y, ptr, linesize, 16, s->dsp.gmc(dest_y, ptr, linesize, 16,
ox, ox, oy,
oy, s->sprite_delta[0][0], s->sprite_delta[0][1],
s->sprite_delta[0][0], s->sprite_delta[0][1], s->sprite_delta[1][0], s->sprite_delta[1][1],
s->sprite_delta[1][0], s->sprite_delta[1][1], a + 1, (1 << (2 * a + 1)) - s->no_rounding,
a+1, (1<<(2*a+1)) - s->no_rounding, s->h_edge_pos, s->v_edge_pos);
s->h_edge_pos, s->v_edge_pos); s->dsp.gmc(dest_y + 8, ptr, linesize, 16,
s->dsp.gmc(dest_y+8, ptr, linesize, 16, ox + s->sprite_delta[0][0] * 8,
ox + s->sprite_delta[0][0]*8, oy + s->sprite_delta[1][0] * 8,
oy + s->sprite_delta[1][0]*8, s->sprite_delta[0][0], s->sprite_delta[0][1],
s->sprite_delta[0][0], s->sprite_delta[0][1], s->sprite_delta[1][0], s->sprite_delta[1][1],
s->sprite_delta[1][0], s->sprite_delta[1][1], a + 1, (1 << (2 * a + 1)) - s->no_rounding,
a+1, (1<<(2*a+1)) - s->no_rounding, s->h_edge_pos, s->v_edge_pos);
s->h_edge_pos, s->v_edge_pos);
if (CONFIG_GRAY && s->flags & CODEC_FLAG_GRAY)
if(CONFIG_GRAY && s->flags&CODEC_FLAG_GRAY) return; return;
ox= s->sprite_offset[1][0] + s->sprite_delta[0][0]*s->mb_x*8 + s->sprite_delta[0][1]*s->mb_y*8; ox = s->sprite_offset[1][0] + s->sprite_delta[0][0] * s->mb_x * 8 +
oy= s->sprite_offset[1][1] + s->sprite_delta[1][0]*s->mb_x*8 + s->sprite_delta[1][1]*s->mb_y*8; s->sprite_delta[0][1] * s->mb_y * 8;
oy = s->sprite_offset[1][1] + s->sprite_delta[1][0] * s->mb_x * 8 +
s->sprite_delta[1][1] * s->mb_y * 8;
ptr = ref_picture[1]; ptr = ref_picture[1];
s->dsp.gmc(dest_cb, ptr, uvlinesize, 8, s->dsp.gmc(dest_cb, ptr, uvlinesize, 8,
ox, ox, oy,
oy, s->sprite_delta[0][0], s->sprite_delta[0][1],
s->sprite_delta[0][0], s->sprite_delta[0][1], s->sprite_delta[1][0], s->sprite_delta[1][1],
s->sprite_delta[1][0], s->sprite_delta[1][1], a + 1, (1 << (2 * a + 1)) - s->no_rounding,
a+1, (1<<(2*a+1)) - s->no_rounding, s->h_edge_pos >> 1, s->v_edge_pos >> 1);
s->h_edge_pos>>1, s->v_edge_pos>>1);
ptr = ref_picture[2]; ptr = ref_picture[2];
s->dsp.gmc(dest_cr, ptr, uvlinesize, 8, s->dsp.gmc(dest_cr, ptr, uvlinesize, 8,
ox, ox, oy,
oy, s->sprite_delta[0][0], s->sprite_delta[0][1],
s->sprite_delta[0][0], s->sprite_delta[0][1], s->sprite_delta[1][0], s->sprite_delta[1][1],
s->sprite_delta[1][0], s->sprite_delta[1][1], a + 1, (1 << (2 * a + 1)) - s->no_rounding,
a+1, (1<<(2*a+1)) - s->no_rounding, s->h_edge_pos >> 1, s->v_edge_pos >> 1);
s->h_edge_pos>>1, s->v_edge_pos>>1);
} }
static inline int hpel_motion(MpegEncContext *s, static inline int hpel_motion(MpegEncContext *s,
uint8_t *dest, uint8_t *src, uint8_t *dest, uint8_t *src,
int src_x, int src_y, int src_x, int src_y,
op_pixels_func *pix_op, op_pixels_func *pix_op,
int motion_x, int motion_y) int motion_x, int motion_y)
{ {
int dxy = 0; int dxy = 0;
int emu=0; int emu = 0;
src_x += motion_x >> 1; src_x += motion_x >> 1;
src_y += motion_y >> 1; src_y += motion_y >> 1;
/* WARNING: do no forget half pels */ /* WARNING: do no forget half pels */
src_x = av_clip(src_x, -16, s->width); //FIXME unneeded for emu? src_x = av_clip(src_x, -16, s->width); // FIXME unneeded for emu?
if (src_x != s->width) if (src_x != s->width)
dxy |= motion_x & 1; dxy |= motion_x & 1;
src_y = av_clip(src_y, -16, s->height); src_y = av_clip(src_y, -16, s->height);
...@@ -195,13 +212,16 @@ static inline int hpel_motion(MpegEncContext *s, ...@@ -195,13 +212,16 @@ static inline int hpel_motion(MpegEncContext *s,
dxy |= (motion_y & 1) << 1; dxy |= (motion_y & 1) << 1;
src += src_y * s->linesize + src_x; src += src_y * s->linesize + src_x;
if(s->unrestricted_mv && (s->flags&CODEC_FLAG_EMU_EDGE)){ if (s->unrestricted_mv && (s->flags & CODEC_FLAG_EMU_EDGE)) {
if( (unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x&1) - 8, 0) if ((unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x & 1) - 8, 0) ||
|| (unsigned)src_y > FFMAX(s->v_edge_pos - (motion_y&1) - 8, 0)){ (unsigned)src_y > FFMAX(s->v_edge_pos - (motion_y & 1) - 8, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, src, s->linesize, 9, 9, s->vdsp.emulated_edge_mc(s->edge_emu_buffer, src,
src_x, src_y, s->h_edge_pos, s->v_edge_pos); s->linesize,
src= s->edge_emu_buffer; 9, 9,
emu=1; src_x, src_y, s->h_edge_pos,
s->v_edge_pos);
src = s->edge_emu_buffer;
emu = 1;
} }
} }
pix_op[dxy](dest, src, s->linesize, 8); pix_op[dxy](dest, src, s->linesize, 8);
...@@ -210,10 +230,19 @@ static inline int hpel_motion(MpegEncContext *s, ...@@ -210,10 +230,19 @@ static inline int hpel_motion(MpegEncContext *s,
static av_always_inline static av_always_inline
void mpeg_motion_internal(MpegEncContext *s, void mpeg_motion_internal(MpegEncContext *s,
uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr, uint8_t *dest_y,
int field_based, int bottom_field, int field_select, uint8_t *dest_cb,
uint8_t **ref_picture, op_pixels_func (*pix_op)[4], uint8_t *dest_cr,
int motion_x, int motion_y, int h, int is_mpeg12, int mb_y) int field_based,
int bottom_field,
int field_select,
uint8_t **ref_picture,
op_pixels_func (*pix_op)[4],
int motion_x,
int motion_y,
int h,
int is_mpeg12,
int mb_y)
{ {
uint8_t *ptr_y, *ptr_cb, *ptr_cr; uint8_t *ptr_y, *ptr_cb, *ptr_cr;
int dxy, uvdxy, mx, my, src_x, src_y, int dxy, uvdxy, mx, my, src_x, src_y,
