//#undef NDEBUG
#include <assert.h>
-static const uint8_t rem6[52]={
-0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
+static const uint8_t rem6[QP_MAX_NUM+1]={
+0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
};
-static const uint8_t div6[52]={
-0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8,
+static const uint8_t div6[QP_MAX_NUM+1]={
+0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9,10,10,10,10,
};
static const enum PixelFormat hwaccel_pixfmt_list_h264_jpeg_420[] = {
h->nal_unit_type= src[0]&0x1F;
src++; length--;
-#if 0
- for(i=0; i<length; i++)
- printf("%2X ", src[i]);
-#endif
#if HAVE_FAST_UNALIGNED
# if HAVE_FAST_64BIT
#undef xStride
#undef stride
-static void chroma_dc_dequant_idct_c(DCTELEM *block, int qmul){
- const int stride= 16*2;
- const int xStride= 16;
- int a,b,c,d,e;
-
- a= block[stride*0 + xStride*0];
- b= block[stride*0 + xStride*1];
- c= block[stride*1 + xStride*0];
- d= block[stride*1 + xStride*1];
-
- e= a-b;
- a= a+b;
- b= c-d;
- c= c+d;
-
- block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
- block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
- block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
- block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
-}
-
#if 0
static void chroma_dc_dct_c(DCTELEM *block){
const int stride= 16*2;
static inline void mc_dir_part(H264Context *h, Picture *pic, int n, int square, int chroma_height, int delta, int list,
uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
int src_x_offset, int src_y_offset,
- qpel_mc_func *qpix_op, h264_chroma_mc_func chroma_op){
+ qpel_mc_func *qpix_op, h264_chroma_mc_func chroma_op,
+ int pixel_shift){
MpegEncContext * const s = &h->s;
const int mx= h->mv_cache[list][ scan8[n] ][0] + src_x_offset*8;
int my= h->mv_cache[list][ scan8[n] ][1] + src_y_offset*8;
const int luma_xy= (mx&3) + ((my&3)<<2);
- uint8_t * src_y = pic->data[0] + (mx>>2) + (my>>2)*h->mb_linesize;
+ uint8_t * src_y = pic->data[0] + ((mx>>2) << pixel_shift) + (my>>2)*h->mb_linesize;
uint8_t * src_cb, * src_cr;
int extra_width= h->emu_edge_width;
int extra_height= h->emu_edge_height;
|| full_my < 0-extra_height
|| full_mx + 16/*FIXME*/ > pic_width + extra_width
|| full_my + 16/*FIXME*/ > pic_height + extra_height){
- s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_y - 2 - 2*h->mb_linesize, h->mb_linesize, 16+5, 16+5/*FIXME*/, full_mx-2, full_my-2, pic_width, pic_height);
- src_y= s->edge_emu_buffer + 2 + 2*h->mb_linesize;
+ s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_y - (2 << pixel_shift) - 2*h->mb_linesize, h->mb_linesize, 16+5, 16+5/*FIXME*/, full_mx-2, full_my-2, pic_width, pic_height);
+ src_y= s->edge_emu_buffer + (2 << pixel_shift) + 2*h->mb_linesize;
emu=1;
}
my += 2 * ((s->mb_y & 1) - (pic->reference - 1));
emu |= (my>>3) < 0 || (my>>3) + 8 >= (pic_height>>1);
}
- src_cb= pic->data[1] + (mx>>3) + (my>>3)*h->mb_uvlinesize;
- src_cr= pic->data[2] + (mx>>3) + (my>>3)*h->mb_uvlinesize;
+ src_cb= pic->data[1] + ((mx>>3) << pixel_shift) + (my>>3)*h->mb_uvlinesize;
+ src_cr= pic->data[2] + ((mx>>3) << pixel_shift) + (my>>3)*h->mb_uvlinesize;
if(emu){
s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_cb, h->mb_uvlinesize, 9, 9/*FIXME*/, (mx>>3), (my>>3), pic_width>>1, pic_height>>1);
int x_offset, int y_offset,
qpel_mc_func *qpix_put, h264_chroma_mc_func chroma_put,
qpel_mc_func *qpix_avg, h264_chroma_mc_func chroma_avg,
- int list0, int list1){
+ int list0, int list1, int pixel_shift){
MpegEncContext * const s = &h->s;
qpel_mc_func *qpix_op= qpix_put;
h264_chroma_mc_func chroma_op= chroma_put;
- dest_y += 2*x_offset + 2*y_offset*h-> mb_linesize;
- dest_cb += x_offset + y_offset*h->mb_uvlinesize;
- dest_cr += x_offset + y_offset*h->mb_uvlinesize;
+ dest_y += (2*x_offset << pixel_shift) + 2*y_offset*h-> mb_linesize;
+ dest_cb += ( x_offset << pixel_shift) + y_offset*h->mb_uvlinesize;
+ dest_cr += ( x_offset << pixel_shift) + y_offset*h->mb_uvlinesize;
x_offset += 8*s->mb_x;
y_offset += 8*(s->mb_y >> MB_FIELD);
Picture *ref= &h->ref_list[0][ h->ref_cache[0][ scan8[n] ] ];
mc_dir_part(h, ref, n, square, chroma_height, delta, 0,
dest_y, dest_cb, dest_cr, x_offset, y_offset,
- qpix_op, chroma_op);
+ qpix_op, chroma_op, pixel_shift);
qpix_op= qpix_avg;
chroma_op= chroma_avg;
Picture *ref= &h->ref_list[1][ h->ref_cache[1][ scan8[n] ] ];
mc_dir_part(h, ref, n, square, chroma_height, delta, 1,
dest_y, dest_cb, dest_cr, x_offset, y_offset,
- qpix_op, chroma_op);
+ qpix_op, chroma_op, pixel_shift);
}
}
qpel_mc_func *qpix_put, h264_chroma_mc_func chroma_put,
h264_weight_func luma_weight_op, h264_weight_func chroma_weight_op,
h264_biweight_func luma_weight_avg, h264_biweight_func chroma_weight_avg,
- int list0, int list1){
+ int list0, int list1, int pixel_shift){
MpegEncContext * const s = &h->s;
- dest_y += 2*x_offset + 2*y_offset*h-> mb_linesize;
- dest_cb += x_offset + y_offset*h->mb_uvlinesize;
- dest_cr += x_offset + y_offset*h->mb_uvlinesize;
+ dest_y += (2*x_offset << pixel_shift) + 2*y_offset*h-> mb_linesize;
+ dest_cb += ( x_offset << pixel_shift) + y_offset*h->mb_uvlinesize;
+ dest_cr += ( x_offset << pixel_shift) + y_offset*h->mb_uvlinesize;
x_offset += 8*s->mb_x;
y_offset += 8*(s->mb_y >> MB_FIELD);
/* don't optimize for luma-only case, since B-frames usually
* use implicit weights => chroma too. */
uint8_t *tmp_cb = s->obmc_scratchpad;
- uint8_t *tmp_cr = s->obmc_scratchpad + 8;