...@@ -221,56 +250,56 @@ void mpeg_motion_internal(MpegEncContext *s, ...@@ -221,56 +250,56 @@ void mpeg_motion_internal(MpegEncContext *s,
ptrdiff_t uvlinesize, linesize; ptrdiff_t uvlinesize, linesize;
#if 0 #if 0
if(s->quarter_sample) if (s->quarter_sample) {
{ motion_x >>= 1;
motion_x>>=1; motion_y >>= 1;
motion_y>>=1; }
}
#endif #endif
v_edge_pos = s->v_edge_pos >> field_based; v_edge_pos = s->v_edge_pos >> field_based;
linesize = s->current_picture.f.linesize[0] << field_based; linesize = s->current_picture.f.linesize[0] << field_based;
uvlinesize = s->current_picture.f.linesize[1] << field_based; uvlinesize = s->current_picture.f.linesize[1] << field_based;
dxy = ((motion_y & 1) << 1) | (motion_x & 1); dxy = ((motion_y & 1) << 1) | (motion_x & 1);
src_x = s->mb_x* 16 + (motion_x >> 1); src_x = s->mb_x * 16 + (motion_x >> 1);
src_y =( mb_y<<(4-field_based)) + (motion_y >> 1); src_y = (mb_y << (4 - field_based)) + (motion_y >> 1);
if (!is_mpeg12 && s->out_format == FMT_H263) { if (!is_mpeg12 && s->out_format == FMT_H263) {
if((s->workaround_bugs & FF_BUG_HPEL_CHROMA) && field_based){ if ((s->workaround_bugs & FF_BUG_HPEL_CHROMA) && field_based) {
mx = (motion_x>>1)|(motion_x&1); mx = (motion_x >> 1) | (motion_x & 1);
my = motion_y >>1; my = motion_y >> 1;
uvdxy = ((my & 1) << 1) | (mx & 1); uvdxy = ((my & 1) << 1) | (mx & 1);
uvsrc_x = s->mb_x* 8 + (mx >> 1); uvsrc_x = s->mb_x * 8 + (mx >> 1);
uvsrc_y =( mb_y<<(3-field_based))+ (my >> 1); uvsrc_y = (mb_y << (3 - field_based)) + (my >> 1);
}else{ } else {
uvdxy = dxy | (motion_y & 2) | ((motion_x & 2) >> 1); uvdxy = dxy | (motion_y & 2) | ((motion_x & 2) >> 1);
uvsrc_x = src_x>>1; uvsrc_x = src_x >> 1;
uvsrc_y = src_y>>1; uvsrc_y = src_y >> 1;
} }
}else if(!is_mpeg12 && s->out_format == FMT_H261){//even chroma mv's are full pel in H261 // Even chroma mv's are full pel in H261
mx = motion_x / 4; } else if (!is_mpeg12 && s->out_format == FMT_H261) {
my = motion_y / 4; mx = motion_x / 4;
uvdxy = 0; my = motion_y / 4;
uvsrc_x = s->mb_x*8 + mx; uvdxy = 0;
uvsrc_y = mb_y*8 + my; uvsrc_x = s->mb_x * 8 + mx;
uvsrc_y = mb_y * 8 + my;
} else { } else {
if(s->chroma_y_shift){ if (s->chroma_y_shift) {
mx = motion_x / 2; mx = motion_x / 2;
my = motion_y / 2; my = motion_y / 2;
uvdxy = ((my & 1) << 1) | (mx & 1); uvdxy = ((my & 1) << 1) | (mx & 1);
uvsrc_x = s->mb_x* 8 + (mx >> 1); uvsrc_x = s->mb_x * 8 + (mx >> 1);
uvsrc_y =( mb_y<<(3-field_based))+ (my >> 1); uvsrc_y = (mb_y << (3 - field_based)) + (my >> 1);
} else { } else {
if(s->chroma_x_shift){ if (s->chroma_x_shift) {
//Chroma422 // Chroma422
mx = motion_x / 2; mx = motion_x / 2;
uvdxy = ((motion_y & 1) << 1) | (mx & 1); uvdxy = ((motion_y & 1) << 1) | (mx & 1);
uvsrc_x = s->mb_x* 8 + (mx >> 1); uvsrc_x = s->mb_x * 8 + (mx >> 1);
uvsrc_y = src_y; uvsrc_y = src_y;
} else { } else {
//Chroma444 // Chroma444
uvdxy = dxy; uvdxy = dxy;
uvsrc_x = src_x; uvsrc_x = src_x;
uvsrc_y = src_y; uvsrc_y = src_y;
} }
...@@ -281,58 +310,63 @@ if(s->quarter_sample) ...@@ -281,58 +310,63 @@ if(s->quarter_sample)
ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x; ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x; ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;
if( (unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x&1) - 16, 0) if ((unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x & 1) - 16, 0) ||
|| (unsigned)src_y > FFMAX( v_edge_pos - (motion_y&1) - h , 0)){ (unsigned)src_y > FFMAX(v_edge_pos - (motion_y & 1) - h, 0)) {
if(is_mpeg12 || s->codec_id == AV_CODEC_ID_MPEG2VIDEO || if (is_mpeg12 ||
s->codec_id == AV_CODEC_ID_MPEG1VIDEO){ s->codec_id == AV_CODEC_ID_MPEG2VIDEO ||
av_log(s->avctx,AV_LOG_DEBUG, s->codec_id == AV_CODEC_ID_MPEG1VIDEO) {
"MPEG motion vector out of boundary (%d %d)\n", src_x, src_y); av_log(s->avctx, AV_LOG_DEBUG,
return; "MPEG motion vector out of boundary (%d %d)\n", src_x,
} src_y);
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr_y, s->linesize, return;
17, 17+field_based, }
src_x, src_y<<field_based, s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr_y,
s->h_edge_pos, s->v_edge_pos); s->linesize,
ptr_y = s->edge_emu_buffer; 17, 17 + field_based,
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ src_x, src_y << field_based,
uint8_t *uvbuf= s->edge_emu_buffer+18*s->linesize; s->h_edge_pos, s->v_edge_pos);
s->vdsp.emulated_edge_mc(uvbuf , ptr_y = s->edge_emu_buffer;
ptr_cb, s->uvlinesize, if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
9, 9+field_based, uint8_t *uvbuf = s->edge_emu_buffer + 18 * s->linesize;
uvsrc_x, uvsrc_y<<field_based, s->vdsp.emulated_edge_mc(uvbuf, ptr_cb,
s->h_edge_pos>>1, s->v_edge_pos>>1); s->uvlinesize,
s->vdsp.emulated_edge_mc(uvbuf+16, 9, 9 + field_based,
ptr_cr, s->uvlinesize, uvsrc_x, uvsrc_y << field_based,
9, 9+field_based, s->h_edge_pos >> 1, s->v_edge_pos >> 1);
uvsrc_x, uvsrc_y<<field_based, s->vdsp.emulated_edge_mc(uvbuf + 16, ptr_cr,
s->h_edge_pos>>1, s->v_edge_pos>>1); s->uvlinesize,
ptr_cb= uvbuf; 9, 9 + field_based,
ptr_cr= uvbuf+16; uvsrc_x, uvsrc_y << field_based,
} s->h_edge_pos >> 1, s->v_edge_pos >> 1);
ptr_cb = uvbuf;
ptr_cr = uvbuf + 16;
}
} }
if(bottom_field){ //FIXME use this for field pix too instead of the obnoxious hack which changes picture.data /* FIXME use this for field pix too instead of the obnoxious hack which
dest_y += s->linesize; * changes picture.data */
dest_cb+= s->uvlinesize; if (bottom_field) {
dest_cr+= s->uvlinesize; dest_y += s->linesize;
dest_cb += s->uvlinesize;
dest_cr += s->uvlinesize;
} }
if(field_select){ if (field_select) {
ptr_y += s->linesize; ptr_y += s->linesize;
ptr_cb+= s->uvlinesize; ptr_cb += s->uvlinesize;
ptr_cr+= s->uvlinesize; ptr_cr += s->uvlinesize;
} }
pix_op[0][dxy](dest_y, ptr_y, linesize, h); pix_op[0][dxy](dest_y, ptr_y, linesize, h);
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
pix_op[s->chroma_x_shift][uvdxy] pix_op[s->chroma_x_shift][uvdxy]
(dest_cb, ptr_cb, uvlinesize, h >> s->chroma_y_shift); (dest_cb, ptr_cb, uvlinesize, h >> s->chroma_y_shift);
pix_op[s->chroma_x_shift][uvdxy] pix_op[s->chroma_x_shift][uvdxy]
(dest_cr, ptr_cr, uvlinesize, h >> s->chroma_y_shift); (dest_cr, ptr_cr, uvlinesize, h >> s->chroma_y_shift);
} }
if(!is_mpeg12 && (CONFIG_H261_ENCODER || CONFIG_H261_DECODER) && if (!is_mpeg12 && (CONFIG_H261_ENCODER || CONFIG_H261_DECODER) &&
s->out_format == FMT_H261){ s->out_format == FMT_H261) {
ff_h261_loop_filter(s); ff_h261_loop_filter(s);
} }
} }
...@@ -344,15 +378,15 @@ static void mpeg_motion(MpegEncContext *s, ...@@ -344,15 +378,15 @@ static void mpeg_motion(MpegEncContext *s,
int motion_x, int motion_y, int h, int mb_y) int motion_x, int motion_y, int h, int mb_y)
{ {
#if !CONFIG_SMALL #if !CONFIG_SMALL
if(s->out_format == FMT_MPEG1) if (s->out_format == FMT_MPEG1)
mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0, mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0,
field_select, ref_picture, pix_op, field_select, ref_picture, pix_op,
motion_x, motion_y, h, 1, mb_y); motion_x, motion_y, h, 1, mb_y);
else else
#endif #endif
mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0, mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 0, 0,
field_select, ref_picture, pix_op, field_select, ref_picture, pix_op,
motion_x, motion_y, h, 0, mb_y); motion_x, motion_y, h, 0, mb_y);
} }
static void mpeg_motion_field(MpegEncContext *s, uint8_t *dest_y, static void mpeg_motion_field(MpegEncContext *s, uint8_t *dest_y,
...@@ -365,23 +399,24 @@ static void mpeg_motion_field(MpegEncContext *s, uint8_t *dest_y, ...@@ -365,23 +399,24 @@ static void mpeg_motion_field(MpegEncContext *s, uint8_t *dest_y,
#if !CONFIG_SMALL #if !CONFIG_SMALL
if(s->out_format == FMT_MPEG1) if(s->out_format == FMT_MPEG1)
mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1, mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1,
bottom_field, field_select, ref_picture, pix_op, bottom_field, field_select, ref_picture, pix_op,
motion_x, motion_y, h, 1, mb_y); motion_x, motion_y, h, 1, mb_y);
else else
#endif #endif
mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1, mpeg_motion_internal(s, dest_y, dest_cb, dest_cr, 1,
bottom_field, field_select, ref_picture, pix_op, bottom_field, field_select, ref_picture, pix_op,
motion_x, motion_y, h, 0, mb_y); motion_x, motion_y, h, 0, mb_y);
} }
//FIXME move to dsputil, avg variant, 16x16 version // FIXME move to dsputil, avg variant, 16x16 version
static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride){ static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride)
{
int x; int x;
uint8_t * const top = src[1]; uint8_t *const top = src[1];
uint8_t * const left = src[2]; uint8_t *const left = src[2];
uint8_t * const mid = src[0]; uint8_t *const mid = src[0];
uint8_t * const right = src[3]; uint8_t *const right = src[3];
uint8_t * const bottom= src[4]; uint8_t *const bottom = src[4];
#define OBMC_FILTER(x, t, l, m, r, b)\ #define OBMC_FILTER(x, t, l, m, r, b)\
dst[x]= (t*top[x] + l*left[x] + m*mid[x] + r*right[x] + b*bottom[x] + 4)>>3 dst[x]= (t*top[x] + l*left[x] + m*mid[x] + r*right[x] + b*bottom[x] + 4)>>3
#define OBMC_FILTER4(x, t, l, m, r, b)\ #define OBMC_FILTER4(x, t, l, m, r, b)\
...@@ -390,40 +425,40 @@ static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride){ ...@@ -390,40 +425,40 @@ static inline void put_obmc(uint8_t *dst, uint8_t *src[5], int stride){
OBMC_FILTER(x +stride, t, l, m, r, b);\ OBMC_FILTER(x +stride, t, l, m, r, b);\
OBMC_FILTER(x+1+stride, t, l, m, r, b); OBMC_FILTER(x+1+stride, t, l, m, r, b);
x=0; x = 0;
OBMC_FILTER (x , 2, 2, 4, 0, 0); OBMC_FILTER (x , 2, 2, 4, 0, 0);
OBMC_FILTER (x+1, 2, 1, 5, 0, 0); OBMC_FILTER (x + 1, 2, 1, 5, 0, 0);
OBMC_FILTER4(x+2, 2, 1, 5, 0, 0); OBMC_FILTER4(x + 2, 2, 1, 5, 0, 0);
OBMC_FILTER4(x+4, 2, 0, 5, 1, 0); OBMC_FILTER4(x + 4, 2, 0, 5, 1, 0);
OBMC_FILTER (x+6, 2, 0, 5, 1, 0); OBMC_FILTER (x + 6, 2, 0, 5, 1, 0);
OBMC_FILTER (x+7, 2, 0, 4, 2, 0); OBMC_FILTER (x + 7, 2, 0, 4, 2, 0);
x+= stride; x += stride;
OBMC_FILTER (x , 1, 2, 5, 0, 0); OBMC_FILTER (x , 1, 2, 5, 0, 0);
OBMC_FILTER (x+1, 1, 2, 5, 0, 0); OBMC_FILTER (x + 1, 1, 2, 5, 0, 0);
OBMC_FILTER (x+6, 1, 0, 5, 2, 0); OBMC_FILTER (x + 6, 1, 0, 5, 2, 0);
OBMC_FILTER (x+7, 1, 0, 5, 2, 0); OBMC_FILTER (x + 7, 1, 0, 5, 2, 0);
x+= stride; x += stride;
OBMC_FILTER4(x , 1, 2, 5, 0, 0); OBMC_FILTER4(x , 1, 2, 5, 0, 0);
OBMC_FILTER4(x+2, 1, 1, 6, 0, 0); OBMC_FILTER4(x + 2, 1, 1, 6, 0, 0);
OBMC_FILTER4(x+4, 1, 0, 6, 1, 0); OBMC_FILTER4(x + 4, 1, 0, 6, 1, 0);
OBMC_FILTER4(x+6, 1, 0, 5, 2, 0); OBMC_FILTER4(x + 6, 1, 0, 5, 2, 0);