+ uint8_t *tmp_cr = s->obmc_scratchpad + (8 << pixel_shift);
uint8_t *tmp_y = s->obmc_scratchpad + 8*h->mb_uvlinesize;
int refn0 = h->ref_cache[0][ scan8[n] ];
int refn1 = h->ref_cache[1][ scan8[n] ];
mc_dir_part(h, &h->ref_list[0][refn0], n, square, chroma_height, delta, 0,
dest_y, dest_cb, dest_cr,
- x_offset, y_offset, qpix_put, chroma_put);
+ x_offset, y_offset, qpix_put, chroma_put, pixel_shift);
mc_dir_part(h, &h->ref_list[1][refn1], n, square, chroma_height, delta, 1,
tmp_y, tmp_cb, tmp_cr,
- x_offset, y_offset, qpix_put, chroma_put);
+ x_offset, y_offset, qpix_put, chroma_put, pixel_shift);
if(h->use_weight == 2){
int weight0 = h->implicit_weight[refn0][refn1][s->mb_y&1];
Picture *ref= &h->ref_list[list][refn];
mc_dir_part(h, ref, n, square, chroma_height, delta, list,
dest_y, dest_cb, dest_cr, x_offset, y_offset,
- qpix_put, chroma_put);
+ qpix_put, chroma_put, pixel_shift);
luma_weight_op(dest_y, h->mb_linesize, h->luma_log2_weight_denom,
h->luma_weight[refn][list][0], h->luma_weight[refn][list][1]);
qpel_mc_func *qpix_put, h264_chroma_mc_func chroma_put,
qpel_mc_func *qpix_avg, h264_chroma_mc_func chroma_avg,
h264_weight_func *weight_op, h264_biweight_func *weight_avg,
- int list0, int list1){
+ int list0, int list1, int pixel_shift){
if((h->use_weight==2 && list0 && list1
&& (h->implicit_weight[ h->ref_cache[0][scan8[n]] ][ h->ref_cache[1][scan8[n]] ][h->s.mb_y&1] != 32))
|| h->use_weight==1)
mc_part_weighted(h, n, square, chroma_height, delta, dest_y, dest_cb, dest_cr,
x_offset, y_offset, qpix_put, chroma_put,
- weight_op[0], weight_op[3], weight_avg[0], weight_avg[3], list0, list1);
+ weight_op[0], weight_op[3], weight_avg[0],
+ weight_avg[3], list0, list1, pixel_shift);
else
mc_part_std(h, n, square, chroma_height, delta, dest_y, dest_cb, dest_cr,
- x_offset, y_offset, qpix_put, chroma_put, qpix_avg, chroma_avg, list0, list1);
+ x_offset, y_offset, qpix_put, chroma_put, qpix_avg,
+ chroma_avg, list0, list1, pixel_shift);
}
-static inline void prefetch_motion(H264Context *h, int list){
+static inline void prefetch_motion(H264Context *h, int list, int pixel_shift){
/* fetch pixels for estimated mv 4 macroblocks ahead
* optimized for 64byte cache lines */
MpegEncContext * const s = &h->s;
const int mx= (h->mv_cache[list][scan8[0]][0]>>2) + 16*s->mb_x + 8;
const int my= (h->mv_cache[list][scan8[0]][1]>>2) + 16*s->mb_y;
uint8_t **src= h->ref_list[list][refn].data;
- int off= mx + (my + (s->mb_x&3)*4)*h->mb_linesize + 64;
+ int off= (mx << pixel_shift) + (my + (s->mb_x&3)*4)*h->mb_linesize + (64 << pixel_shift);
s->dsp.prefetch(src[0]+off, s->linesize, 4);
- off= (mx>>1) + ((my>>1) + (s->mb_x&7))*s->uvlinesize + 64;
+ off= ((mx>>1) << pixel_shift) + ((my>>1) + (s->mb_x&7))*s->uvlinesize + (64 << pixel_shift);
s->dsp.prefetch(src[1]+off, src[2]-src[1], 2);
}
}
-static void hl_motion(H264Context *h, uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
+static av_always_inline void hl_motion(H264Context *h, uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
qpel_mc_func (*qpix_put)[16], h264_chroma_mc_func (*chroma_put),
qpel_mc_func (*qpix_avg)[16], h264_chroma_mc_func (*chroma_avg),
- h264_weight_func *weight_op, h264_biweight_func *weight_avg){
+ h264_weight_func *weight_op, h264_biweight_func *weight_avg,
+ int pixel_shift){
MpegEncContext * const s = &h->s;
const int mb_xy= h->mb_xy;
const int mb_type= s->current_picture.mb_type[mb_xy];
assert(IS_INTER(mb_type));
- prefetch_motion(h, 0);
+ prefetch_motion(h, 0, pixel_shift);
if(IS_16X16(mb_type)){
mc_part(h, 0, 1, 8, 0, dest_y, dest_cb, dest_cr, 0, 0,
qpix_put[0], chroma_put[0], qpix_avg[0], chroma_avg[0],
weight_op, weight_avg,
- IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
+ IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
+ pixel_shift);
}else if(IS_16X8(mb_type)){
- mc_part(h, 0, 0, 4, 8, dest_y, dest_cb, dest_cr, 0, 0,
+ mc_part(h, 0, 0, 4, 8 << pixel_shift, dest_y, dest_cb, dest_cr, 0, 0,
qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
&weight_op[1], &weight_avg[1],
- IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
- mc_part(h, 8, 0, 4, 8, dest_y, dest_cb, dest_cr, 0, 4,
+ IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
+ pixel_shift);
+ mc_part(h, 8, 0, 4, 8 << pixel_shift, dest_y, dest_cb, dest_cr, 0, 4,
qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
&weight_op[1], &weight_avg[1],
- IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1));
+ IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1),
+ pixel_shift);
}else if(IS_8X16(mb_type)){
mc_part(h, 0, 0, 8, 8*h->mb_linesize, dest_y, dest_cb, dest_cr, 0, 0,
qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
&weight_op[2], &weight_avg[2],
- IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
+ IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
+ pixel_shift);
mc_part(h, 4, 0, 8, 8*h->mb_linesize, dest_y, dest_cb, dest_cr, 4, 0,
qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
&weight_op[2], &weight_avg[2],
- IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1));
+ IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1),
+ pixel_shift);
}else{
int i;
mc_part(h, n, 1, 4, 0, dest_y, dest_cb, dest_cr, x_offset, y_offset,
qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
&weight_op[3], &weight_avg[3],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
}else if(IS_SUB_8X4(sub_mb_type)){