x+= 2*stride; x += 2 * stride;
OBMC_FILTER4(x , 0, 2, 5, 0, 1); OBMC_FILTER4(x , 0, 2, 5, 0, 1);
OBMC_FILTER4(x+2, 0, 1, 6, 0, 1); OBMC_FILTER4(x + 2, 0, 1, 6, 0, 1);
OBMC_FILTER4(x+4, 0, 0, 6, 1, 1); OBMC_FILTER4(x + 4, 0, 0, 6, 1, 1);
OBMC_FILTER4(x+6, 0, 0, 5, 2, 1); OBMC_FILTER4(x + 6, 0, 0, 5, 2, 1);
x+= 2*stride; x += 2*stride;
OBMC_FILTER (x , 0, 2, 5, 0, 1); OBMC_FILTER (x , 0, 2, 5, 0, 1);
OBMC_FILTER (x+1, 0, 2, 5, 0, 1); OBMC_FILTER (x + 1, 0, 2, 5, 0, 1);
OBMC_FILTER4(x+2, 0, 1, 5, 0, 2); OBMC_FILTER4(x + 2, 0, 1, 5, 0, 2);
OBMC_FILTER4(x+4, 0, 0, 5, 1, 2); OBMC_FILTER4(x + 4, 0, 0, 5, 1, 2);
OBMC_FILTER (x+6, 0, 0, 5, 2, 1); OBMC_FILTER (x + 6, 0, 0, 5, 2, 1);
OBMC_FILTER (x+7, 0, 0, 5, 2, 1); OBMC_FILTER (x + 7, 0, 0, 5, 2, 1);
x+= stride; x += stride;
OBMC_FILTER (x , 0, 2, 4, 0, 2); OBMC_FILTER (x , 0, 2, 4, 0, 2);
OBMC_FILTER (x+1, 0, 1, 5, 0, 2); OBMC_FILTER (x + 1, 0, 1, 5, 0, 2);
OBMC_FILTER (x+6, 0, 0, 5, 1, 2); OBMC_FILTER (x + 6, 0, 0, 5, 1, 2);
OBMC_FILTER (x+7, 0, 0, 4, 2, 2); OBMC_FILTER (x + 7, 0, 0, 4, 2, 2);
} }
/* obmc for 1 8x8 luma block */ /* obmc for 1 8x8 luma block */
...@@ -431,22 +466,21 @@ static inline void obmc_motion(MpegEncContext *s, ...@@ -431,22 +466,21 @@ static inline void obmc_motion(MpegEncContext *s,
uint8_t *dest, uint8_t *src, uint8_t *dest, uint8_t *src,
int src_x, int src_y, int src_x, int src_y,
op_pixels_func *pix_op, op_pixels_func *pix_op,
int16_t mv[5][2]/* mid top left right bottom*/) int16_t mv[5][2] /* mid top left right bottom */)
#define MID 0 #define MID 0
{ {
int i; int i;
uint8_t *ptr[5]; uint8_t *ptr[5];
assert(s->quarter_sample==0); assert(s->quarter_sample == 0);
for(i=0; i<5; i++){ for (i = 0; i < 5; i++) {
if(i && mv[i][0]==mv[MID][0] && mv[i][1]==mv[MID][1]){ if (i && mv[i][0] == mv[MID][0] && mv[i][1] == mv[MID][1]) {
ptr[i]= ptr[MID]; ptr[i] = ptr[MID];
}else{ } else {
ptr[i]= s->obmc_scratchpad + 8*(i&1) + s->linesize*8*(i>>1); ptr[i] = s->obmc_scratchpad + 8 * (i & 1) +
hpel_motion(s, ptr[i], src, s->linesize * 8 * (i >> 1);
src_x, src_y, hpel_motion(s, ptr[i], src, src_x, src_y, pix_op,
pix_op,
mv[i][0], mv[i][1]); mv[i][0], mv[i][1]);
} }
} }
...@@ -455,9 +489,12 @@ static inline void obmc_motion(MpegEncContext *s, ...@@ -455,9 +489,12 @@ static inline void obmc_motion(MpegEncContext *s,
} }
static inline void qpel_motion(MpegEncContext *s, static inline void qpel_motion(MpegEncContext *s,
uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr, uint8_t *dest_y,
int field_based, int bottom_field, int field_select, uint8_t *dest_cb,
uint8_t **ref_picture, op_pixels_func (*pix_op)[4], uint8_t *dest_cr,
int field_based, int bottom_field,
int field_select, uint8_t **ref_picture,
op_pixels_func (*pix_op)[4],
qpel_mc_func (*qpix_op)[16], qpel_mc_func (*qpix_op)[16],
int motion_x, int motion_y, int h) int motion_x, int motion_y, int h)
{ {
...@@ -465,83 +502,88 @@ static inline void qpel_motion(MpegEncContext *s, ...@@ -465,83 +502,88 @@ static inline void qpel_motion(MpegEncContext *s,
int dxy, uvdxy, mx, my, src_x, src_y, uvsrc_x, uvsrc_y, v_edge_pos; int dxy, uvdxy, mx, my, src_x, src_y, uvsrc_x, uvsrc_y, v_edge_pos;
ptrdiff_t linesize, uvlinesize; ptrdiff_t linesize, uvlinesize;
dxy = ((motion_y & 3) << 2) | (motion_x & 3); dxy = ((motion_y & 3) << 2) | (motion_x & 3);
src_x = s->mb_x * 16 + (motion_x >> 2); src_x = s->mb_x * 16 + (motion_x >> 2);
src_y = s->mb_y * (16 >> field_based) + (motion_y >> 2); src_y = s->mb_y * (16 >> field_based) + (motion_y >> 2);
v_edge_pos = s->v_edge_pos >> field_based; v_edge_pos = s->v_edge_pos >> field_based;
linesize = s->linesize << field_based; linesize = s->linesize << field_based;
uvlinesize = s->uvlinesize << field_based; uvlinesize = s->uvlinesize << field_based;
if(field_based){ if (field_based) {
mx= motion_x/2; mx = motion_x / 2;
my= motion_y>>1; my = motion_y >> 1;
}else if(s->workaround_bugs&FF_BUG_QPEL_CHROMA2){ } else if (s->workaround_bugs & FF_BUG_QPEL_CHROMA2) {
static const int rtab[8]= {0,0,1,1,0,0,0,1}; static const int rtab[8] = { 0, 0, 1, 1, 0, 0, 0, 1 };
mx= (motion_x>>1) + rtab[motion_x&7]; mx = (motion_x >> 1) + rtab[motion_x & 7];
my= (motion_y>>1) + rtab[motion_y&7]; my = (motion_y >> 1) + rtab[motion_y & 7];
}else if(s->workaround_bugs&FF_BUG_QPEL_CHROMA){ } else if (s->workaround_bugs & FF_BUG_QPEL_CHROMA) {
mx= (motion_x>>1)|(motion_x&1); mx = (motion_x >> 1) | (motion_x & 1);
my= (motion_y>>1)|(motion_y&1); my = (motion_y >> 1) | (motion_y & 1);
}else{ } else {
mx= motion_x/2; mx = motion_x / 2;
my= motion_y/2; my = motion_y / 2;
} }
mx= (mx>>1)|(mx&1); mx = (mx >> 1) | (mx & 1);
my= (my>>1)|(my&1); my = (my >> 1) | (my & 1);
uvdxy= (mx&1) | ((my&1)<<1); uvdxy = (mx & 1) | ((my & 1) << 1);
mx>>=1; mx >>= 1;
my>>=1; my >>= 1;
uvsrc_x = s->mb_x * 8 + mx; uvsrc_x = s->mb_x * 8 + mx;
uvsrc_y = s->mb_y * (8 >> field_based) + my; uvsrc_y = s->mb_y * (8 >> field_based) + my;
ptr_y = ref_picture[0] + src_y * linesize + src_x; ptr_y = ref_picture[0] + src_y * linesize + src_x;
ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x; ptr_cb = ref_picture[1] + uvsrc_y * uvlinesize + uvsrc_x;
ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x; ptr_cr = ref_picture[2] + uvsrc_y * uvlinesize + uvsrc_x;