- mc_part(h, n , 0, 2, 4, dest_y, dest_cb, dest_cr, x_offset, y_offset,
+ mc_part(h, n , 0, 2, 4 << pixel_shift, dest_y, dest_cb, dest_cr, x_offset, y_offset,
qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
&weight_op[4], &weight_avg[4],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
- mc_part(h, n+2, 0, 2, 4, dest_y, dest_cb, dest_cr, x_offset, y_offset+2,
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
+ mc_part(h, n+2, 0, 2, 4 << pixel_shift, dest_y, dest_cb, dest_cr, x_offset, y_offset+2,
qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
&weight_op[4], &weight_avg[4],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
}else if(IS_SUB_4X8(sub_mb_type)){
mc_part(h, n , 0, 4, 4*h->mb_linesize, dest_y, dest_cb, dest_cr, x_offset, y_offset,
qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
&weight_op[5], &weight_avg[5],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
mc_part(h, n+1, 0, 4, 4*h->mb_linesize, dest_y, dest_cb, dest_cr, x_offset+2, y_offset,
qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
&weight_op[5], &weight_avg[5],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
}else{
int j;
assert(IS_SUB_4X4(sub_mb_type));
mc_part(h, n+j, 1, 2, 0, dest_y, dest_cb, dest_cr, sub_x_offset, sub_y_offset,
qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
&weight_op[6], &weight_avg[6],
- IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+ IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
+ pixel_shift);
}
}
}
}
- prefetch_motion(h, 1);
+ prefetch_motion(h, 1, pixel_shift);
}
+#define hl_motion_fn(sh, bits) \
+static av_always_inline void hl_motion_ ## bits(H264Context *h, \
+ uint8_t *dest_y, \
+ uint8_t *dest_cb, uint8_t *dest_cr, \
+ qpel_mc_func (*qpix_put)[16], \
+ h264_chroma_mc_func (*chroma_put), \
+ qpel_mc_func (*qpix_avg)[16], \
+ h264_chroma_mc_func (*chroma_avg), \
+ h264_weight_func *weight_op, \
+ h264_biweight_func *weight_avg) \
+{ \
+ hl_motion(h, dest_y, dest_cb, dest_cr, qpix_put, chroma_put, \
+ qpix_avg, chroma_avg, weight_op, weight_avg, sh); \
+}
+hl_motion_fn(0, 8);
+hl_motion_fn(1, 16);
static void free_tables(H264Context *h, int free_rbsp){
int i;
static void init_dequant8_coeff_table(H264Context *h){
int i,q,x;
+ const int max_qp = 51 + 6*(h->sps.bit_depth_luma-8);
h->dequant8_coeff[0] = h->dequant8_buffer[0];
h->dequant8_coeff[1] = h->dequant8_buffer[1];
break;
}
- for(q=0; q<52; q++){
+ for(q=0; q<max_qp+1; q++){
int shift = div6[q];
int idx = rem6[q];
for(x=0; x<64; x++)
static void init_dequant4_coeff_table(H264Context *h){
int i,j,q,x;
+ const int max_qp = 51 + 6*(h->sps.bit_depth_luma-8);
for(i=0; i<6; i++ ){
h->dequant4_coeff[i] = h->dequant4_buffer[i];
for(j=0; j<i; j++){
if(j<i)
continue;
- for(q=0; q<52; q++){
+ for(q=0; q<max_qp+1; q++){
int shift = div6[q] + 2;
int idx = rem6[q];
for(x=0; x<16; x++)
dst->list_counts = src->list_counts;
dst->s.obmc_scratchpad = NULL;
- ff_h264_pred_init(&dst->hpc, src->s.codec_id);
+ ff_h264_pred_init(&dst->hpc, src->s.codec_id, src->sps.bit_depth_luma);
}
/**
* Allocate buffers which are not shared amongst multiple threads.
*/
static int context_init(H264Context *h){
- FF_ALLOCZ_OR_GOTO(h->s.avctx, h->top_borders[0], h->s.mb_width * (16+8+8) * sizeof(uint8_t), fail)
- FF_ALLOCZ_OR_GOTO(h->s.avctx, h->top_borders[1], h->s.mb_width * (16+8+8) * sizeof(uint8_t), fail)
+ FF_ALLOCZ_OR_GOTO(h->s.avctx, h->top_borders[0], h->s.mb_width * (16+8+8) * sizeof(uint8_t)*2, fail)
+ FF_ALLOCZ_OR_GOTO(h->s.avctx, h->top_borders[1], h->s.mb_width * (16+8+8) * sizeof(uint8_t)*2, fail)
h->ref_cache[0][scan8[5 ]+1] = h->ref_cache[0][scan8[7 ]+1] = h->ref_cache[0][scan8[13]+1] =
h->ref_cache[1][scan8[5 ]+1] = h->ref_cache[1][scan8[7 ]+1] = h->ref_cache[1][scan8[13]+1] = PART_NOT_AVAILABLE;
s->height = s->avctx->height;
s->codec_id= s->avctx->codec->id;
- ff_h264dsp_init(&h->h264dsp);
- ff_h264_pred_init(&h->hpc, s->codec_id);
+ ff_h264dsp_init(&h->h264dsp, 8);
+ ff_h264_pred_init(&h->hpc, s->codec_id, 8);
h->dequant_coeff_pps= -1;
s->unrestricted_mv=1;
ff_h264_decode_init_vlc();
+ h->pixel_shift = 0;
+ h->sps.bit_depth_luma = avctx->bits_per_raw_sample = 8;
+
h->thread_context[0] = h;
h->outputed_poc = INT_MIN;
h->prev_poc_msb= 1<<16;
int ff_h264_frame_start(H264Context *h){
MpegEncContext * const s = &h->s;
int i;
+ const int pixel_shift = h->pixel_shift;
if(MPV_frame_start(s, s->avctx) < 0)
return -1;
assert(s->linesize && s->uvlinesize);
for(i=0; i<16; i++){
- h->block_offset[i]= 4*((scan8[i] - scan8[0])&7) + 4*s->linesize*((scan8[i] - scan8[0])>>3);
- h->block_offset[24+i]= 4*((scan8[i] - scan8[0])&7) + 8*s->linesize*((scan8[i] - scan8[0])>>3);
+ h->block_offset[i]= (4*((scan8[i] - scan8[0])&7) << pixel_shift) + 4*s->linesize*((scan8[i] - scan8[0])>>3);
+ h->block_offset[24+i]= (4*((scan8[i] - scan8[0])&7) << pixel_shift) + 8*s->linesize*((scan8[i] - scan8[0])>>3);
}
for(i=0; i<4; i++){
h->block_offset[16+i]=
- h->block_offset[20+i]= 4*((scan8[i] - scan8[0])&7) + 4*s->uvlinesize*((scan8[i] - scan8[0])>>3);
+ h->block_offset[20+i]= (4*((scan8[i] - scan8[0])&7) << pixel_shift) + 4*s->uvlinesize*((scan8[i] - scan8[0])>>3);
h->block_offset[24+16+i]=
- h->block_offset[24+20+i]= 4*((scan8[i] - scan8[0])&7) + 8*s->uvlinesize*((scan8[i] - scan8[0])>>3);
+ h->block_offset[24+20+i]= (4*((scan8[i] - scan8[0])&7) << pixel_shift) + 8*s->uvlinesize*((scan8[i] - scan8[0])>>3);
}
/* can't be in alloc_tables because linesize isn't known there.
MpegEncContext * const s = &h->s;
uint8_t *top_border;
int top_idx = 1;
+ const int pixel_shift = h->pixel_shift;
src_y -= linesize;
src_cb -= uvlinesize;
if(!MB_MBAFF){
top_border = h->top_borders[0][s->mb_x];
AV_COPY128(top_border, src_y + 15*linesize);
+ if (pixel_shift)
+ AV_COPY128(top_border+16, src_y+15*linesize+16);
if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if (pixel_shift) {
+ AV_COPY128(top_border+32, src_cb+7*uvlinesize);
+ AV_COPY128(top_border+48, src_cr+7*uvlinesize);
+ } else {
AV_COPY64(top_border+16, src_cb+7*uvlinesize);
AV_COPY64(top_border+24, src_cr+7*uvlinesize);
+ }
}
}
}else if(MB_MBAFF){
// There are two lines saved, the line above the the top macroblock of a pair,
// and the line above the bottom macroblock
AV_COPY128(top_border, src_y + 16*linesize);
+ if (pixel_shift)
+ AV_COPY128(top_border+16, src_y+16*linesize+16);
if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
+ if (pixel_shift) {
+ AV_COPY128(top_border+32, src_cb+8*uvlinesize);
+ AV_COPY128(top_border+48, src_cr+8*uvlinesize);
+ } else {
AV_COPY64(top_border+16, src_cb+8*uvlinesize);
AV_COPY64(top_border+24, src_cr+8*uvlinesize);
+ }
}
}
-static inline void xchg_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg, int simple){
+static inline void xchg_mb_border(H264Context *h, uint8_t *src_y,
+ uint8_t *src_cb, uint8_t *src_cr,
+ int linesize, int uvlinesize,
+ int xchg, int simple, int pixel_shift){
MpegEncContext * const s = &h->s;
- int deblock_left;
+ int deblock_topleft;
int deblock_top;
int top_idx = 1;
uint8_t *top_border_m1;
}
if(h->deblocking_filter == 2) {
- deblock_left = h->left_type[0];
- deblock_top = h->top_type;
+ deblock_topleft = h->slice_table[h->mb_xy - 1 - s->mb_stride] == h->slice_num;
+ deblock_top = h->top_type;
} else {
- deblock_left = (s->mb_x > 0);
- deblock_top = (s->mb_y > !!MB_FIELD);
+ deblock_topleft = (s->mb_x > 0);
+ deblock_top = (s->mb_y > !!MB_FIELD);
}
- src_y -= linesize + 1;
- src_cb -= uvlinesize + 1;
- src_cr -= uvlinesize + 1;
+ src_y -= linesize + 1 + pixel_shift;
+ src_cb -= uvlinesize + 1 + pixel_shift;
+ src_cr -= uvlinesize + 1 + pixel_shift;
top_border_m1 = h->top_borders[top_idx][s->mb_x-1];
top_border = h->top_borders[top_idx][s->mb_x];
#define XCHG(a,b,xchg)\
+ if (pixel_shift) {\
+ if (xchg) {\
+ AV_SWAP64(b+0,a+0);\
+ AV_SWAP64(b+8,a+8);\
+ } else {\
+ AV_COPY128(b,a); \
+ }\
+ } else \
if (xchg) AV_SWAP64(b,a);\
else AV_COPY64(b,a);
if(deblock_top){
- if(deblock_left){
- XCHG(top_border_m1+8, src_y -7, 1);
+ if(deblock_topleft){
+ XCHG(top_border_m1 + (8 << pixel_shift), src_y - (7 << pixel_shift), 1);
}
- XCHG(top_border+0, src_y +1, xchg);
- XCHG(top_border+8, src_y +9, 1);
+ XCHG(top_border + (0 << pixel_shift), src_y + (1 << pixel_shift), xchg);
+ XCHG(top_border + (8 << pixel_shift), src_y + (9 << pixel_shift), 1);
if(s->mb_x+1 < s->mb_width){
- XCHG(h->top_borders[top_idx][s->mb_x+1], src_y +17, 1);
+ XCHG(h->top_borders[top_idx][s->mb_x+1], src_y + (17 << pixel_shift), 1);
}
}
-
if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
if(deblock_top){
- if(deblock_left){
- XCHG(top_border_m1+16, src_cb -7, 1);
- XCHG(top_border_m1+24, src_cr -7, 1);
+ if(deblock_topleft){
+ XCHG(top_border_m1 + (16 << pixel_shift), src_cb - (7 << pixel_shift), 1);
+ XCHG(top_border_m1 + (24 << pixel_shift), src_cr - (7 << pixel_shift), 1);
}
- XCHG(top_border+16, src_cb+1, 1);
- XCHG(top_border+24, src_cr+1, 1);
+ XCHG(top_border + (16 << pixel_shift), src_cb+1+pixel_shift, 1);
+ XCHG(top_border + (24 << pixel_shift), src_cr+1+pixel_shift, 1);
}
}
}
-static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){
+static av_always_inline int dctcoef_get(DCTELEM *mb, int high_bit_depth, int index) {
+ if (high_bit_depth) {
+ return AV_RN32A(((int32_t*)mb) + index);
+ } else
+ return AV_RN16A(mb + index);
+}
+
+static av_always_inline void dctcoef_set(DCTELEM *mb, int high_bit_depth, int index, int value) {
+ if (high_bit_depth) {
+ AV_WN32A(((int32_t*)mb) + index, value);
+ } else
+ AV_WN16A(mb + index, value);
+}
+
+static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, int pixel_shift){
MpegEncContext * const s = &h->s;
const int mb_x= s->mb_x;
const int mb_y= s->mb_y;
void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
void (*idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
- dest_y = s->current_picture.data[0] + (mb_x + mb_y * s->linesize ) * 16;
- dest_cb = s->current_picture.data[1] + (mb_x + mb_y * s->uvlinesize) * 8;
- dest_cr = s->current_picture.data[2] + (mb_x + mb_y * s->uvlinesize) * 8;
+ dest_y = s->current_picture.data[0] + ((mb_x << pixel_shift) + mb_y * s->linesize ) * 16;
+ dest_cb = s->current_picture.data[1] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * 8;
+ dest_cr = s->current_picture.data[2] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * 8;
- s->dsp.prefetch(dest_y + (s->mb_x&3)*4*s->linesize + 64, s->linesize, 4);
- s->dsp.prefetch(dest_cb + (s->mb_x&7)*s->uvlinesize + 64, dest_cr - dest_cb, 2);
+ s->dsp.prefetch(dest_y + (s->mb_x&3)*4*s->linesize + (64 << pixel_shift), s->linesize, 4);
+ s->dsp.prefetch(dest_cb + (s->mb_x&7)*s->uvlinesize + (64 << pixel_shift), dest_cr - dest_cb, 2);
h->list_counts[mb_xy]= h->list_count;
}
if (!simple && IS_INTRA_PCM(mb_type)) {
+ if (pixel_shift) {
+ const int bit_depth = h->sps.bit_depth_luma;
+ int j;
+ GetBitContext gb;
+ init_get_bits(&gb, (uint8_t*)h->mb, 384*bit_depth);
+
+ for (i = 0; i < 16; i++) {
+ uint16_t *tmp_y = (uint16_t*)(dest_y + i*linesize);
+ for (j = 0; j < 16; j++)
+ tmp_y[j] = get_bits(&gb, bit_depth);
+ }
+ for (i = 0; i < 8; i++) {
+ uint16_t *tmp_cb = (uint16_t*)(dest_cb + i*uvlinesize);
+ for (j = 0; j < 8; j++)
+ tmp_cb[j] = get_bits(&gb, bit_depth);
+ }
+ for (i = 0; i < 8; i++) {
+ uint16_t *tmp_cr = (uint16_t*)(dest_cr + i*uvlinesize);
+ for (j = 0; j < 8; j++)
+ tmp_cr[j] = get_bits(&gb, bit_depth);
+ }
+ } else {
for (i=0; i<16; i++) {
memcpy(dest_y + i* linesize, h->mb + i*8, 16);
}
memcpy(dest_cb+ i*uvlinesize, h->mb + 128 + i*4, 8);
memcpy(dest_cr+ i*uvlinesize, h->mb + 160 + i*4, 8);
}
+ }
} else {
if(IS_INTRA(mb_type)){
if(h->deblocking_filter)
- xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1, simple);
+ xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1, simple, pixel_shift);
if(simple || !CONFIG_GRAY || !(s->flags&CODEC_FLAG_GRAY)){
h->hpc.pred8x8[ h->chroma_pred_mode ](dest_cb, uvlinesize);