if( (unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x&3) - 16, 0) if ((unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x & 3) - 16, 0) ||
|| (unsigned)src_y > FFMAX( v_edge_pos - (motion_y&3) - h , 0)){ (unsigned)src_y > FFMAX(v_edge_pos - (motion_y & 3) - h, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr_y, s->linesize, s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr_y,
17, 17+field_based, src_x, src_y<<field_based, s->linesize,
s->h_edge_pos, s->v_edge_pos); 17, 17 + field_based,
ptr_y= s->edge_emu_buffer; src_x, src_y << field_based,
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ s->h_edge_pos, s->v_edge_pos);
uint8_t *uvbuf= s->edge_emu_buffer + 18*s->linesize; ptr_y = s->edge_emu_buffer;
s->vdsp.emulated_edge_mc(uvbuf, ptr_cb, s->uvlinesize, if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
9, 9 + field_based, uint8_t *uvbuf = s->edge_emu_buffer + 18 * s->linesize;
uvsrc_x, uvsrc_y<<field_based, s->vdsp.emulated_edge_mc(uvbuf, ptr_cb,
s->h_edge_pos>>1, s->v_edge_pos>>1); s->uvlinesize,
s->vdsp.emulated_edge_mc(uvbuf + 16, ptr_cr, s->uvlinesize, 9, 9 + field_based,
9, 9 + field_based, uvsrc_x, uvsrc_y << field_based,
uvsrc_x, uvsrc_y<<field_based, s->h_edge_pos >> 1, s->v_edge_pos >> 1);
s->h_edge_pos>>1, s->v_edge_pos>>1); s->vdsp.emulated_edge_mc(uvbuf + 16, ptr_cr,
ptr_cb= uvbuf; s->uvlinesize,
ptr_cr= uvbuf + 16; 9, 9 + field_based,
uvsrc_x, uvsrc_y << field_based,
s->h_edge_pos >> 1, s->v_edge_pos >> 1);
ptr_cb = uvbuf;
ptr_cr = uvbuf + 16;
} }
} }
if(!field_based) if (!field_based)
qpix_op[0][dxy](dest_y, ptr_y, linesize); qpix_op[0][dxy](dest_y, ptr_y, linesize);
else{ else {
if(bottom_field){ if (bottom_field) {
dest_y += s->linesize; dest_y += s->linesize;
dest_cb+= s->uvlinesize; dest_cb += s->uvlinesize;
dest_cr+= s->uvlinesize; dest_cr += s->uvlinesize;
} }
if(field_select){ if (field_select) {
ptr_y += s->linesize; ptr_y += s->linesize;
ptr_cb += s->uvlinesize; ptr_cb += s->uvlinesize;
ptr_cr += s->uvlinesize; ptr_cr += s->uvlinesize;
} }
//damn interlaced mode // damn interlaced mode
//FIXME boundary mirroring is not exactly correct here // FIXME boundary mirroring is not exactly correct here
qpix_op[1][dxy](dest_y , ptr_y , linesize); qpix_op[1][dxy](dest_y, ptr_y, linesize);
qpix_op[1][dxy](dest_y+8, ptr_y+8, linesize); qpix_op[1][dxy](dest_y + 8, ptr_y + 8, linesize);
} }
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
pix_op[1][uvdxy](dest_cr, ptr_cr, uvlinesize, h >> 1); pix_op[1][uvdxy](dest_cr, ptr_cr, uvlinesize, h >> 1);
pix_op[1][uvdxy](dest_cb, ptr_cb, uvlinesize, h >> 1); pix_op[1][uvdxy](dest_cb, ptr_cb, uvlinesize, h >> 1);
} }
...@@ -561,11 +603,11 @@ static void chroma_4mv_motion(MpegEncContext *s, ...@@ -561,11 +603,11 @@ static void chroma_4mv_motion(MpegEncContext *s,
ptrdiff_t offset; ptrdiff_t offset;
/* In case of 8X8, we construct a single chroma motion vector /* In case of 8X8, we construct a single chroma motion vector
with a special rounding */ * with a special rounding */
mx= ff_h263_round_chroma(mx); mx = ff_h263_round_chroma(mx);
my= ff_h263_round_chroma(my); my = ff_h263_round_chroma(my);
dxy = ((my & 1) << 1) | (mx & 1); dxy = ((my & 1) << 1) | (mx & 1);
mx >>= 1; mx >>= 1;
my >>= 1; my >>= 1;
...@@ -579,39 +621,43 @@ static void chroma_4mv_motion(MpegEncContext *s, ...@@ -579,39 +621,43 @@ static void chroma_4mv_motion(MpegEncContext *s,
dxy &= ~2; dxy &= ~2;
offset = src_y * s->uvlinesize + src_x; offset = src_y * s->uvlinesize + src_x;
ptr = ref_picture[1] + offset; ptr = ref_picture[1] + offset;
if(s->flags&CODEC_FLAG_EMU_EDGE){ if (s->flags & CODEC_FLAG_EMU_EDGE) {
if( (unsigned)src_x > FFMAX((s->h_edge_pos>>1) - (dxy &1) - 8, 0) if ((unsigned)src_x > FFMAX((s->h_edge_pos >> 1) - (dxy & 1) - 8, 0) ||
|| (unsigned)src_y > FFMAX((s->v_edge_pos>>1) - (dxy>>1) - 8, 0)){ (unsigned)src_y > FFMAX((s->v_edge_pos >> 1) - (dxy >> 1) - 8, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, s->vdsp.emulated_edge_mc(s->edge_emu_buffer,
9, 9, src_x, src_y, ptr, s->uvlinesize,
s->h_edge_pos>>1, s->v_edge_pos>>1); 9, 9, src_x, src_y,
ptr= s->edge_emu_buffer; s->h_edge_pos >> 1, s->v_edge_pos >> 1);
emu=1; ptr = s->edge_emu_buffer;
emu = 1;
} }
} }
pix_op[dxy](dest_cb, ptr, s->uvlinesize, 8); pix_op[dxy](dest_cb, ptr, s->uvlinesize, 8);
ptr = ref_picture[2] + offset; ptr = ref_picture[2] + offset;
if(emu){ if (emu) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, s->uvlinesize, s->vdsp.emulated_edge_mc(s->edge_emu_buffer,
9, 9, src_x, src_y, ptr, s->uvlinesize,
s->h_edge_pos>>1, s->v_edge_pos>>1); 9, 9, src_x, src_y,
ptr= s->edge_emu_buffer; s->h_edge_pos >> 1, s->v_edge_pos >> 1);
ptr = s->edge_emu_buffer;
} }
pix_op[dxy](dest_cr, ptr, s->uvlinesize, 8); pix_op[dxy](dest_cr, ptr, s->uvlinesize, 8);
} }
static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir){ static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir)
{
/* fetch pixels for estimated mv 4 macroblocks ahead /* fetch pixels for estimated mv 4 macroblocks ahead
* optimized for 64byte cache lines */ * optimized for 64byte cache lines */
const int shift = s->quarter_sample ? 2 : 1; const int shift = s->quarter_sample ? 2 : 1;
const int mx= (s->mv[dir][0][0]>>shift) + 16*s->mb_x + 8; const int mx = (s->mv[dir][0][0] >> shift) + 16 * s->mb_x + 8;
const int my= (s->mv[dir][0][1]>>shift) + 16*s->mb_y; const int my = (s->mv[dir][0][1] >> shift) + 16 * s->mb_y;
int off= mx + (my + (s->mb_x&3)*4)*s->linesize + 64; int off = mx + (my + (s->mb_x & 3) * 4) * s->linesize + 64;
s->vdsp.prefetch(pix[0]+off, s->linesize, 4);
off= (mx>>1) + ((my>>1) + (s->mb_x&7))*s->uvlinesize + 64; s->vdsp.prefetch(pix[0] + off, s->linesize, 4);
s->vdsp.prefetch(pix[1]+off, pix[2]-pix[1], 2); off = (mx >> 1) + ((my >> 1) + (s->mb_x & 7)) * s->uvlinesize + 64;
s->vdsp.prefetch(pix[1] + off, pix[2] - pix[1], 2);
} }
/** /**
...@@ -627,11 +673,14 @@ static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir){ ...@@ -627,11 +673,14 @@ static inline void prefetch_motion(MpegEncContext *s, uint8_t **pix, int dir){
* the motion vectors are taken from s->mv and the MV type from s->mv_type * the motion vectors are taken from s->mv and the MV type from s->mv_type
*/ */
static av_always_inline void MPV_motion_internal(MpegEncContext *s, static av_always_inline void MPV_motion_internal(MpegEncContext *s,
uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_y,
uint8_t *dest_cr, int dir, uint8_t *dest_cb,
uint8_t **ref_picture, uint8_t *dest_cr,
op_pixels_func (*pix_op)[4], int dir,
qpel_mc_func (*qpix_op)[16], int is_mpeg12) uint8_t **ref_picture,
op_pixels_func (*pix_op)[4],
qpel_mc_func (*qpix_op)[16],
int is_mpeg12)
{ {
int dxy, mx, my, src_x, src_y, motion_x, motion_y; int dxy, mx, my, src_x, src_y, motion_x, motion_y;
int mb_x, mb_y, i; int mb_x, mb_y, i;
...@@ -642,86 +691,97 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s, ...@@ -642,86 +691,97 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s,
prefetch_motion(s, ref_picture, dir); prefetch_motion(s, ref_picture, dir);
if(!is_mpeg12 && s->obmc && s->pict_type != AV_PICTURE_TYPE_B){ if (!is_mpeg12 && s->obmc && s->pict_type != AV_PICTURE_TYPE_B) {
LOCAL_ALIGNED_8(int16_t, mv_cache, [4], [4][2]); LOCAL_ALIGNED_8(int16_t, mv_cache, [4], [4][2]);
Picture *cur_frame = &s->current_picture; Picture *cur_frame = &s->current_picture;
const int xy= s->mb_x + s->mb_y*s->mb_stride; const int xy = s->mb_x + s->mb_y * s->mb_stride;
const int mot_stride= s->b8_stride; const int mot_stride = s->b8_stride;
const int mot_xy= mb_x*2 + mb_y*2*mot_stride; const int mot_xy = mb_x * 2 + mb_y * 2 * mot_stride;
assert(!s->mb_skipped); assert(!s->mb_skipped);
AV_COPY32(mv_cache[1][1], cur_frame->motion_val[0][mot_xy ]); AV_COPY32(mv_cache[1][1], cur_frame->motion_val[0][mot_xy]);
AV_COPY32(mv_cache[1][2], cur_frame->motion_val[0][mot_xy + 1]); AV_COPY32(mv_cache[1][2], cur_frame->motion_val[0][mot_xy + 1]);
AV_COPY32(mv_cache[2][1], cur_frame->motion_val[0][mot_xy + mot_stride ]); AV_COPY32(mv_cache[2][1],
AV_COPY32(mv_cache[2][2], cur_frame->motion_val[0][mot_xy + mot_stride + 1]); cur_frame->motion_val[0][mot_xy + mot_stride]);
AV_COPY32(mv_cache[2][2],
cur_frame->motion_val[0][mot_xy + mot_stride + 1]);
AV_COPY32(mv_cache[3][1], cur_frame->motion_val[0][mot_xy + mot_stride ]); AV_COPY32(mv_cache[3][1],
AV_COPY32(mv_cache[3][2], cur_frame->motion_val[0][mot_xy + mot_stride + 1]); cur_frame->motion_val[0][mot_xy + mot_stride]);
AV_COPY32(mv_cache[3][2],
cur_frame->motion_val[0][mot_xy + mot_stride + 1]);
if (mb_y == 0 || IS_INTRA(cur_frame->mb_type[xy - s->mb_stride])) { if (mb_y == 0 || IS_INTRA(cur_frame->mb_type[xy - s->mb_stride])) {
AV_COPY32(mv_cache[0][1], mv_cache[1][1]); AV_COPY32(mv_cache[0][1], mv_cache[1][1]);
AV_COPY32(mv_cache[0][2], mv_cache[1][2]); AV_COPY32(mv_cache[0][2], mv_cache[1][2]);
}else{ } else {
AV_COPY32(mv_cache[0][1], cur_frame->motion_val[0][mot_xy - mot_stride ]); AV_COPY32(mv_cache[0][1],
AV_COPY32(mv_cache[0][2], cur_frame->motion_val[0][mot_xy - mot_stride + 1]); cur_frame->motion_val[0][mot_xy - mot_stride]);
AV_COPY32(mv_cache[0][2],
cur_frame->motion_val[0][mot_xy - mot_stride + 1]);
} }
if (mb_x == 0 || IS_INTRA(cur_frame->mb_type[xy - 1])) { if (mb_x == 0 || IS_INTRA(cur_frame->mb_type[xy - 1])) {
AV_COPY32(mv_cache[1][0], mv_cache[1][1]); AV_COPY32(mv_cache[1][0], mv_cache[1][1]);
AV_COPY32(mv_cache[2][0], mv_cache[2][1]); AV_COPY32(mv_cache[2][0], mv_cache[2][1]);
}else{ } else {
AV_COPY32(mv_cache[1][0], cur_frame->motion_val[0][mot_xy - 1]); AV_COPY32(mv_cache[1][0], cur_frame->motion_val[0][mot_xy - 1]);
AV_COPY32(mv_cache[2][0], cur_frame->motion_val[0][mot_xy - 1 + mot_stride]); AV_COPY32(mv_cache[2][0],
cur_frame->motion_val[0][mot_xy - 1 + mot_stride]);
} }
if (mb_x + 1 >= s->mb_width || IS_INTRA(cur_frame->mb_type[xy + 1])) { if (mb_x + 1 >= s->mb_width || IS_INTRA(cur_frame->mb_type[xy + 1])) {
AV_COPY32(mv_cache[1][3], mv_cache[1][2]); AV_COPY32(mv_cache[1][3], mv_cache[1][2]);
AV_COPY32(mv_cache[2][3], mv_cache[2][2]); AV_COPY32(mv_cache[2][3], mv_cache[2][2]);
}else{ } else {
AV_COPY32(mv_cache[1][3], cur_frame->motion_val[0][mot_xy + 2]); AV_COPY32(mv_cache[1][3], cur_frame->motion_val[0][mot_xy + 2]);
AV_COPY32(mv_cache[2][3], cur_frame->motion_val[0][mot_xy + 2 + mot_stride]); AV_COPY32(mv_cache[2][3],
cur_frame->motion_val[0][mot_xy + 2 + mot_stride]);
} }
mx = 0; mx = 0;
my = 0; my = 0;
for(i=0;i<4;i++) { for (i = 0; i < 4; i++) {