uint8_t * const ptr= dest_y + block_offset[i];
const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ];
if(transform_bypass && h->sps.profile_idc==244 && dir<=1){
- h->hpc.pred8x8l_add[dir](ptr, h->mb + i*16, linesize);
+ h->hpc.pred8x8l_add[dir](ptr, h->mb + (i*16 << pixel_shift), linesize);
}else{
const int nnz = h->non_zero_count_cache[ scan8[i] ];
h->hpc.pred8x8l[ dir ](ptr, (h->topleft_samples_available<<i)&0x8000,
(h->topright_samples_available<<i)&0x4000, linesize);
if(nnz){
- if(nnz == 1 && h->mb[i*16])
- idct_dc_add(ptr, h->mb + i*16, linesize);
+ if(nnz == 1 && dctcoef_get(h->mb, pixel_shift, i*16))
+ idct_dc_add(ptr, h->mb + (i*16 << pixel_shift), linesize);
else
- idct_add (ptr, h->mb + i*16, linesize);
+ idct_add (ptr, h->mb + (i*16 << pixel_shift), linesize);
}
}
}
const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ];
if(transform_bypass && h->sps.profile_idc==244 && dir<=1){
- h->hpc.pred4x4_add[dir](ptr, h->mb + i*16, linesize);
+ h->hpc.pred4x4_add[dir](ptr, h->mb + (i*16 << pixel_shift), linesize);
}else{
uint8_t *topright;
int nnz, tr;
+ uint64_t tr_high;
if(dir == DIAG_DOWN_LEFT_PRED || dir == VERT_LEFT_PRED){
const int topright_avail= (h->topright_samples_available<<i)&0x8000;
assert(mb_y || linesize <= block_offset[i]);
if(!topright_avail){
+ if (pixel_shift) {
+ tr_high= ((uint16_t*)ptr)[3 - linesize/2]*0x0001000100010001ULL;
+ topright= (uint8_t*) &tr_high;
+ } else {
tr= ptr[3 - linesize]*0x01010101;
topright= (uint8_t*) &tr;
+ }
}else
- topright= ptr + 4 - linesize;
+ topright= ptr + (4 << pixel_shift) - linesize;
}else
topright= NULL;
nnz = h->non_zero_count_cache[ scan8[i] ];
if(nnz){
if(is_h264){
- if(nnz == 1 && h->mb[i*16])
- idct_dc_add(ptr, h->mb + i*16, linesize);
+ if(nnz == 1 && dctcoef_get(h->mb, pixel_shift, i*16))
+ idct_dc_add(ptr, h->mb + (i*16 << pixel_shift), linesize);
else
- idct_add (ptr, h->mb + i*16, linesize);
+ idct_add (ptr, h->mb + (i*16 << pixel_shift), linesize);
}else
ff_svq3_add_idct_c(ptr, h->mb + i*16, linesize, s->qscale, 0);
}
static const uint8_t dc_mapping[16] = { 0*16, 1*16, 4*16, 5*16, 2*16, 3*16, 6*16, 7*16,
8*16, 9*16,12*16,13*16,10*16,11*16,14*16,15*16};
for(i = 0; i < 16; i++)
- h->mb[dc_mapping[i]] = h->mb_luma_dc[i];
+ dctcoef_set(h->mb, pixel_shift, dc_mapping[i], dctcoef_get(h->mb_luma_dc, pixel_shift, i));
}
}
}else
ff_svq3_luma_dc_dequant_idct_c(h->mb, h->mb_luma_dc, s->qscale);
}
if(h->deblocking_filter)
- xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0, simple);
+ xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0, simple, pixel_shift);
}else if(is_h264){
- hl_motion(h, dest_y, dest_cb, dest_cr,
- s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
- s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
- h->h264dsp.weight_h264_pixels_tab, h->h264dsp.biweight_h264_pixels_tab);
+ if (pixel_shift) {
+ hl_motion_16(h, dest_y, dest_cb, dest_cr,
+ s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+ s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
+ h->h264dsp.weight_h264_pixels_tab,
+ h->h264dsp.biweight_h264_pixels_tab);
+ } else
+ hl_motion_8(h, dest_y, dest_cb, dest_cr,
+ s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+ s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
+ h->h264dsp.weight_h264_pixels_tab,
+ h->h264dsp.biweight_h264_pixels_tab);
}
h->hpc.pred16x16_add[h->intra16x16_pred_mode](dest_y, block_offset, h->mb, linesize);
}else{
for(i=0; i<16; i++){
- if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16])
- s->dsp.add_pixels4(dest_y + block_offset[i], h->mb + i*16, linesize);
+ if(h->non_zero_count_cache[ scan8[i] ] || dctcoef_get(h->mb, pixel_shift, i*16))
+ s->dsp.add_pixels4(dest_y + block_offset[i], h->mb + (i*16 << pixel_shift), linesize);
}
}
}else{
idct_add= IS_8x8DCT(mb_type) ? s->dsp.add_pixels8 : s->dsp.add_pixels4;
for(i=0; i<16; i+=di){
if(h->non_zero_count_cache[ scan8[i] ]){
- idct_add(dest_y + block_offset[i], h->mb + i*16, linesize);
+ idct_add(dest_y + block_offset[i], h->mb + (i*16 << pixel_shift), linesize);
}
}
}else{
uint8_t *dest[2] = {dest_cb, dest_cr};
if(transform_bypass){
if(IS_INTRA(mb_type) && h->sps.profile_idc==244 && (h->chroma_pred_mode==VERT_PRED8x8 || h->chroma_pred_mode==HOR_PRED8x8)){
- h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0], block_offset + 16, h->mb + 16*16, uvlinesize);
- h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1], block_offset + 20, h->mb + 20*16, uvlinesize);
+ h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0], block_offset + 16, h->mb + (16*16 << pixel_shift), uvlinesize);
+ h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1], block_offset + 20, h->mb + (20*16 << pixel_shift), uvlinesize);
}else{
idct_add = s->dsp.add_pixels4;
for(i=16; i<16+8; i++){
- if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16])
- idct_add (dest[(i&4)>>2] + block_offset[i], h->mb + i*16, uvlinesize);
+ if(h->non_zero_count_cache[ scan8[i] ] || dctcoef_get(h->mb, pixel_shift, i*16))
+ idct_add (dest[(i&4)>>2] + block_offset[i], h->mb + (i*16 << pixel_shift), uvlinesize);
}
}
}else{
if(is_h264){
if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+0] ])
- chroma_dc_dequant_idct_c(h->mb + 16*16 , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+ h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16*16 << pixel_shift) , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+1] ])
- chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+ h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + ((16*16+4*16) << pixel_shift), h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
h->h264dsp.h264_idct_add8(dest, block_offset,
h->mb, uvlinesize,
h->non_zero_count_cache);
}else{
- chroma_dc_dequant_idct_c(h->mb + 16*16 , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
- chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+ h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16 , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+ h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
for(i=16; i<16+8; i++){
if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]){
uint8_t * const ptr= dest[(i&4)>>2] + block_offset[i];
- ff_svq3_add_idct_c(ptr, h->mb + i*16, uvlinesize, ff_h264_chroma_qp[s->qscale + 12] - 12, 2);
+ ff_svq3_add_idct_c(ptr, h->mb + i*16, uvlinesize, ff_h264_chroma_qp[0][s->qscale + 12] - 12, 2);
}
}
}
/**
* Process a macroblock; this case avoids checks for expensive uncommon cases.