const int x= (i&1)+1; const int x = (i & 1) + 1;
const int y= (i>>1)+1; const int y = (i >> 1) + 1;
int16_t mv[5][2]= { int16_t mv[5][2] = {
{mv_cache[y][x ][0], mv_cache[y][x ][1]}, { mv_cache[y][x][0], mv_cache[y][x][1] },
{mv_cache[y-1][x][0], mv_cache[y-1][x][1]}, { mv_cache[y - 1][x][0], mv_cache[y - 1][x][1] },
{mv_cache[y][x-1][0], mv_cache[y][x-1][1]}, { mv_cache[y][x - 1][0], mv_cache[y][x - 1][1] },
{mv_cache[y][x+1][0], mv_cache[y][x+1][1]}, { mv_cache[y][x + 1][0], mv_cache[y][x + 1][1] },
{mv_cache[y+1][x][0], mv_cache[y+1][x][1]}}; { mv_cache[y + 1][x][0], mv_cache[y + 1][x][1] }
//FIXME cleanup };
// FIXME cleanup
obmc_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize, obmc_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
ref_picture[0], ref_picture[0],
mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >>1) * 8, mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >> 1) * 8,
pix_op[1], pix_op[1],
mv); mv);
mx += mv[0][0]; mx += mv[0][0];
my += mv[0][1]; my += mv[0][1];
} }
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)) if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY))
chroma_4mv_motion(s, dest_cb, dest_cr, ref_picture, pix_op[1], mx, my); chroma_4mv_motion(s, dest_cb, dest_cr,
ref_picture, pix_op[1],
mx, my);
return; return;
} }
switch(s->mv_type) { switch (s->mv_type) {
case MV_TYPE_16X16: case MV_TYPE_16X16:
if(s->mcsel){ if (s->mcsel) {
if(s->real_sprite_warping_points==1){ if (s->real_sprite_warping_points == 1) {
gmc1_motion(s, dest_y, dest_cb, dest_cr, gmc1_motion(s, dest_y, dest_cb, dest_cr,
ref_picture); ref_picture);
}else{ } else {
gmc_motion(s, dest_y, dest_cb, dest_cr, gmc_motion(s, dest_y, dest_cb, dest_cr,
ref_picture); ref_picture);
} }
}else if(!is_mpeg12 && s->quarter_sample){ } else if (!is_mpeg12 && s->quarter_sample) {
qpel_motion(s, dest_y, dest_cb, dest_cr, qpel_motion(s, dest_y, dest_cb, dest_cr,
0, 0, 0, 0, 0, 0,
ref_picture, pix_op, qpix_op, ref_picture, pix_op, qpix_op,
...@@ -729,80 +789,84 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s, ...@@ -729,80 +789,84 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s,
} else if (!is_mpeg12 && (CONFIG_WMV2_DECODER || CONFIG_WMV2_ENCODER) && } else if (!is_mpeg12 && (CONFIG_WMV2_DECODER || CONFIG_WMV2_ENCODER) &&
s->mspel && s->codec_id == AV_CODEC_ID_WMV2) { s->mspel && s->codec_id == AV_CODEC_ID_WMV2) {
ff_mspel_motion(s, dest_y, dest_cb, dest_cr, ff_mspel_motion(s, dest_y, dest_cb, dest_cr,
ref_picture, pix_op, ref_picture, pix_op,
s->mv[dir][0][0], s->mv[dir][0][1], 16); s->mv[dir][0][0], s->mv[dir][0][1], 16);
}else } else {
{
mpeg_motion(s, dest_y, dest_cb, dest_cr, 0, mpeg_motion(s, dest_y, dest_cb, dest_cr, 0,
ref_picture, pix_op, ref_picture, pix_op,
s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y); s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y);
} }
break; break;
case MV_TYPE_8X8: case MV_TYPE_8X8:
if (!is_mpeg12) { if (!is_mpeg12) {
mx = 0; mx = 0;
my = 0; my = 0;
if(s->quarter_sample){ if (s->quarter_sample) {
for(i=0;i<4;i++) { for (i = 0; i < 4; i++) {
motion_x = s->mv[dir][i][0]; motion_x = s->mv[dir][i][0];
motion_y = s->mv[dir][i][1]; motion_y = s->mv[dir][i][1];
dxy = ((motion_y & 3) << 2) | (motion_x & 3); dxy = ((motion_y & 3) << 2) | (motion_x & 3);
src_x = mb_x * 16 + (motion_x >> 2) + (i & 1) * 8; src_x = mb_x * 16 + (motion_x >> 2) + (i & 1) * 8;
src_y = mb_y * 16 + (motion_y >> 2) + (i >>1) * 8; src_y = mb_y * 16 + (motion_y >> 2) + (i >> 1) * 8;
/* WARNING: do no forget half pels */ /* WARNING: do no forget half pels */
src_x = av_clip(src_x, -16, s->width); src_x = av_clip(src_x, -16, s->width);
if (src_x == s->width) if (src_x == s->width)
dxy &= ~3; dxy &= ~3;
src_y = av_clip(src_y, -16, s->height); src_y = av_clip(src_y, -16, s->height);
if (src_y == s->height) if (src_y == s->height)
dxy &= ~12; dxy &= ~12;
ptr = ref_picture[0] + (src_y * s->linesize) + (src_x); ptr = ref_picture[0] + (src_y * s->linesize) + (src_x);
if(s->flags&CODEC_FLAG_EMU_EDGE){ if (s->flags & CODEC_FLAG_EMU_EDGE) {
if( (unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x&3) - 8, 0) if ((unsigned)src_x > FFMAX(s->h_edge_pos - (motion_x & 3) - 8, 0) ||
|| (unsigned)src_y > FFMAX(s->v_edge_pos - (motion_y&3) - 8, 0)){ (unsigned)src_y > FFMAX(s->v_edge_pos - (motion_y & 3) - 8, 0)) {
s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr, s->vdsp.emulated_edge_mc(s->edge_emu_buffer, ptr,
s->linesize, 9, 9, s->linesize,
src_x, src_y, 9, 9,
s->h_edge_pos, s->v_edge_pos); src_x, src_y,
ptr= s->edge_emu_buffer; s->h_edge_pos,
s->v_edge_pos);
ptr = s->edge_emu_buffer;
}
} }
} dest = dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize;
dest = dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize; qpix_op[1][dxy](dest, ptr, s->linesize);
qpix_op[1][dxy](dest, ptr, s->linesize);
mx += s->mv[dir][i][0]/2; mx += s->mv[dir][i][0] / 2;
my += s->mv[dir][i][1]/2; my += s->mv[dir][i][1] / 2;
} }
}else{ } else {
for(i=0;i<4;i++) { for (i = 0; i < 4; i++) {
hpel_motion(s, dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize, hpel_motion(s,
ref_picture[0], dest_y + ((i & 1) * 8) + (i >> 1) * 8 * s->linesize,
mb_x * 16 + (i & 1) * 8, mb_y * 16 + (i >>1) * 8, ref_picture[0],