*/
-static void hl_decode_mb_simple(H264Context *h){
- hl_decode_mb_internal(h, 1);
+#define hl_decode_mb_simple(sh, bits) \
+static void hl_decode_mb_simple_ ## bits(H264Context *h){ \
+ hl_decode_mb_internal(h, 1, sh); \
}
+hl_decode_mb_simple(0, 8);
+hl_decode_mb_simple(1, 16);
/**
* Process a macroblock; this handles edge cases, such as interlacing.
*/
static void av_noinline hl_decode_mb_complex(H264Context *h){
- hl_decode_mb_internal(h, 0);
+ hl_decode_mb_internal(h, 0, h->pixel_shift);
}
void ff_h264_hl_decode_mb(H264Context *h){
const int mb_type= s->current_picture.mb_type[mb_xy];
int is_complex = CONFIG_SMALL || h->is_complex || IS_INTRA_PCM(mb_type) || s->qscale == 0;
- if (is_complex)
+ if (is_complex) {
hl_decode_mb_complex(h);
- else hl_decode_mb_simple(h);
+ } else if (h->pixel_shift) {
+ hl_decode_mb_simple_16(h);
+ } else
+ hl_decode_mb_simple_8(h);
}
static int pred_weight_table(H264Context *h){
}
}
}
- if(h->slice_type_nos != FF_B_TYPE) break;
+ if(h->slice_type_nos != AV_PICTURE_TYPE_B) break;
}
h->use_weight= h->use_weight || h->use_weight_chroma;
return 0;
h->slice_type_fixed=0;
slice_type= golomb_to_pict_type[ slice_type ];
- if (slice_type == FF_I_TYPE
+ if (slice_type == AV_PICTURE_TYPE_I
|| (h0->current_slice != 0 && slice_type == h0->last_slice_type) ) {
default_ref_list_done = 1;
}
s->avctx->sample_aspect_ratio= h->sps.sar;
av_assert0(s->avctx->sample_aspect_ratio.den);
+ h->s.avctx->coded_width = 16*s->mb_width;
+ h->s.avctx->coded_height = 16*s->mb_height;
+
if(h->sps.video_signal_type_present_flag){
s->avctx->color_range = h->sps.full_range ? AVCOL_RANGE_JPEG : AVCOL_RANGE_MPEG;
if(h->sps.colour_description_present_flag){
av_reduce(&s->avctx->time_base.num, &s->avctx->time_base.den,
h->sps.num_units_in_tick, den, 1<<30);
}
+
+ switch (h->sps.bit_depth_luma) {
+ case 9 :
+ s->avctx->pix_fmt = PIX_FMT_YUV420P9;
+ break;
+ case 10 :
+ s->avctx->pix_fmt = PIX_FMT_YUV420P10;
+ break;
+ default:
s->avctx->pix_fmt = s->avctx->get_format(s->avctx,
s->avctx->codec->pix_fmts ?
s->avctx->codec->pix_fmts :
s->avctx->color_range == AVCOL_RANGE_JPEG ?
hwaccel_pixfmt_list_h264_jpeg_420 :
ff_hwaccel_pixfmt_list_420);
+ }
+
s->avctx->hwaccel = ff_find_hwaccel(s->avctx->codec->id, s->avctx->pix_fmt);
if (MPV_common_init(s) < 0)
c->h264dsp = h->h264dsp;
c->sps = h->sps;
c->pps = h->pps;
+ c->pixel_shift = h->pixel_shift;
init_scan_tables(c);
clone_tables(c, h, i);
}
h->ref_count[0]= h->pps.ref_count[0];
h->ref_count[1]= h->pps.ref_count[1];
- if(h->slice_type_nos != FF_I_TYPE){
- if(h->slice_type_nos == FF_B_TYPE){
+ if(h->slice_type_nos != AV_PICTURE_TYPE_I){
+ if(h->slice_type_nos == AV_PICTURE_TYPE_B){
h->direct_spatial_mv_pred= get_bits1(&s->gb);
}
num_ref_idx_active_override_flag= get_bits1(&s->gb);
if(num_ref_idx_active_override_flag){
h->ref_count[0]= get_ue_golomb(&s->gb) + 1;
- if(h->slice_type_nos==FF_B_TYPE)
+ if(h->slice_type_nos==AV_PICTURE_TYPE_B)
h->ref_count[1]= get_ue_golomb(&s->gb) + 1;
if(h->ref_count[0]-1 > 32-1 || h->ref_count[1]-1 > 32-1){
return -1;
}
}
- if(h->slice_type_nos == FF_B_TYPE)
+ if(h->slice_type_nos == AV_PICTURE_TYPE_B)
h->list_count= 2;
else
h->list_count= 1;
ff_h264_fill_default_ref_list(h);
}
- if(h->slice_type_nos!=FF_I_TYPE && ff_h264_decode_ref_pic_list_reordering(h) < 0)
+ if(h->slice_type_nos!=AV_PICTURE_TYPE_I && ff_h264_decode_ref_pic_list_reordering(h) < 0)
return -1;
- if(h->slice_type_nos!=FF_I_TYPE){
+ if(h->slice_type_nos!=AV_PICTURE_TYPE_I){
s->last_picture_ptr= &h->ref_list[0][0];
ff_copy_picture(&s->last_picture, s->last_picture_ptr);
}
- if(h->slice_type_nos==FF_B_TYPE){
+ if(h->slice_type_nos==AV_PICTURE_TYPE_B){
s->next_picture_ptr= &h->ref_list[1][0];
ff_copy_picture(&s->next_picture, s->next_picture_ptr);
}
- if( (h->pps.weighted_pred && h->slice_type_nos == FF_P_TYPE )
- || (h->pps.weighted_bipred_idc==1 && h->slice_type_nos== FF_B_TYPE ) )
+ if( (h->pps.weighted_pred && h->slice_type_nos == AV_PICTURE_TYPE_P )
+ || (h->pps.weighted_bipred_idc==1 && h->slice_type_nos== AV_PICTURE_TYPE_B ) )
pred_weight_table(h);
- else if(h->pps.weighted_bipred_idc==2 && h->slice_type_nos== FF_B_TYPE){
+ else if(h->pps.weighted_bipred_idc==2 && h->slice_type_nos== AV_PICTURE_TYPE_B){
implicit_weight_table(h, -1);
}else {
h->use_weight = 0;
if(FRAME_MBAFF){
ff_h264_fill_mbaff_ref_list(h);
- if(h->pps.weighted_bipred_idc==2 && h->slice_type_nos== FF_B_TYPE){
+ if(h->pps.weighted_bipred_idc==2 && h->slice_type_nos== AV_PICTURE_TYPE_B){
implicit_weight_table(h, 0);
implicit_weight_table(h, 1);
}
}
- if(h->slice_type_nos==FF_B_TYPE && !h->direct_spatial_mv_pred)
+ if(h->slice_type_nos==AV_PICTURE_TYPE_B && !h->direct_spatial_mv_pred)
ff_h264_direct_dist_scale_factor(h);
ff_h264_direct_ref_list_init(h);
- if( h->slice_type_nos != FF_I_TYPE && h->pps.cabac ){
+ if( h->slice_type_nos != AV_PICTURE_TYPE_I && h->pps.cabac ){
tmp = get_ue_golomb_31(&s->gb);
if(tmp > 2){
av_log(s->avctx, AV_LOG_ERROR, "cabac_init_idc overflow\n");
h->last_qscale_diff = 0;
tmp = h->pps.init_qp + get_se_golomb(&s->gb);
- if(tmp>51){
+ if(tmp>51+6*(h->sps.bit_depth_luma-8)){
av_log(s->avctx, AV_LOG_ERROR, "QP %u out of range\n", tmp);
return -1;
}
h->chroma_qp[0] = get_chroma_qp(h, 0, s->qscale);
h->chroma_qp[1] = get_chroma_qp(h, 1, s->qscale);
//FIXME qscale / qp ... stuff
- if(h->slice_type == FF_SP_TYPE){
+ if(h->slice_type == AV_PICTURE_TYPE_SP){
get_bits1(&s->gb); /* sp_for_switch_flag */
}
- if(h->slice_type==FF_SP_TYPE || h->slice_type == FF_SI_TYPE){
+ if(h->slice_type==AV_PICTURE_TYPE_SP || h->slice_type == AV_PICTURE_TYPE_SI){
get_se_golomb(&s->gb); /* slice_qs_delta */
}
}
if( s->avctx->skip_loop_filter >= AVDISCARD_ALL
- ||(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY && h->slice_type_nos != FF_I_TYPE)
- ||(s->avctx->skip_loop_filter >= AVDISCARD_BIDIR && h->slice_type_nos == FF_B_TYPE)
+ ||(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY && h->slice_type_nos != AV_PICTURE_TYPE_I)
+ ||(s->avctx->skip_loop_filter >= AVDISCARD_BIDIR && h->slice_type_nos == AV_PICTURE_TYPE_B)
||(s->avctx->skip_loop_filter >= AVDISCARD_NONREF && h->nal_ref_idc == 0))
h->deblocking_filter= 0;
h->slice_num,
(s->picture_structure==PICT_FRAME ? "F" : s->picture_structure==PICT_TOP_FIELD ? "T" : "B"),
first_mb_in_slice,
- av_get_pict_type_char(h->slice_type), h->slice_type_fixed ? " fix" : "", h->nal_unit_type == NAL_IDR_SLICE ? " IDR" : "",
+ av_get_picture_type_char(h->slice_type), h->slice_type_fixed ? " fix" : "", h->nal_unit_type == NAL_IDR_SLICE ? " IDR" : "",
pps_id, h->frame_num,
s->current_picture_ptr->field_poc[0], s->current_picture_ptr->field_poc[1],
h->ref_count[0], h->ref_count[1],
h->deblocking_filter, h->slice_alpha_c0_offset/2-26, h->slice_beta_offset/2-26,
h->use_weight,
h->use_weight==1 && h->use_weight_chroma ? "c" : "",
- h->slice_type == FF_B_TYPE ? (h->direct_spatial_mv_pred ? "SPAT" : "TEMP") : ""
+ h->slice_type == AV_PICTURE_TYPE_B ? (h->direct_spatial_mv_pred ? "SPAT" : "TEMP") : ""
);
}
int ff_h264_get_slice_type(const H264Context *h)
{
switch (h->slice_type) {
- case FF_P_TYPE: return 0;
- case FF_B_TYPE: return 1;
- case FF_I_TYPE: return 2;
- case FF_SP_TYPE: return 3;
- case FF_SI_TYPE: return 4;
+ case AV_PICTURE_TYPE_P: return 0;
+ case AV_PICTURE_TYPE_B: return 1;
+ case AV_PICTURE_TYPE_I: return 2;
+ case AV_PICTURE_TYPE_SP: return 3;
+ case AV_PICTURE_TYPE_SI: return 4;
default: return -1;
}
}
return 0;
}
-static void loop_filter(H264Context *h){
+static void loop_filter(H264Context *h, int start_x, int end_x){
MpegEncContext * const s = &h->s;
uint8_t *dest_y, *dest_cb, *dest_cr;
int linesize, uvlinesize, mb_x, mb_y;
const int end_mb_y= s->mb_y + FRAME_MBAFF;
const int old_slice_type= h->slice_type;
+ const int pixel_shift = h->pixel_shift;
if(h->deblocking_filter) {
- for(mb_x= 0; mb_x<s->mb_width; mb_x++){
+ for(mb_x= start_x; mb_x<end_x; mb_x++){
for(mb_y=end_mb_y - FRAME_MBAFF; mb_y<= end_mb_y; mb_y++){
int mb_xy, mb_type;
mb_xy = h->mb_xy = mb_x + mb_y*s->mb_stride;
s->mb_x= mb_x;
s->mb_y= mb_y;
- dest_y = s->current_picture.data[0] + (mb_x + mb_y * s->linesize ) * 16;
- dest_cb = s->current_picture.data[1] + (mb_x + mb_y * s->uvlinesize) * 8;
- dest_cr = s->current_picture.data[2] + (mb_x + mb_y * s->uvlinesize) * 8;
+ dest_y = s->current_picture.data[0] + ((mb_x << pixel_shift) + mb_y * s->linesize ) * 16;
+ dest_cb = s->current_picture.data[1] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * 8;
+ dest_cr = s->current_picture.data[2] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * 8;
//FIXME simplify above
if (MB_FIELD) {
}
}
h->slice_type= old_slice_type;
- s->mb_x= 0;
+ s->mb_x= end_x;
s->mb_y= end_mb_y - FRAME_MBAFF;
h->chroma_qp[0] = get_chroma_qp(h, 0, s->qscale);
h->chroma_qp[1] = get_chroma_qp(h, 1, s->qscale);
H264Context *h = *(void**)arg;
MpegEncContext * const s = &h->s;
const int part_mask= s->partitioned_frame ? (AC_END|AC_ERROR) : 0x7F;
+ int lf_x_start = s->mb_x;
s->mb_skip_run= -1;
if((s->workaround_bugs & FF_BUG_TRUNCATED) && h->cabac.bytestream > h->cabac.bytestream_end + 2){
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
+ if (s->mb_x >= lf_x_start) loop_filter(h, lf_x_start, s->mb_x + 1);
return 0;
}
if( ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 2) {
}
if( ++s->mb_x >= s->mb_width ) {
- s->mb_x = 0;
- loop_filter(h);
+ loop_filter(h, lf_x_start, s->mb_x);
+ s->mb_x = lf_x_start = 0;
ff_draw_horiz_band(s, 16*s->mb_y, 16);
++s->mb_y;
if(FIELD_OR_MBAFF_PICTURE) {
if( eos || s->mb_y >= s->mb_height ) {
tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
+ if (s->mb_x > lf_x_start) loop_filter(h, lf_x_start, s->mb_x);
return 0;
}
}
if(ret<0){
av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d\n", s->mb_x, s->mb_y);
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x, s->mb_y, (AC_ERROR|DC_ERROR|MV_ERROR)&part_mask);
-
return -1;
}
if(++s->mb_x >= s->mb_width){
- s->mb_x=0;
- loop_filter(h);
+ loop_filter(h, lf_x_start, s->mb_x);
+ s->mb_x = lf_x_start = 0;
ff_draw_horiz_band(s, 16*s->mb_y, 16);
++s->mb_y;
if(FIELD_OR_MBAFF_PICTURE) {
tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
if(get_bits_count(&s->gb) == s->gb.size_in_bits ){
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
+ if (s->mb_x > lf_x_start) loop_filter(h, lf_x_start, s->mb_x);
return 0;
}else{
int next_avc= h->is_avc ? 0 : buf_size;
h->max_contexts = avctx->thread_count;
-#if 0
- int i;
- for(i=0; i<50; i++){
- av_log(NULL, AV_LOG_ERROR,"%02X ", buf[i]);
- }
-#endif
if(!(s->flags2 & CODEC_FLAG2_CHUNKS)){
h->current_slice = 0;
if (!s->first_field)
buf_index += consumed;
- if( (s->hurry_up == 1 && h->nal_ref_idc == 0) //FIXME do not discard SEI id
- ||(avctx->skip_frame >= AVDISCARD_NONREF && h->nal_ref_idc == 0))
+ //FIXME do not discard SEI id
+ if(avctx->skip_frame >= AVDISCARD_NONREF && h->nal_ref_idc == 0)
continue;
again:
s->current_picture_ptr->key_frame |=
(hx->nal_unit_type == NAL_IDR_SLICE) ||
(h->sei_recovery_frame_cnt >= 0);
- if(hx->redundant_pic_count==0 && hx->s.hurry_up < 5
+ if(hx->redundant_pic_count==0
&& (avctx->skip_frame < AVDISCARD_NONREF || hx->nal_ref_idc)
- && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type_nos!=FF_B_TYPE)
- && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type_nos==FF_I_TYPE)
+ && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type_nos!=AV_PICTURE_TYPE_B)
+ && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type_nos==AV_PICTURE_TYPE_I)
&& avctx->skip_frame < AVDISCARD_ALL){
if(avctx->hwaccel) {
if (avctx->hwaccel->decode_slice(avctx, &buf[buf_index - consumed], consumed) < 0)
if(hx->redundant_pic_count==0 && hx->intra_gb_ptr && hx->s.data_partitioning
&& s->context_initialized
- && s->hurry_up < 5
&& (avctx->skip_frame < AVDISCARD_NONREF || hx->nal_ref_idc)
- && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type_nos!=FF_B_TYPE)
- && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type_nos==FF_I_TYPE)
+ && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type_nos!=AV_PICTURE_TYPE_B)
+ && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type_nos==AV_PICTURE_TYPE_I)
&& avctx->skip_frame < AVDISCARD_ALL)
context_count++;
break;
if(avctx->has_b_frames < 2)
avctx->has_b_frames= !s->low_delay;
+
+ if (avctx->bits_per_raw_sample != h->sps.bit_depth_luma) {
+ if (h->sps.bit_depth_luma >= 8 && h->sps.bit_depth_luma <= 10) {
+ avctx->bits_per_raw_sample = h->sps.bit_depth_luma;
+ h->pixel_shift = h->sps.bit_depth_luma > 8;
+
+ ff_h264dsp_init(&h->h264dsp, h->sps.bit_depth_luma);
+ ff_h264_pred_init(&h->hpc, s->codec_id, h->sps.bit_depth_luma);
+ dsputil_init(&s->dsp, s->avctx);
+ } else {
+ av_log(avctx, AV_LOG_DEBUG, "Unsupported bit depth: %d\n", h->sps.bit_depth_luma);
+ return -1;
+ }
+ }
break;
case NAL_PPS:
init_get_bits(&s->gb, ptr, bit_length);
}
if(!(s->flags2 & CODEC_FLAG2_CHUNKS) && !s->current_picture_ptr){
- if (avctx->skip_frame >= AVDISCARD_NONREF || s->hurry_up) return 0;
+ if (avctx->skip_frame >= AVDISCARD_NONREF)
+ return 0;
av_log(avctx, AV_LOG_ERROR, "no frame!\n");
return -1;
}
else if((out_of_order && pics-1 == s->avctx->has_b_frames && s->avctx->has_b_frames < MAX_DELAYED_PIC_COUNT)
|| (s->low_delay &&
((h->outputed_poc != INT_MIN && out->poc > h->outputed_poc + 2)
- || cur->pict_type == FF_B_TYPE)))
+ || cur->pict_type == AV_PICTURE_TYPE_B)))
{
s->low_delay = 0;
s->avctx->has_b_frames++;
NULL,
ff_h264_decode_end,
decode_frame,
- /*CODEC_CAP_DRAW_HORIZ_BAND |*/ CODEC_CAP_DR1 | CODEC_CAP_DELAY,
+ /*CODEC_CAP_DRAW_HORIZ_BAND |*/ CODEC_CAP_DR1 | CODEC_CAP_DELAY |
+ CODEC_CAP_SLICE_THREADS,
.flush= flush_dpb,
.long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
.profiles = NULL_IF_CONFIG_SMALL(profiles),
projects / ffmpeg / blobdiff