pix_op[1], mb_x * 16 + (i & 1) * 8,
s->mv[dir][i][0], s->mv[dir][i][1]); mb_y * 16 + (i >> 1) * 8,
pix_op[1],
mx += s->mv[dir][i][0]; s->mv[dir][i][0],
my += s->mv[dir][i][1]; s->mv[dir][i][1]);
mx += s->mv[dir][i][0];
my += s->mv[dir][i][1];
}
} }
}
if(!CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)) if (!CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY))
chroma_4mv_motion(s, dest_cb, dest_cr, ref_picture, pix_op[1], mx, my); chroma_4mv_motion(s, dest_cb, dest_cr,
} ref_picture, pix_op[1], mx, my);
}
break; break;
case MV_TYPE_FIELD: case MV_TYPE_FIELD:
if (s->picture_structure == PICT_FRAME) { if (s->picture_structure == PICT_FRAME) {
if(!is_mpeg12 && s->quarter_sample){ if (!is_mpeg12 && s->quarter_sample) {
for(i=0; i<2; i++){ for (i = 0; i < 2; i++)
qpel_motion(s, dest_y, dest_cb, dest_cr, qpel_motion(s, dest_y, dest_cb, dest_cr,
1, i, s->field_select[dir][i], 1, i, s->field_select[dir][i],
ref_picture, pix_op, qpix_op, ref_picture, pix_op, qpix_op,
s->mv[dir][i][0], s->mv[dir][i][1], 8); s->mv[dir][i][0], s->mv[dir][i][1], 8);
} } else {
}else{
/* top field */ /* top field */
mpeg_motion_field(s, dest_y, dest_cb, dest_cr, mpeg_motion_field(s, dest_y, dest_cb, dest_cr,
0, s->field_select[dir][0], 0, s->field_select[dir][0],
...@@ -815,66 +879,69 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s, ...@@ -815,66 +879,69 @@ static av_always_inline void MPV_motion_internal(MpegEncContext *s,
s->mv[dir][1][0], s->mv[dir][1][1], 8, mb_y); s->mv[dir][1][0], s->mv[dir][1][1], 8, mb_y);
} }
} else { } else {
if(s->picture_structure != s->field_select[dir][0] + 1 && s->pict_type != AV_PICTURE_TYPE_B && !s->first_field){ if (s->picture_structure != s->field_select[dir][0] + 1 &&
s->pict_type != AV_PICTURE_TYPE_B && !s->first_field) {
ref_picture = s->current_picture_ptr->f.data; ref_picture = s->current_picture_ptr->f.data;
} }
mpeg_motion(s, dest_y, dest_cb, dest_cr, mpeg_motion(s, dest_y, dest_cb, dest_cr,
s->field_select[dir][0], s->field_select[dir][0],
ref_picture, pix_op, ref_picture, pix_op,
s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y>>1); s->mv[dir][0][0], s->mv[dir][0][1], 16, mb_y >> 1);
} }
break; break;
case MV_TYPE_16X8: case MV_TYPE_16X8:
for(i=0; i<2; i++){ for (i = 0; i < 2; i++) {
uint8_t ** ref2picture; uint8_t **ref2picture;
if(s->picture_structure == s->field_select[dir][i] + 1 if (s->picture_structure == s->field_select[dir][i] + 1
|| s->pict_type == AV_PICTURE_TYPE_B || s->first_field){ || s->pict_type == AV_PICTURE_TYPE_B || s->first_field) {
ref2picture= ref_picture; ref2picture = ref_picture;
}else{ } else {
ref2picture = s->current_picture_ptr->f.data; ref2picture = s->current_picture_ptr->f.data;
} }
mpeg_motion(s, dest_y, dest_cb, dest_cr, mpeg_motion(s, dest_y, dest_cb, dest_cr,
s->field_select[dir][i], s->field_select[dir][i],
ref2picture, pix_op, ref2picture, pix_op,
s->mv[dir][i][0], s->mv[dir][i][1] + 16*i, 8, mb_y>>1); s->mv[dir][i][0], s->mv[dir][i][1] + 16 * i,
8, mb_y >> 1);
dest_y += 16*s->linesize; dest_y += 16 * s->linesize;
dest_cb+= (16>>s->chroma_y_shift)*s->uvlinesize; dest_cb += (16 >> s->chroma_y_shift) * s->uvlinesize;
dest_cr+= (16>>s->chroma_y_shift)*s->uvlinesize; dest_cr += (16 >> s->chroma_y_shift) * s->uvlinesize;
} }
break; break;
case MV_TYPE_DMV: case MV_TYPE_DMV:
if(s->picture_structure == PICT_FRAME){ if (s->picture_structure == PICT_FRAME) {
for(i=0; i<2; i++){ for (i = 0; i < 2; i++) {
int j; int j;
for(j=0; j<2; j++){ for (j = 0; j < 2; j++)
mpeg_motion_field(s, dest_y, dest_cb, dest_cr, mpeg_motion_field(s, dest_y, dest_cb, dest_cr,
j, j^i, ref_picture, pix_op, j, j ^ i, ref_picture, pix_op,
s->mv[dir][2*i + j][0], s->mv[dir][2 * i + j][0],
s->mv[dir][2*i + j][1], 8, mb_y); s->mv[dir][2 * i + j][1], 8, mb_y);
}
pix_op = s->hdsp.avg_pixels_tab; pix_op = s->hdsp.avg_pixels_tab;
} }
}else{ } else {
for(i=0; i<2; i++){ for (i = 0; i < 2; i++) {
mpeg_motion(s, dest_y, dest_cb, dest_cr, mpeg_motion(s, dest_y, dest_cb, dest_cr,
s->picture_structure != i+1, s->picture_structure != i + 1,
ref_picture, pix_op, ref_picture, pix_op,
s->mv[dir][2*i][0],s->mv[dir][2*i][1],16, mb_y>>1); s->mv[dir][2 * i][0], s->mv[dir][2 * i][1],
16, mb_y >> 1);
// after put we make avg of the same block // after put we make avg of the same block
pix_op=s->hdsp.avg_pixels_tab; pix_op = s->hdsp.avg_pixels_tab;
//opposite parity is always in the same frame if this is second field /* opposite parity is always in the same frame if this is
if(!s->first_field){ * second field */
if (!s->first_field) {
ref_picture = s->current_picture_ptr->f.data; ref_picture = s->current_picture_ptr->f.data;
} }
} }
} }
break; break;
default: assert(0); default: assert(0);
} }
} }
...@@ -887,7 +954,7 @@ void ff_MPV_motion(MpegEncContext *s, ...@@ -887,7 +954,7 @@ void ff_MPV_motion(MpegEncContext *s,
qpel_mc_func (*qpix_op)[16]) qpel_mc_func (*qpix_op)[16])
{ {
#if !CONFIG_SMALL #if !CONFIG_SMALL
if(s->out_format == FMT_MPEG1) if (s->out_format == FMT_MPEG1)
MPV_motion_internal(s, dest_y, dest_cb, dest_cr, dir, MPV_motion_internal(s, dest_y, dest_cb, dest_cr, dir,
ref_picture, pix_op, qpix_op, 1); ref_picture, pix_op, qpix_op, 1);
else else
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment