多媒体编程

开发平台：

Visual C++

h264.c：源码内容

decode_intra_mb:
partition_count = 0;
cbp= i_mb_type_info[mb_type].cbp;
h->intra16x16_pred_mode= i_mb_type_info[mb_type].pred_mode;
mb_type= i_mb_type_info[mb_type].type;
}
if(MB_FIELD)
mb_type |= MB_TYPE_INTERLACED;
h->slice_table[ mb_xy ]= h->slice_num;
if(IS_INTRA_PCM(mb_type)) {
const uint8_t *ptr;
// We assume these blocks are very rare so we do not optimize it.
// FIXME The two following lines get the bitstream position in the cabac
// decode, I think it should be done by a function in cabac.h (or cabac.c).
ptr= h->cabac.bytestream;
if(h->cabac.low&0x1) ptr--;
if(CABAC_BITS==16){
if(h->cabac.low&0x1FF) ptr--;
}
// The pixels are stored in the same order as levels in h->mb array.
memcpy(h->mb, ptr, 256); ptr+=256;
if(CHROMA){
memcpy(h->mb+128, ptr, 128); ptr+=128;
}
ff_init_cabac_decoder(&h->cabac, ptr, h->cabac.bytestream_end - ptr);
// All blocks are present
h->cbp_table[mb_xy] = 0x1ef;
h->chroma_pred_mode_table[mb_xy] = 0;
// In deblocking, the quantizer is 0
s->current_picture.qscale_table[mb_xy]= 0;
// All coeffs are present
memset(h->non_zero_count[mb_xy], 16, 16);
s->current_picture.mb_type[mb_xy]= mb_type;
h->last_qscale_diff = 0;
return 0;
}
if(MB_MBAFF){
h->ref_count[0] <<= 1;
h->ref_count[1] <<= 1;
}
fill_caches(h, mb_type, 0);
if( IS_INTRA( mb_type ) ) {
int i, pred_mode;
if( IS_INTRA4x4( mb_type ) ) {
if( dct8x8_allowed && decode_cabac_mb_transform_size( h ) ) {
mb_type |= MB_TYPE_8x8DCT;
for( i = 0; i < 16; i+=4 ) {
int pred = pred_intra_mode( h, i );
int mode = decode_cabac_mb_intra4x4_pred_mode( h, pred );
fill_rectangle( &h->intra4x4_pred_mode_cache[ scan8[i] ], 2, 2, 8, mode, 1 );
}
} else {
for( i = 0; i < 16; i++ ) {
int pred = pred_intra_mode( h, i );
h->intra4x4_pred_mode_cache[ scan8[i] ] = decode_cabac_mb_intra4x4_pred_mode( h, pred );
//av_log( s->avctx, AV_LOG_ERROR, "i4x4 pred=%d mode=%dn", pred, h->intra4x4_pred_mode_cache[ scan8[i] ] );
}
}
write_back_intra_pred_mode(h);
if( check_intra4x4_pred_mode(h) < 0 ) return -1;
} else {
h->intra16x16_pred_mode= check_intra_pred_mode( h, h->intra16x16_pred_mode );
if( h->intra16x16_pred_mode < 0 ) return -1;
}
if(CHROMA){
h->chroma_pred_mode_table[mb_xy] =
pred_mode = decode_cabac_mb_chroma_pre_mode( h );
pred_mode= check_intra_pred_mode( h, pred_mode );
if( pred_mode < 0 ) return -1;
h->chroma_pred_mode= pred_mode;
}
} else if( partition_count == 4 ) {
int i, j, sub_partition_count[4], list, ref[2][4];
if( h->slice_type_nos == FF_B_TYPE ) {
for( i = 0; i < 4; i++ ) {
h->sub_mb_type[i] = decode_cabac_b_mb_sub_type( h );
sub_partition_count[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count;
h->sub_mb_type[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].type;
}
if( IS_DIRECT(h->sub_mb_type[0] | h->sub_mb_type[1] |
h->sub_mb_type[2] | h->sub_mb_type[3]) ) {
pred_direct_motion(h, &mb_type);
h->ref_cache[0][scan8[4]] =
h->ref_cache[1][scan8[4]] =
h->ref_cache[0][scan8[12]] =
h->ref_cache[1][scan8[12]] = PART_NOT_AVAILABLE;
if( h->ref_count[0] > 1 || h->ref_count[1] > 1 ) {
for( i = 0; i < 4; i++ )
if( IS_DIRECT(h->sub_mb_type[i]) )
fill_rectangle( &h->direct_cache[scan8[4*i]], 2, 2, 8, 1, 1 );
}
}
} else {
for( i = 0; i < 4; i++ ) {
h->sub_mb_type[i] = decode_cabac_p_mb_sub_type( h );
sub_partition_count[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count;
h->sub_mb_type[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].type;
}
}
for( list = 0; list < h->list_count; list++ ) {
for( i = 0; i < 4; i++ ) {
if(IS_DIRECT(h->sub_mb_type[i])) continue;
if(IS_DIR(h->sub_mb_type[i], 0, list)){
if( h->ref_count[list] > 1 ){
ref[list][i] = decode_cabac_mb_ref( h, list, 4*i );
if(ref[list][i] >= h->ref_count[list]){
av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %dn", ref[list][i], h->ref_count[list]);
return -1;
}
}else
ref[list][i] = 0;
} else {
ref[list][i] = -1;
}
h->ref_cache[list][ scan8[4*i]+1 ]=
h->ref_cache[list][ scan8[4*i]+8 ]=h->ref_cache[list][ scan8[4*i]+9 ]= ref[list][i];
}
}
if(dct8x8_allowed)
dct8x8_allowed = get_dct8x8_allowed(h);
for(list=0; list<h->list_count; list++){
for(i=0; i<4; i++){
h->ref_cache[list][ scan8[4*i] ]=h->ref_cache[list][ scan8[4*i]+1 ];
if(IS_DIRECT(h->sub_mb_type[i])){
fill_rectangle(h->mvd_cache[list][scan8[4*i]], 2, 2, 8, 0, 4);
continue;
}
if(IS_DIR(h->sub_mb_type[i], 0, list) && !IS_DIRECT(h->sub_mb_type[i])){
const int sub_mb_type= h->sub_mb_type[i];
const int block_width= (sub_mb_type & (MB_TYPE_16x16|MB_TYPE_16x8)) ? 2 : 1;
for(j=0; j<sub_partition_count[i]; j++){
int mpx, mpy;
int mx, my;
const int index= 4*i + block_width*j;
int16_t (* mv_cache)[2]= &h->mv_cache[list][ scan8[index] ];
int16_t (* mvd_cache)[2]= &h->mvd_cache[list][ scan8[index] ];
pred_motion(h, index, block_width, list, h->ref_cache[list][ scan8[index] ], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, index, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, index, 1 );
tprintf(s->avctx, "final mv:%d %dn", mx, my);
if(IS_SUB_8X8(sub_mb_type)){
mv_cache[ 1 ][0]=
mv_cache[ 8 ][0]= mv_cache[ 9 ][0]= mx;
mv_cache[ 1 ][1]=
mv_cache[ 8 ][1]= mv_cache[ 9 ][1]= my;
mvd_cache[ 1 ][0]=
mvd_cache[ 8 ][0]= mvd_cache[ 9 ][0]= mx - mpx;
mvd_cache[ 1 ][1]=
mvd_cache[ 8 ][1]= mvd_cache[ 9 ][1]= my - mpy;
}else if(IS_SUB_8X4(sub_mb_type)){
mv_cache[ 1 ][0]= mx;
mv_cache[ 1 ][1]= my;
mvd_cache[ 1 ][0]= mx - mpx;
mvd_cache[ 1 ][1]= my - mpy;
}else if(IS_SUB_4X8(sub_mb_type)){
mv_cache[ 8 ][0]= mx;
mv_cache[ 8 ][1]= my;
mvd_cache[ 8 ][0]= mx - mpx;
mvd_cache[ 8 ][1]= my - mpy;
}
mv_cache[ 0 ][0]= mx;
mv_cache[ 0 ][1]= my;
mvd_cache[ 0 ][0]= mx - mpx;
mvd_cache[ 0 ][1]= my - mpy;
}
}else{
uint32_t *p= (uint32_t *)&h->mv_cache[list][ scan8[4*i] ][0];
uint32_t *pd= (uint32_t *)&h->mvd_cache[list][ scan8[4*i] ][0];
p[0] = p[1] = p[8] = p[9] = 0;
pd[0]= pd[1]= pd[8]= pd[9]= 0;
}
}
}
} else if( IS_DIRECT(mb_type) ) {
pred_direct_motion(h, &mb_type);
fill_rectangle(h->mvd_cache[0][scan8[0]], 4, 4, 8, 0, 4);
fill_rectangle(h->mvd_cache[1][scan8[0]], 4, 4, 8, 0, 4);
dct8x8_allowed &= h->sps.direct_8x8_inference_flag;
} else {
int list, mx, my, i, mpx, mpy;
if(IS_16X16(mb_type)){
for(list=0; list<h->list_count; list++){
if(IS_DIR(mb_type, 0, list)){
int ref;
if(h->ref_count[list] > 1){
ref= decode_cabac_mb_ref(h, list, 0);
if(ref >= h->ref_count[list])
{
av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %dn", ref, h->ref_count[list]);
return -1;
}
}else
ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, (uint8_t)LIST_NOT_USED, 1); //FIXME factorize and the other fill_rect below too
}
for(list=0; list<h->list_count; list++){
if(IS_DIR(mb_type, 0, list)){
pred_motion(h, 0, 4, list, h->ref_cache[list][ scan8[0] ], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 0, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 0, 1 );
tprintf(s->avctx, "final mv:%d %dn", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx,my), 4);
}else
fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, 0, 4);
}
}
else if(IS_16X8(mb_type)){
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
int ref;
if(h->ref_count[list] > 1){
ref= decode_cabac_mb_ref( h, list, 8*i );
if(ref >= h->ref_count[list])
{
av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %dn", ref, h->ref_count[list]);
return -1;
}
}else
ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, (LIST_NOT_USED&0xFF), 1);
}
}
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
pred_16x8_motion(h, 8*i, list, h->ref_cache[list][scan8[0] + 16*i], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 8*i, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 8*i, 1 );
tprintf(s->avctx, "final mv:%d %dn", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx,my), 4);
}else{
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, 0, 4);
fill_rectangle(h-> mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, 0, 4);
}
}
}
}else{
assert(IS_8X16(mb_type));
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){ //FIXME optimize
int ref;
if(h->ref_count[list] > 1){
ref= decode_cabac_mb_ref( h, list, 4*i );
if(ref >= h->ref_count[list])
{
av_log(s->avctx, AV_LOG_ERROR, "Reference %d >= %dn", ref, h->ref_count[list]);
return -1;
}
}else
ref=0;
fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, ref, 1);
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, (LIST_NOT_USED&0xFF), 1);
}
}
for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
pred_8x16_motion(h, i*4, list, h->ref_cache[list][ scan8[0] + 2*i ], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 4*i, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 4*i, 1 );
tprintf(s->avctx, "final mv:%d %dn", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx,my), 4);
}else{
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, 0, 4);
fill_rectangle(h-> mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, 0, 4);
}
}
}
}
}
if( IS_INTER( mb_type ) ) {
h->chroma_pred_mode_table[mb_xy] = 0;
write_back_motion( h, mb_type );
}
if( !IS_INTRA16x16( mb_type ) ) {
cbp = decode_cabac_mb_cbp_luma( h );
if(CHROMA)
cbp |= decode_cabac_mb_cbp_chroma( h ) << 4;
}
h->cbp_table[mb_xy] = h->cbp = cbp;
if( dct8x8_allowed && (cbp&15) && !IS_INTRA( mb_type ) ) {
if( decode_cabac_mb_transform_size( h ) )
mb_type |= MB_TYPE_8x8DCT;
}
s->current_picture.mb_type[mb_xy]= mb_type;
if( cbp || IS_INTRA16x16( mb_type ) ) {
const uint8_t *scan, *scan8x8, *dc_scan;
const uint32_t *qmul;
int dqp;
if(IS_INTERLACED(mb_type)){
scan8x8= s->qscale ? h->field_scan8x8 : h->field_scan8x8_q0;
scan= s->qscale ? h->field_scan : h->field_scan_q0;
dc_scan= luma_dc_field_scan;
}else{
scan8x8= s->qscale ? h->zigzag_scan8x8 : h->zigzag_scan8x8_q0;
scan= s->qscale ? h->zigzag_scan : h->zigzag_scan_q0;
dc_scan= luma_dc_zigzag_scan;
}
h->last_qscale_diff = dqp = decode_cabac_mb_dqp( h );
if( dqp == INT_MIN ){
av_log(h->s.avctx, AV_LOG_ERROR, "cabac decode of qscale diff failed at %d %dn", s->mb_x, s->mb_y);
return -1;
}
s->qscale += dqp;
if(((unsigned)s->qscale) > 51){
if(s->qscale<0) s->qscale+= 52;
else s->qscale-= 52;
}
h->chroma_qp[0] = get_chroma_qp(h, 0, s->qscale);
h->chroma_qp[1] = get_chroma_qp(h, 1, s->qscale);
if( IS_INTRA16x16( mb_type ) ) {
int i;
//av_log( s->avctx, AV_LOG_ERROR, "INTRA16x16 DCn" );
decode_cabac_residual( h, h->mb, 0, 0, dc_scan, NULL, 16);
if( cbp&15 ) {
qmul = h->dequant4_coeff[0][s->qscale];
for( i = 0; i < 16; i++ ) {
//av_log( s->avctx, AV_LOG_ERROR, "INTRA16x16 AC:%dn", i );
decode_cabac_residual(h, h->mb + 16*i, 1, i, scan + 1, qmul, 15);
}
} else {
fill_rectangle(&h->non_zero_count_cache[scan8[0]], 4, 4, 8, 0, 1);
}
} else {
int i8x8, i4x4;
for( i8x8 = 0; i8x8 < 4; i8x8++ ) {
if( cbp & (1<<i8x8) ) {
if( IS_8x8DCT(mb_type) ) {
decode_cabac_residual(h, h->mb + 64*i8x8, 5, 4*i8x8,
scan8x8, h->dequant8_coeff[IS_INTRA( mb_type ) ? 0:1][s->qscale], 64);
} else {
qmul = h->dequant4_coeff[IS_INTRA( mb_type ) ? 0:3][s->qscale];
for( i4x4 = 0; i4x4 < 4; i4x4++ ) {
const int index = 4*i8x8 + i4x4;
//av_log( s->avctx, AV_LOG_ERROR, "Luma4x4: %dn", index );
//START_TIMER
decode_cabac_residual(h, h->mb + 16*index, 2, index, scan, qmul, 16);
//STOP_TIMER("decode_residual")
}
}
} else {
uint8_t * const nnz= &h->non_zero_count_cache[ scan8[4*i8x8] ];
nnz[0] = nnz[1] = nnz[8] = nnz[9] = 0;
}
}
}
if( cbp&0x30 ){
int c;
for( c = 0; c < 2; c++ ) {
//av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-DCn",c );
decode_cabac_residual(h, h->mb + 256 + 16*4*c, 3, c, chroma_dc_scan, NULL, 4);
}
}
if( cbp&0x20 ) {
int c, i;
for( c = 0; c < 2; c++ ) {
qmul = h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[c]];
for( i = 0; i < 4; i++ ) {
const int index = 16 + 4 * c + i;
//av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-AC %dn",c, index - 16 );
decode_cabac_residual(h, h->mb + 16*index, 4, index, scan + 1, qmul, 15);
}
}
} else {
uint8_t * const nnz= &h->non_zero_count_cache[0];
nnz[ scan8[16]+0 ] = nnz[ scan8[16]+1 ] =nnz[ scan8[16]+8 ] =nnz[ scan8[16]+9 ] =
nnz[ scan8[20]+0 ] = nnz[ scan8[20]+1 ] =nnz[ scan8[20]+8 ] =nnz[ scan8[20]+9 ] = 0;
}
} else {
uint8_t * const nnz= &h->non_zero_count_cache[0];
fill_rectangle(&nnz[scan8[0]], 4, 4, 8, 0, 1);
nnz[ scan8[16]+0 ] = nnz[ scan8[16]+1 ] =nnz[ scan8[16]+8 ] =nnz[ scan8[16]+9 ] =
nnz[ scan8[20]+0 ] = nnz[ scan8[20]+1 ] =nnz[ scan8[20]+8 ] =nnz[ scan8[20]+9 ] = 0;
h->last_qscale_diff = 0;
}
s->current_picture.qscale_table[mb_xy]= s->qscale;
write_back_non_zero_count(h);
if(MB_MBAFF){
h->ref_count[0] >>= 1;
h->ref_count[1] >>= 1;
}
return 0;
}
static void filter_mb_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
tc[0] = (tc0_table+52)[index_a][bS[0]];
tc[1] = (tc0_table+52)[index_a][bS[1]];
tc[2] = (tc0_table+52)[index_a][bS[2]];
tc[3] = (tc0_table+52)[index_a][bS[3]];
h->s.dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
tc[0] = (tc0_table+52)[index_a][bS[0]]+1;
tc[1] = (tc0_table+52)[index_a][bS[1]]+1;
tc[2] = (tc0_table+52)[index_a][bS[2]]+1;
tc[3] = (tc0_table+52)[index_a][bS[3]]+1;
h->s.dsp.h264_h_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_h_loop_filter_chroma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[8], int qp[2] ) {
int i;
for( i = 0; i < 16; i++, pix += stride) {
int index_a;
int alpha;
int beta;
int qp_index;
int bS_index = (i >> 1);
if (!MB_FIELD) {
bS_index &= ~1;
bS_index |= (i & 1);
}
if( bS[bS_index] == 0 ) {
continue;
}
qp_index = MB_FIELD ? (i >> 3) : (i & 1);
index_a = qp[qp_index] + h->slice_alpha_c0_offset;
alpha = (alpha_table+52)[index_a];
beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
const int tc0 = (tc0_table+52)[index_a][bS[bS_index]];
const int p0 = pix[-1];
const int p1 = pix[-2];
const int p2 = pix[-3];
const int q0 = pix[0];
const int q1 = pix[1];
const int q2 = pix[2];
if( FFABS( p0 - q0 ) < alpha &&
FFABS( p1 - p0 ) < beta &&
FFABS( q1 - q0 ) < beta ) {
int tc = tc0;
int i_delta;
if( FFABS( p2 - p0 ) < beta ) {
pix[-2] = p1 + av_clip( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
tc++;
}
if( FFABS( q2 - q0 ) < beta ) {
pix[1] = q1 + av_clip( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
tc++;
}
i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%dn# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
}
}else{
const int p0 = pix[-1];
const int p1 = pix[-2];
const int p2 = pix[-3];
const int q0 = pix[0];
const int q1 = pix[1];
const int q2 = pix[2];
if( FFABS( p0 - q0 ) < alpha &&
FFABS( p1 - p0 ) < beta &&
FFABS( q1 - q0 ) < beta ) {
if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
if( FFABS( p2 - p0 ) < beta)
{
const int p3 = pix[-4];
/* p0', p1', p2' */
pix[-1] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
pix[-2] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
pix[-3] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
} else {
/* p0' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
}
if( FFABS( q2 - q0 ) < beta)
{
const int q3 = pix[3];
/* q0', q1', q2' */
pix[0] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
pix[1] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
pix[2] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
} else {
/* q0' */
pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
}
}else{
/* p0', q0' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
pix[ 0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
}
tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%dn# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]n", i, qp[qp_index], index_a, alpha, beta, p2, p1, p0, q0, q1, q2, pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
}
}
}
}
static void filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[8], int qp[2] ) {
int i;
for( i = 0; i < 8; i++, pix += stride) {
int index_a;
int alpha;
int beta;
int qp_index;
int bS_index = i;
if( bS[bS_index] == 0 ) {
continue;
}
qp_index = MB_FIELD ? (i >> 2) : (i & 1);
index_a = qp[qp_index] + h->slice_alpha_c0_offset;
alpha = (alpha_table+52)[index_a];
beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
const int tc = (tc0_table+52)[index_a][bS[bS_index]] + 1;
const int p0 = pix[-1];
const int p1 = pix[-2];
const int q0 = pix[0];
const int q1 = pix[1];
if( FFABS( p0 - q0 ) < alpha &&
FFABS( p1 - p0 ) < beta &&
FFABS( q1 - q0 ) < beta ) {
const int i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%dn# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
}
}else{
const int p0 = pix[-1];
const int p1 = pix[-2];
const int q0 = pix[0];
const int q1 = pix[1];
if( FFABS( p0 - q0 ) < alpha &&
FFABS( p1 - p0 ) < beta &&
FFABS( q1 - q0 ) < beta ) {
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2; /* p0' */
pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2; /* q0' */
tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%dn# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]n", i, pix[-3], p1, p0, q0, q1, pix[2], pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
}
}
}
}
static void filter_mb_edgeh( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
tc[0] = (tc0_table+52)[index_a][bS[0]];
tc[1] = (tc0_table+52)[index_a][bS[1]];
tc[2] = (tc0_table+52)[index_a][bS[2]];
tc[3] = (tc0_table+52)[index_a][bS[3]];
h->s.dsp.h264_v_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_v_loop_filter_luma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_edgech( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
const int index_a = qp + h->slice_alpha_c0_offset;
const int alpha = (alpha_table+52)[index_a];
const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
tc[0] = (tc0_table+52)[index_a][bS[0]]+1;
tc[1] = (tc0_table+52)[index_a][bS[1]]+1;
tc[2] = (tc0_table+52)[index_a][bS[2]]+1;
tc[3] = (tc0_table+52)[index_a][bS[3]]+1;
h->s.dsp.h264_v_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_v_loop_filter_chroma_intra(pix, stride, alpha, beta);
}
}
static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
MpegEncContext * const s = &h->s;
int mb_y_firstrow = s->picture_structure == PICT_BOTTOM_FIELD;
int mb_xy, mb_type;
int qp, qp0, qp1, qpc, qpc0, qpc1, qp_thresh;
mb_xy = h->mb_xy;
if(mb_x==0 || mb_y==mb_y_firstrow || !s->dsp.h264_loop_filter_strength || h->pps.chroma_qp_diff ||
!(s->flags2 & CODEC_FLAG2_FAST) || //FIXME filter_mb_fast is broken, thus hasto be, but should not under CODEC_FLAG2_FAST
(h->deblocking_filter == 2 && (h->slice_table[mb_xy] != h->slice_table[h->top_mb_xy] ||
h->slice_table[mb_xy] != h->slice_table[mb_xy - 1]))) {
filter_mb(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize);
return;
}
assert(!FRAME_MBAFF);
mb_type = s->current_picture.mb_type[mb_xy];
qp = s->current_picture.qscale_table[mb_xy];
qp0 = s->current_picture.qscale_table[mb_xy-1];
qp1 = s->current_picture.qscale_table[h->top_mb_xy];
qpc = get_chroma_qp( h, 0, qp );
qpc0 = get_chroma_qp( h, 0, qp0 );
qpc1 = get_chroma_qp( h, 0, qp1 );
qp0 = (qp + qp0 + 1) >> 1;
qp1 = (qp + qp1 + 1) >> 1;
qpc0 = (qpc + qpc0 + 1) >> 1;
qpc1 = (qpc + qpc1 + 1) >> 1;
qp_thresh = 15 - h->slice_alpha_c0_offset;
if(qp <= qp_thresh && qp0 <= qp_thresh && qp1 <= qp_thresh &&
qpc <= qp_thresh && qpc0 <= qp_thresh && qpc1 <= qp_thresh)
return;
if( IS_INTRA(mb_type) ) {
int16_t bS4[4] = {4,4,4,4};
int16_t bS3[4] = {3,3,3,3};
int16_t *bSH = FIELD_PICTURE ? bS3 : bS4;
if( IS_8x8DCT(mb_type) ) {
filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 );
filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp );
filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bSH, qp1 );
filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp );
} else {
filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 );
filter_mb_edgev( h, &img_y[4*1], linesize, bS3, qp );
filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp );
filter_mb_edgev( h, &img_y[4*3], linesize, bS3, qp );
filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bSH, qp1 );
filter_mb_edgeh( h, &img_y[4*1*linesize], linesize, bS3, qp );
filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp );
filter_mb_edgeh( h, &img_y[4*3*linesize], linesize, bS3, qp );
}
filter_mb_edgecv( h, &img_cb[2*0], uvlinesize, bS4, qpc0 );
filter_mb_edgecv( h, &img_cb[2*2], uvlinesize, bS3, qpc );
filter_mb_edgecv( h, &img_cr[2*0], uvlinesize, bS4, qpc0 );
filter_mb_edgecv( h, &img_cr[2*2], uvlinesize, bS3, qpc );
filter_mb_edgech( h, &img_cb[2*0*uvlinesize], uvlinesize, bSH, qpc1 );
filter_mb_edgech( h, &img_cb[2*2*uvlinesize], uvlinesize, bS3, qpc );
filter_mb_edgech( h, &img_cr[2*0*uvlinesize], uvlinesize, bSH, qpc1 );
filter_mb_edgech( h, &img_cr[2*2*uvlinesize], uvlinesize, bS3, qpc );
return;
} else {
DECLARE_ALIGNED_8(int16_t, bS[2][4][4]);
uint64_t (*bSv)[4] = (uint64_t(*)[4])bS;
int edges;
if( IS_8x8DCT(mb_type) && (h->cbp&7) == 7 ) {
edges = 4;
bSv[0][0] = bSv[0][2] = bSv[1][0] = bSv[1][2] = 0x0002000200020002ULL;
} else {
int mask_edge1 = (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16)) ? 3 :
(mb_type & MB_TYPE_16x8) ? 1 : 0;
int mask_edge0 = (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16))
&& (s->current_picture.mb_type[mb_xy-1] & (MB_TYPE_16x16 | MB_TYPE_8x16))
? 3 : 0;
int step = IS_8x8DCT(mb_type) ? 2 : 1;
edges = (mb_type & MB_TYPE_16x16) && !(h->cbp & 15) ? 1 : 4;
s->dsp.h264_loop_filter_strength( bS, h->non_zero_count_cache, h->ref_cache, h->mv_cache,
(h->slice_type_nos == FF_B_TYPE), edges, step, mask_edge0, mask_edge1, FIELD_PICTURE);
}
if( IS_INTRA(s->current_picture.mb_type[mb_xy-1]) )
bSv[0][0] = 0x0004000400040004ULL;
if( IS_INTRA(s->current_picture.mb_type[h->top_mb_xy]) )
bSv[1][0] = FIELD_PICTURE ? 0x0003000300030003ULL : 0x0004000400040004ULL;
#define FILTER(hv,dir,edge)
if(bSv[dir][edge]) {
filter_mb_edge##hv( h, &img_y[4*edge*(dir?linesize:1)], linesize, bS[dir][edge], edge ? qp : qp##dir );
if(!(edge&1)) {
filter_mb_edgec##hv( h, &img_cb[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir );
filter_mb_edgec##hv( h, &img_cr[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir );
}
}
if( edges == 1 ) {
FILTER(v,0,0);
FILTER(h,1,0);
} else if( IS_8x8DCT(mb_type) ) {
FILTER(v,0,0);
FILTER(v,0,2);
FILTER(h,1,0);
FILTER(h,1,2);
} else {
FILTER(v,0,0);
FILTER(v,0,1);
FILTER(v,0,2);
FILTER(v,0,3);
FILTER(h,1,0);
FILTER(h,1,1);
FILTER(h,1,2);
FILTER(h,1,3);
}
#undef FILTER
}
}
static void av_always_inline filter_mb_dir(H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int dir) {
MpegEncContext * const s = &h->s;
int edge;
const int mbm_xy = dir == 0 ? mb_xy -1 : h->top_mb_xy;
const int mbm_type = s->current_picture.mb_type[mbm_xy];
//int (*ref2frm) [64] = h->ref2frm[ h->slice_num &(MAX_SLICES-1) ][0] + (MB_MBAFF ? 20 : 2);
//int (*ref2frmm)[64] = h->ref2frm[ h->slice_table[mbm_xy]&(MAX_SLICES-1) ][0] + (MB_MBAFF ? 20 : 2);
int* ref2frm[2];
int* ref2frmm[2];
int start = h->slice_table[mbm_xy] == 0xFFFF ? 1 : 0;
const int edges = (mb_type & (MB_TYPE_16x16|MB_TYPE_SKIP))
== (MB_TYPE_16x16|MB_TYPE_SKIP) ? 1 : 4;
// how often to recheck mv-based bS when iterating between edges
const int mask_edge = (mb_type & (MB_TYPE_16x16 | (MB_TYPE_16x8 << dir))) ? 3 :
(mb_type & (MB_TYPE_8x16 >> dir)) ? 1 : 0;
// how often to recheck mv-based bS when iterating along each edge
const int mask_par0 = mb_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir));
ref2frm[0] = h->ref2frm[ h->slice_num &(MAX_SLICES-1) ][0] + (MB_MBAFF ? 20 : 2);
ref2frm[1] = h->ref2frm[ h->slice_num &(MAX_SLICES-1) ][1] + (MB_MBAFF ? 20 : 2);
ref2frmm[0] = h->ref2frm[ h->slice_table[mbm_xy]&(MAX_SLICES-1) ][0] + (MB_MBAFF ? 20 : 2);
ref2frmm[1] = h->ref2frm[ h->slice_table[mbm_xy]&(MAX_SLICES-1) ][1] + (MB_MBAFF ? 20 : 2);
if (first_vertical_edge_done) {
start = 1;
}
if (h->deblocking_filter==2 && h->slice_table[mbm_xy] != h->slice_table[mb_xy])
start = 1;
if (FRAME_MBAFF && (dir == 1) && ((mb_y&1) == 0) && start == 0
&& !IS_INTERLACED(mb_type)
&& IS_INTERLACED(mbm_type)
) {
// This is a special case in the norm where the filtering must
// be done twice (one each of the field) even if we are in a
// frame macroblock.
//
static const int nnz_idx[4] = {4,5,6,3};
unsigned int tmp_linesize = 2 * linesize;
unsigned int tmp_uvlinesize = 2 * uvlinesize;
int mbn_xy = mb_xy - 2 * s->mb_stride;
int qp;
int i, j;
int16_t bS[4];
for(j=0; j<2; j++, mbn_xy += s->mb_stride){
if( IS_INTRA(mb_type) ||
IS_INTRA(s->current_picture.mb_type[mbn_xy]) ) {
bS[0] = bS[1] = bS[2] = bS[3] = 3;
} else {
const uint8_t *mbn_nnz = h->non_zero_count[mbn_xy];
for( i = 0; i < 4; i++ ) {
if( h->non_zero_count_cache[scan8[0]+i] != 0 ||
mbn_nnz[nnz_idx[i]] != 0 )
bS[i] = 2;
else
bS[i] = 1;
}
}
// Do not use s->qscale as luma quantizer because it has not the same
// value in IPCM macroblocks.
qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbn_xy] + 1 ) >> 1;
tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, tmp_linesize, tmp_uvlinesize);
{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "n"); }
filter_mb_edgeh( h, &img_y[j*linesize], tmp_linesize, bS, qp );
filter_mb_edgech( h, &img_cb[j*uvlinesize], tmp_uvlinesize, bS,
( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
filter_mb_edgech( h, &img_cr[j*uvlinesize], tmp_uvlinesize, bS,
( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
}
start = 1;
}
/* Calculate bS */
for( edge = start; edge < edges; edge++ ) {
/* mbn_xy: neighbor macroblock */
const int mbn_xy = edge > 0 ? mb_xy : mbm_xy;
const int mbn_type = s->current_picture.mb_type[mbn_xy];
//int (*ref2frmn)[64] = edge > 0 ? ref2frm : ref2frmm;
int* ref2frmn[2];
int16_t bS[4];
int qp;
ref2frmn[0] = edge > 0 ? ref2frm[0] : ref2frmm[0];
ref2frmn[1] = edge > 0 ? ref2frm[1] : ref2frmm[1];
if( (edge&1) && IS_8x8DCT(mb_type) )
continue;
if( IS_INTRA(mb_type) ||
IS_INTRA(mbn_type) ) {
int value;
if (edge == 0) {
if ( (!IS_INTERLACED(mb_type) && !IS_INTERLACED(mbm_type))
|| ((FRAME_MBAFF || (s->picture_structure != PICT_FRAME)) && (dir == 0))
) {
value = 4;
} else {
value = 3;
}
} else {
value = 3;
}
bS[0] = bS[1] = bS[2] = bS[3] = value;
} else {
int i, l;
int mv_done;
if( edge & mask_edge ) {
bS[0] = bS[1] = bS[2] = bS[3] = 0;
mv_done = 1;
}
else if( FRAME_MBAFF && IS_INTERLACED(mb_type ^ mbn_type)) {
bS[0] = bS[1] = bS[2] = bS[3] = 1;
mv_done = 1;
}
else if( mask_par0 && (edge || (mbn_type & (MB_TYPE_16x16 | (MB_TYPE_8x16 >> dir)))) ) {
int b_idx= 8 + 4 + edge * (dir ? 8:1);
int bn_idx= b_idx - (dir ? 8:1);
int v = 0;
for( l = 0; !v && l < 1 + (h->slice_type_nos == FF_B_TYPE); l++ ) {
v |= ref2frm[l][h->ref_cache[l][b_idx]] != ref2frmn[l][h->ref_cache[l][bn_idx]] ||
FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit;
}
if(h->slice_type_nos == FF_B_TYPE && v){
v=0;
for( l = 0; !v && l < 2; l++ ) {
int ln= 1-l;
v |= ref2frm[l][h->ref_cache[l][b_idx]] != ref2frmn[ln][h->ref_cache[ln][bn_idx]] ||
FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[ln][bn_idx][0] ) >= 4 ||
FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[ln][bn_idx][1] ) >= mvy_limit;
}
}
bS[0] = bS[1] = bS[2] = bS[3] = v;
mv_done = 1;
}
else
mv_done = 0;
for( i = 0; i < 4; i++ ) {
int x = dir == 0 ? edge : i;
int y = dir == 0 ? i : edge;
int b_idx= 8 + 4 + x + 8*y;
int bn_idx= b_idx - (dir ? 8:1);
if( h->non_zero_count_cache[b_idx] |
h->non_zero_count_cache[bn_idx] ) {
bS[i] = 2;
}
else if(!mv_done)
{
bS[i] = 0;
for( l = 0; l < 1 + (h->slice_type_nos == FF_B_TYPE); l++ ) {
if( ref2frm[l][h->ref_cache[l][b_idx]] != ref2frmn[l][h->ref_cache[l][bn_idx]] ||
FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit ) {
bS[i] = 1;
break;
}
}
if(h->slice_type_nos == FF_B_TYPE && bS[i]){
bS[i] = 0;
for( l = 0; l < 2; l++ ) {
int ln= 1-l;
if( ref2frm[l][h->ref_cache[l][b_idx]] != ref2frmn[ln][h->ref_cache[ln][bn_idx]] ||
FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[ln][bn_idx][0] ) >= 4 ||
FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[ln][bn_idx][1] ) >= mvy_limit ) {
bS[i] = 1;
break;
}
}
}
}
}
if(bS[0]+bS[1]+bS[2]+bS[3] == 0)
continue;
}
/* Filter edge */
// Do not use s->qscale as luma quantizer because it has not the same
// value in IPCM macroblocks.
qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbn_xy] + 1 ) >> 1;
//tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d, QPc:%d, QPcn:%dn", mb_x, mb_y, dir, edge, qp, h->chroma_qp, s->current_picture.qscale_table[mbn_xy]);
tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, linesize, uvlinesize);
{ int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "n"); }
if( dir == 0 ) {
filter_mb_edgev( h, &img_y[4*edge], linesize, bS, qp );
if( (edge&1) == 0 ) {
filter_mb_edgecv( h, &img_cb[2*edge], uvlinesize, bS,
( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
filter_mb_edgecv( h, &img_cr[2*edge], uvlinesize, bS,
( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
}
} else {
filter_mb_edgeh( h, &img_y[4*edge*linesize], linesize, bS, qp );
if( (edge&1) == 0 ) {
filter_mb_edgech( h, &img_cb[2*edge*uvlinesize], uvlinesize, bS,
( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
filter_mb_edgech( h, &img_cr[2*edge*uvlinesize], uvlinesize, bS,
( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1);
}
}
}
}
static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
MpegEncContext * const s = &h->s;
const int mb_xy= mb_x + mb_y*s->mb_stride;
const int mb_type = s->current_picture.mb_type[mb_xy];
const int mvy_limit = IS_INTERLACED(mb_type) ? 2 : 4;
int first_vertical_edge_done = 0;
//for sufficiently low qp, filtering wouldn't do anything
//this is a conservative estimate: could also check beta_offset and more accurate chroma_qp
if(!FRAME_MBAFF){
int qp_thresh = 15 - h->slice_alpha_c0_offset - FFMAX3(0, h->pps.chroma_qp_index_offset[0], h->pps.chroma_qp_index_offset[1]);
int qp = s->current_picture.qscale_table[mb_xy];
if(qp <= qp_thresh
&& (mb_x == 0 || ((qp + s->current_picture.qscale_table[mb_xy-1] + 1)>>1) <= qp_thresh)
&& (mb_y == 0 || ((qp + s->current_picture.qscale_table[h->top_mb_xy] + 1)>>1) <= qp_thresh)){
return;
}
}
// CAVLC 8x8dct requires NNZ values for residual decoding that differ from what the loop filter needs
if(!h->pps.cabac && h->pps.transform_8x8_mode){
int top_type, left_type[2];
top_type = s->current_picture.mb_type[h->top_mb_xy] ;
left_type[0] = s->current_picture.mb_type[h->left_mb_xy[0]];
left_type[1] = s->current_picture.mb_type[h->left_mb_xy[1]];
if(IS_8x8DCT(top_type)){
h->non_zero_count_cache[4+8*0]=
h->non_zero_count_cache[5+8*0]= h->cbp_table[h->top_mb_xy] & 4;
h->non_zero_count_cache[6+8*0]=
h->non_zero_count_cache[7+8*0]= h->cbp_table[h->top_mb_xy] & 8;
}
if(IS_8x8DCT(left_type[0])){
h->non_zero_count_cache[3+8*1]=
h->non_zero_count_cache[3+8*2]= h->cbp_table[h->left_mb_xy[0]]&2; //FIXME check MBAFF
}
if(IS_8x8DCT(left_type[1])){
h->non_zero_count_cache[3+8*3]=
h->non_zero_count_cache[3+8*4]= h->cbp_table[h->left_mb_xy[1]]&8; //FIXME check MBAFF
}
if(IS_8x8DCT(mb_type)){
h->non_zero_count_cache[scan8[0 ]]= h->non_zero_count_cache[scan8[1 ]]=
h->non_zero_count_cache[scan8[2 ]]= h->non_zero_count_cache[scan8[3 ]]= h->cbp & 1;
h->non_zero_count_cache[scan8[0+ 4]]= h->non_zero_count_cache[scan8[1+ 4]]=
h->non_zero_count_cache[scan8[2+ 4]]= h->non_zero_count_cache[scan8[3+ 4]]= h->cbp & 2;
h->non_zero_count_cache[scan8[0+ 8]]= h->non_zero_count_cache[scan8[1+ 8]]=
h->non_zero_count_cache[scan8[2+ 8]]= h->non_zero_count_cache[scan8[3+ 8]]= h->cbp & 4;
h->non_zero_count_cache[scan8[0+12]]= h->non_zero_count_cache[scan8[1+12]]=
h->non_zero_count_cache[scan8[2+12]]= h->non_zero_count_cache[scan8[3+12]]= h->cbp & 8;
}
}
if (FRAME_MBAFF
// left mb is in picture
&& h->slice_table[mb_xy-1] != 0xFFFF
// and current and left pair do not have the same interlaced type
&& (IS_INTERLACED(mb_type) != IS_INTERLACED(s->current_picture.mb_type[mb_xy-1]))
// and left mb is in the same slice if deblocking_filter == 2
&& (h->deblocking_filter!=2 || h->slice_table[mb_xy-1] == h->slice_table[mb_xy])) {
/* First vertical edge is different in MBAFF frames
* There are 8 different bS to compute and 2 different Qp
*/
const int pair_xy = mb_x + (mb_y&~1)*s->mb_stride;
const int left_mb_xy[2] = { pair_xy-1, pair_xy-1+s->mb_stride };
int16_t bS[8];
int qp[2];
int bqp[2];
int rqp[2];
int mb_qp, mbn0_qp, mbn1_qp;
int i;
first_vertical_edge_done = 1;
if( IS_INTRA(mb_type) )
bS[0] = bS[1] = bS[2] = bS[3] = bS[4] = bS[5] = bS[6] = bS[7] = 4;
else {
for( i = 0; i < 8; i++ ) {
int mbn_xy = MB_FIELD ? left_mb_xy[i>>2] : left_mb_xy[i&1];
if( IS_INTRA( s->current_picture.mb_type[mbn_xy] ) )
bS[i] = 4;
else if( h->non_zero_count_cache[12+8*(i>>1)] != 0 ||
((!h->pps.cabac && IS_8x8DCT(s->current_picture.mb_type[mbn_xy])) ?
(h->cbp_table[mbn_xy] & ((MB_FIELD ? (i&2) : (mb_y&1)) ? 8 : 2))
:
h->non_zero_count[mbn_xy][MB_FIELD ? i&3 : (i>>2)+(mb_y&1)*2]))
bS[i] = 2;
else
bS[i] = 1;
}
}
mb_qp = s->current_picture.qscale_table[mb_xy];
mbn0_qp = s->current_picture.qscale_table[left_mb_xy[0]];
mbn1_qp = s->current_picture.qscale_table[left_mb_xy[1]];
qp[0] = ( mb_qp + mbn0_qp + 1 ) >> 1;
bqp[0] = ( get_chroma_qp( h, 0, mb_qp ) +
get_chroma_qp( h, 0, mbn0_qp ) + 1 ) >> 1;
rqp[0] = ( get_chroma_qp( h, 1, mb_qp ) +
get_chroma_qp( h, 1, mbn0_qp ) + 1 ) >> 1;
qp[1] = ( mb_qp + mbn1_qp + 1 ) >> 1;
bqp[1] = ( get_chroma_qp( h, 0, mb_qp ) +
get_chroma_qp( h, 0, mbn1_qp ) + 1 ) >> 1;
rqp[1] = ( get_chroma_qp( h, 1, mb_qp ) +
get_chroma_qp( h, 1, mbn1_qp ) + 1 ) >> 1;
/* Filter edge */
tprintf(s->avctx, "filter mb:%d/%d MBAFF, QPy:%d/%d, QPb:%d/%d QPr:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], bqp[0], bqp[1], rqp[0], rqp[1], linesize, uvlinesize);
{ int i; for (i = 0; i < 8; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "n"); }
filter_mb_mbaff_edgev ( h, &img_y [0], linesize, bS, qp );
filter_mb_mbaff_edgecv( h, &img_cb[0], uvlinesize, bS, bqp );
filter_mb_mbaff_edgecv( h, &img_cr[0], uvlinesize, bS, rqp );
}
#ifdef CONFIG_SMALL
for( dir = 0; dir < 2; dir++ )
filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, dir ? 0 : first_vertical_edge_done, dir);
#else
filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, first_vertical_edge_done, 0);
filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, 0, 1);
#endif
}
static int decode_slice(struct AVCodecContext *avctx, void *arg){
H264Context *h = *(void**)arg;
MpegEncContext * const s = &h->s;
const int part_mask= s->partitioned_frame ? (AC_END|AC_ERROR) : 0x7F;
s->mb_skip_run= -1;
h->is_complex = FRAME_MBAFF || s->picture_structure != PICT_FRAME || s->codec_id != CODEC_ID_H264 ||
(ENABLE_GRAY && (s->flags&CODEC_FLAG_GRAY)) || (ENABLE_H264_ENCODER && s->encoding);
if( h->pps.cabac ) {
int i;
/* realign */
align_get_bits( &s->gb );
/* init cabac */
ff_init_cabac_states( &h->cabac);
ff_init_cabac_decoder( &h->cabac,
s->gb.buffer + get_bits_count(&s->gb)/8,
( s->gb.size_in_bits - get_bits_count(&s->gb) + 7)/8);
/* calculate pre-state */
for( i= 0; i < 460; i++ ) {
int pre;
if( h->slice_type_nos == FF_I_TYPE )
pre = av_clip( ((cabac_context_init_I[i][0] * s->qscale) >>4 ) + cabac_context_init_I[i][1], 1, 126 );
else
pre = av_clip( ((cabac_context_init_PB[h->cabac_init_idc][i][0] * s->qscale) >>4 ) + cabac_context_init_PB[h->cabac_init_idc][i][1], 1, 126 );
if( pre <= 63 )
h->cabac_state[i] = 2 * ( 63 - pre ) + 0;
else
h->cabac_state[i] = 2 * ( pre - 64 ) + 1;
}
for(;;){
//START_TIMER
int ret = decode_mb_cabac(h);
int eos;
//STOP_TIMER("decode_mb_cabac")
if(ret>=0) hl_decode_mb(h);
if( ret >= 0 && FRAME_MBAFF ) { //FIXME optimal? or let mb_decode decode 16x32 ?
s->mb_y++;
ret = decode_mb_cabac(h);
if(ret>=0) hl_decode_mb(h);
s->mb_y--;
}
eos = get_cabac_terminate( &h->cabac );
if( ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 2) {
av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d, bytestream (%td)n", s->mb_x, s->mb_y, h->cabac.bytestream_end - h->cabac.bytestream);
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;
ff_draw_horiz_band(s, 16*s->mb_y, 16);
++s->mb_y;
if(FIELD_OR_MBAFF_PICTURE) {
++s->mb_y;
}
}
if( eos || s->mb_y >= s->mb_height ) {
tprintf(s->avctx, "slice end %d %dn", 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);
return 0;
}
}
} else {
for(;;){
int ret = decode_mb_cavlc(h);
if(ret>=0) hl_decode_mb(h);
if(ret>=0 && FRAME_MBAFF){ //FIXME optimal? or let mb_decode decode 16x32 ?
s->mb_y++;
ret = decode_mb_cavlc(h);
if(ret>=0) hl_decode_mb(h);
s->mb_y--;
}
if(ret<0){
av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %dn", 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;
ff_draw_horiz_band(s, 16*s->mb_y, 16);
++s->mb_y;
if(FIELD_OR_MBAFF_PICTURE) {
++s->mb_y;
}
if(s->mb_y >= s->mb_height){
tprintf(s->avctx, "slice end %d %dn", 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);
return 0;
}else{
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
return -1;
}
}
}
if(get_bits_count(&s->gb) >= s->gb.size_in_bits && s->mb_skip_run<=0){
tprintf(s->avctx, "slice end %d %dn", 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);
return 0;
}else{
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 0
for(;s->mb_y < s->mb_height; s->mb_y++){
for(;s->mb_x < s->mb_width; s->mb_x++){
int ret= decode_mb(h);
hl_decode_mb(h);
if(ret<0){
av_log(s->avctx, AV_LOG_ERROR, "error while decoding MB %d %dn", 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;
if(++s->mb_y >= s->mb_height){
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);
return 0;
}else{
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
return -1;
}
}
}
if(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);
return 0;
}else{
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;
}
}
}
s->mb_x=0;
ff_draw_horiz_band(s, 16*s->mb_y, 16);
}
#endif
return -1; //not reached
}
static int decode_picture_timing(H264Context *h){
MpegEncContext * const s = &h->s;
if(h->sps.nal_hrd_parameters_present_flag || h->sps.vcl_hrd_parameters_present_flag){
skip_bits(&s->gb, h->sps.cpb_removal_delay_length); /* cpb_removal_delay */
skip_bits(&s->gb, h->sps.dpb_output_delay_length); /* dpb_output_delay */
}
if(h->sps.pic_struct_present_flag){
unsigned int i, num_clock_ts;
h->sei_pic_struct = (SEI_PicStructType)get_bits(&s->gb, 4);
if (h->sei_pic_struct > SEI_PIC_STRUCT_FRAME_TRIPLING)
return -1;
num_clock_ts = sei_num_clock_ts_table[h->sei_pic_struct];
for (i = 0 ; i < num_clock_ts ; i++){
if(get_bits(&s->gb, 1)){ /* clock_timestamp_flag */
unsigned int full_timestamp_flag;
skip_bits(&s->gb, 2); /* ct_type */
skip_bits(&s->gb, 1); /* nuit_field_based_flag */
skip_bits(&s->gb, 5); /* counting_type */
full_timestamp_flag = get_bits(&s->gb, 1);
skip_bits(&s->gb, 1); /* discontinuity_flag */
skip_bits(&s->gb, 1); /* cnt_dropped_flag */
skip_bits(&s->gb, 8); /* n_frames */
if(full_timestamp_flag){
skip_bits(&s->gb, 6); /* seconds_value 0..59 */
skip_bits(&s->gb, 6); /* minutes_value 0..59 */
skip_bits(&s->gb, 5); /* hours_value 0..23 */
}else{
if(get_bits(&s->gb, 1)){ /* seconds_flag */
skip_bits(&s->gb, 6); /* seconds_value range 0..59 */
if(get_bits(&s->gb, 1)){ /* minutes_flag */
skip_bits(&s->gb, 6); /* minutes_value 0..59 */
if(get_bits(&s->gb, 1)) /* hours_flag */
skip_bits(&s->gb, 5); /* hours_value 0..23 */
}
}
}
if(h->sps.time_offset_length > 0)
skip_bits(&s->gb, h->sps.time_offset_length); /* time_offset */
}
}
}
return 0;
}
static int decode_unregistered_user_data(H264Context *h, int size){
MpegEncContext * const s = &h->s;
uint8_t user_data[16+256];
int e, build, i;
if(size<16)
return -1;
for(i=0; i<sizeof(user_data)-1 && i<size; i++){
user_data[i]= get_bits(&s->gb, 8);
}
user_data[i]= 0;
e= sscanf(user_data+16, "x264 - core %d"/*%s - H.264/MPEG-4 AVC codec - Copyleft 2005 - http://www.videolan.org/x264.html*/, &build);
if(e==1 && build>=0)
h->x264_build= build;
if(s->avctx->debug & FF_DEBUG_BUGS)
av_log(s->avctx, AV_LOG_DEBUG, "user data:"%s"n", user_data+16);
for(; i<size; i++)
skip_bits(&s->gb, 8);
return 0;
}
static int decode_sei(H264Context *h){
MpegEncContext * const s = &h->s;
while(get_bits_count(&s->gb) + 16 < s->gb.size_in_bits){
int size, type;
type=0;
do{
type+= show_bits(&s->gb, 8);
}while(get_bits(&s->gb, 8) == 255);
size=0;
do{
size+= show_bits(&s->gb, 8);
}while(get_bits(&s->gb, 8) == 255);
switch(type){
case 1: // Picture timing SEI
if(decode_picture_timing(h) < 0)
return -1;
break;
case 5:
if(decode_unregistered_user_data(h, size) < 0)
return -1;
break;
default:
skip_bits(&s->gb, 8*size);
}
//FIXME check bits here
align_get_bits(&s->gb);
}
return 0;
}
static inline int decode_hrd_parameters(H264Context *h, SPS *sps){
MpegEncContext * const s = &h->s;
int cpb_count, i;
cpb_count = get_ue_golomb_31(&s->gb) + 1;
if(cpb_count > 32U){
av_log(h->s.avctx, AV_LOG_ERROR, "cpb_count %d invalidn", cpb_count);
return -1;
}
get_bits(&s->gb, 4); /* bit_rate_scale */
get_bits(&s->gb, 4); /* cpb_size_scale */
for(i=0; i<cpb_count; i++){
get_ue_golomb(&s->gb); /* bit_rate_value_minus1 */
get_ue_golomb(&s->gb); /* cpb_size_value_minus1 */
get_bits1(&s->gb); /* cbr_flag */
}
get_bits(&s->gb, 5); /* initial_cpb_removal_delay_length_minus1 */
sps->cpb_removal_delay_length = get_bits(&s->gb, 5) + 1;
sps->dpb_output_delay_length = get_bits(&s->gb, 5) + 1;
sps->time_offset_length = get_bits(&s->gb, 5);
return 0;
}
static inline int decode_vui_parameters(H264Context *h, SPS *sps){
MpegEncContext * const s = &h->s;
int aspect_ratio_info_present_flag;
unsigned int aspect_ratio_idc;
aspect_ratio_info_present_flag= get_bits1(&s->gb);
if( aspect_ratio_info_present_flag ) {
aspect_ratio_idc= get_bits(&s->gb, 8);
if( aspect_ratio_idc == EXTENDED_SAR ) {
sps->sar.num= get_bits(&s->gb, 16);
sps->sar.den= get_bits(&s->gb, 16);
}else if(aspect_ratio_idc < FF_ARRAY_ELEMS(pixel_aspect)){
sps->sar= pixel_aspect[aspect_ratio_idc];
}else{
av_log(h->s.avctx, AV_LOG_ERROR, "illegal aspect ration");
return -1;
}
}else{
sps->sar.num=
sps->sar.den= 0;
}
// s->avctx->aspect_ratio= sar_width*s->width / (float)(s->height*sar_height);
if(get_bits1(&s->gb)){ /* overscan_info_present_flag */
get_bits1(&s->gb); /* overscan_appropriate_flag */
}
if(get_bits1(&s->gb)){ /* video_signal_type_present_flag */
get_bits(&s->gb, 3); /* video_format */
get_bits1(&s->gb); /* video_full_range_flag */
if(get_bits1(&s->gb)){ /* colour_description_present_flag */
get_bits(&s->gb, 8); /* colour_primaries */
get_bits(&s->gb, 8); /* transfer_characteristics */
get_bits(&s->gb, 8); /* matrix_coefficients */
}
}
if(get_bits1(&s->gb)){ /* chroma_location_info_present_flag */
get_ue_golomb(&s->gb); /* chroma_sample_location_type_top_field */
get_ue_golomb(&s->gb); /* chroma_sample_location_type_bottom_field */
}
sps->timing_info_present_flag = get_bits1(&s->gb);
if(sps->timing_info_present_flag){
sps->num_units_in_tick = get_bits_long(&s->gb, 32);
sps->time_scale = get_bits_long(&s->gb, 32);
sps->fixed_frame_rate_flag = get_bits1(&s->gb);
}
sps->nal_hrd_parameters_present_flag = get_bits1(&s->gb);
if(sps->nal_hrd_parameters_present_flag)
if(decode_hrd_parameters(h, sps) < 0)
return -1;
sps->vcl_hrd_parameters_present_flag = get_bits1(&s->gb);
if(sps->vcl_hrd_parameters_present_flag)
if(decode_hrd_parameters(h, sps) < 0)
return -1;
if(sps->nal_hrd_parameters_present_flag || sps->vcl_hrd_parameters_present_flag)
get_bits1(&s->gb); /* low_delay_hrd_flag */
sps->pic_struct_present_flag = get_bits1(&s->gb);
sps->bitstream_restriction_flag = get_bits1(&s->gb);
if(sps->bitstream_restriction_flag){
get_bits1(&s->gb); /* motion_vectors_over_pic_boundaries_flag */
get_ue_golomb(&s->gb); /* max_bytes_per_pic_denom */
get_ue_golomb(&s->gb); /* max_bits_per_mb_denom */
get_ue_golomb(&s->gb); /* log2_max_mv_length_horizontal */
get_ue_golomb(&s->gb); /* log2_max_mv_length_vertical */
sps->num_reorder_frames= get_ue_golomb(&s->gb);
get_ue_golomb(&s->gb); /*max_dec_frame_buffering*/
if(sps->num_reorder_frames > 16U /*max_dec_frame_buffering || max_dec_frame_buffering > 16*/){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal num_reorder_frames %dn", sps->num_reorder_frames);
return -1;
}
}
return 0;
}
static void decode_scaling_list(H264Context *h, uint8_t *factors, int size,
const uint8_t *jvt_list, const uint8_t *fallback_list){
MpegEncContext * const s = &h->s;
int i, last = 8, next = 8;
const uint8_t *scan = size == 16 ? zigzag_scan : zigzag_scan8x8;
if(!get_bits1(&s->gb)) /* matrix not written, we use the predicted one */
memcpy(factors, fallback_list, size*sizeof(uint8_t));
else
for(i=0;i<size;i++){
if(next)
next = (last + get_se_golomb(&s->gb)) & 0xff;
if(!i && !next){ /* matrix not written, we use the preset one */
memcpy(factors, jvt_list, size*sizeof(uint8_t));
break;
}
last = factors[scan[i]] = next ? next : last;
}
}
static void decode_scaling_matrices(H264Context *h, SPS *sps, PPS *pps, int is_sps,
uint8_t (*scaling_matrix4)[16], uint8_t (*scaling_matrix8)[64]){
MpegEncContext * const s = &h->s;
int fallback_sps = !is_sps && sps->scaling_matrix_present;
const uint8_t *fallback[4] = {
fallback_sps ? sps->scaling_matrix4[0] : default_scaling4[0],
fallback_sps ? sps->scaling_matrix4[3] : default_scaling4[1],
fallback_sps ? sps->scaling_matrix8[0] : default_scaling8[0],
fallback_sps ? sps->scaling_matrix8[1] : default_scaling8[1]
};
if(get_bits1(&s->gb)){
sps->scaling_matrix_present |= is_sps;
decode_scaling_list(h,scaling_matrix4[0],16,default_scaling4[0],fallback[0]); // Intra, Y
decode_scaling_list(h,scaling_matrix4[1],16,default_scaling4[0],scaling_matrix4[0]); // Intra, Cr
decode_scaling_list(h,scaling_matrix4[2],16,default_scaling4[0],scaling_matrix4[1]); // Intra, Cb
decode_scaling_list(h,scaling_matrix4[3],16,default_scaling4[1],fallback[1]); // Inter, Y
decode_scaling_list(h,scaling_matrix4[4],16,default_scaling4[1],scaling_matrix4[3]); // Inter, Cr
decode_scaling_list(h,scaling_matrix4[5],16,default_scaling4[1],scaling_matrix4[4]); // Inter, Cb
if(is_sps || pps->transform_8x8_mode){
decode_scaling_list(h,scaling_matrix8[0],64,default_scaling8[0],fallback[2]); // Intra, Y
decode_scaling_list(h,scaling_matrix8[1],64,default_scaling8[1],fallback[3]); // Inter, Y
}
}
}
static inline int decode_seq_parameter_set(H264Context *h){
MpegEncContext * const s = &h->s;
int profile_idc, level_idc;
unsigned int sps_id;
int i;
SPS *sps;
profile_idc= get_bits(&s->gb, 8);
get_bits1(&s->gb); //constraint_set0_flag
get_bits1(&s->gb); //constraint_set1_flag
get_bits1(&s->gb); //constraint_set2_flag
get_bits1(&s->gb); //constraint_set3_flag
get_bits(&s->gb, 4); // reserved
level_idc= get_bits(&s->gb, 8);
sps_id= get_ue_golomb_31(&s->gb);
if(sps_id >= MAX_SPS_COUNT) {
av_log(h->s.avctx, AV_LOG_ERROR, "sps_id (%d) out of rangen", sps_id);
return -1;
}
sps= av_mallocz(sizeof(SPS));
if(sps == NULL)
return -1;
sps->profile_idc= profile_idc;
sps->level_idc= level_idc;
memset(sps->scaling_matrix4, 16, sizeof(sps->scaling_matrix4));
memset(sps->scaling_matrix8, 16, sizeof(sps->scaling_matrix8));
sps->scaling_matrix_present = 0;
if(sps->profile_idc >= 100){ //high profile
sps->chroma_format_idc= get_ue_golomb_31(&s->gb);
if(sps->chroma_format_idc == 3)
get_bits1(&s->gb); //residual_color_transform_flag
get_ue_golomb(&s->gb); //bit_depth_luma_minus8
get_ue_golomb(&s->gb); //bit_depth_chroma_minus8
sps->transform_bypass = get_bits1(&s->gb);
decode_scaling_matrices(h, sps, NULL, 1, sps->scaling_matrix4, sps->scaling_matrix8);
}else{
sps->chroma_format_idc= 1;
}
sps->log2_max_frame_num= get_ue_golomb(&s->gb) + 4;
sps->poc_type= get_ue_golomb_31(&s->gb);
if(sps->poc_type == 0){ //FIXME #define
sps->log2_max_poc_lsb= get_ue_golomb(&s->gb) + 4;
} else if(sps->poc_type == 1){//FIXME #define
sps->delta_pic_order_always_zero_flag= get_bits1(&s->gb);
sps->offset_for_non_ref_pic= get_se_golomb(&s->gb);
sps->offset_for_top_to_bottom_field= get_se_golomb(&s->gb);
sps->poc_cycle_length = get_ue_golomb(&s->gb);
if((unsigned)sps->poc_cycle_length >= FF_ARRAY_ELEMS(sps->offset_for_ref_frame)){
av_log(h->s.avctx, AV_LOG_ERROR, "poc_cycle_length overflow %un", sps->poc_cycle_length);
goto fail;
}
for(i=0; i<sps->poc_cycle_length; i++)
sps->offset_for_ref_frame[i]= get_se_golomb(&s->gb);
}else if(sps->poc_type != 2){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal POC type %dn", sps->poc_type);
goto fail;
}
sps->ref_frame_count= get_ue_golomb_31(&s->gb);
if(sps->ref_frame_count > MAX_PICTURE_COUNT-2 || sps->ref_frame_count >= 32U){
av_log(h->s.avctx, AV_LOG_ERROR, "too many reference framesn");
goto fail;
}
sps->gaps_in_frame_num_allowed_flag= get_bits1(&s->gb);
sps->mb_width = get_ue_golomb(&s->gb) + 1;
sps->mb_height= get_ue_golomb(&s->gb) + 1;
if((unsigned)sps->mb_width >= INT_MAX/16 || (unsigned)sps->mb_height >= INT_MAX/16 ||
avcodec_check_dimensions(NULL, 16*sps->mb_width, 16*sps->mb_height)){
av_log(h->s.avctx, AV_LOG_ERROR, "mb_width/height overflown");
goto fail;
}
sps->frame_mbs_only_flag= get_bits1(&s->gb);
if(!sps->frame_mbs_only_flag)
sps->mb_aff= get_bits1(&s->gb);
else
sps->mb_aff= 0;
sps->direct_8x8_inference_flag= get_bits1(&s->gb);
#ifndef ALLOW_INTERLACE
if(sps->mb_aff)
av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF support not included; enable it at compile-time.n");
#endif
sps->crop= get_bits1(&s->gb);
if(sps->crop){
sps->crop_left = get_ue_golomb(&s->gb);
sps->crop_right = get_ue_golomb(&s->gb);
sps->crop_top = get_ue_golomb(&s->gb);
sps->crop_bottom= get_ue_golomb(&s->gb);
if(sps->crop_left || sps->crop_top){
av_log(h->s.avctx, AV_LOG_ERROR, "insane cropping not completely supported, this could look slightly wrong ...n");
}
if((int)sps->crop_right >= 8 || (int)sps->crop_bottom >= (8>> !sps->frame_mbs_only_flag)){
av_log(h->s.avctx, AV_LOG_ERROR, "brainfart cropping not supported, this could look slightly wrong ...n");
}
}else{
sps->crop_left =
sps->crop_right =
sps->crop_top =
sps->crop_bottom= 0;
}
sps->vui_parameters_present_flag= get_bits1(&s->gb);
if( sps->vui_parameters_present_flag )
decode_vui_parameters(h, sps);
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
av_log(h->s.avctx, AV_LOG_DEBUG, "sps:%u profile:%d/%d poc:%d ref:%d %dx%d %s %s crop:%d/%d/%d/%d %s %sn",
sps_id, sps->profile_idc, sps->level_idc,
sps->poc_type,
sps->ref_frame_count,
sps->mb_width, sps->mb_height,
sps->frame_mbs_only_flag ? "FRM" : (sps->mb_aff ? "MB-AFF" : "PIC-AFF"),
sps->direct_8x8_inference_flag ? "8B8" : "",
sps->crop_left, sps->crop_right,
sps->crop_top, sps->crop_bottom,
sps->vui_parameters_present_flag ? "VUI" : ""
//((const char*[]){"Gray","420","422","444"})[sps->chroma_format_idc]
);
}
av_free(h->sps_buffers[sps_id]);
h->sps_buffers[sps_id]= sps;
return 0;
fail:
av_free(sps);
return -1;
}
static void
build_qp_table(PPS *pps, int t, int index)
{
int i;
for(i = 0; i < 52; i++)
pps->chroma_qp_table[t][i] = chroma_qp[av_clip(i + index, 0, 51)];
}
static inline int decode_picture_parameter_set(H264Context *h, int bit_length){
MpegEncContext * const s = &h->s;
unsigned int pps_id= get_ue_golomb(&s->gb);
PPS *pps;
if(pps_id >= MAX_PPS_COUNT) {
av_log(h->s.avctx, AV_LOG_ERROR, "pps_id (%d) out of rangen", pps_id);
return -1;
}
pps= av_mallocz(sizeof(PPS));
if(pps == NULL)
return -1;
pps->sps_id= get_ue_golomb_31(&s->gb);
if((unsigned)pps->sps_id>=MAX_SPS_COUNT || h->sps_buffers[pps->sps_id] == NULL){
av_log(h->s.avctx, AV_LOG_ERROR, "sps_id out of rangen");
goto fail;
}
pps->cabac= get_bits1(&s->gb);
pps->pic_order_present= get_bits1(&s->gb);
pps->slice_group_count= get_ue_golomb(&s->gb) + 1;
if(pps->slice_group_count > 1 ){
pps->mb_slice_group_map_type= get_ue_golomb(&s->gb);
av_log(h->s.avctx, AV_LOG_ERROR, "FMO not supportedn");
switch(pps->mb_slice_group_map_type){
case 0:
#if 0
| for( i = 0; i <= num_slice_groups_minus1; i++ ) | | |
| run_length[ i ] |1 |ue(v) |
#endif
break;
case 2:
#if 0
| for( i = 0; i < num_slice_groups_minus1; i++ ) | | |
|{ | | |
| top_left_mb[ i ] |1 |ue(v) |
| bottom_right_mb[ i ] |1 |ue(v) |
| } | | |
#endif
break;
case 3:
case 4:
case 5:
#if 0
| slice_group_change_direction_flag |1 |u(1) |
| slice_group_change_rate_minus1 |1 |ue(v) |
#endif
break;
case 6:
#if 0
| slice_group_id_cnt_minus1 |1 |ue(v) |
| for( i = 0; i <= slice_group_id_cnt_minus1; i++ | | |
|) | | |
| slice_group_id[ i ] |1 |u(v) |
#endif
break;
}
}
pps->ref_count[0]= get_ue_golomb(&s->gb) + 1;
pps->ref_count[1]= get_ue_golomb(&s->gb) + 1;
if(pps->ref_count[0]-1 > 32-1 || pps->ref_count[1]-1 > 32-1){
av_log(h->s.avctx, AV_LOG_ERROR, "reference overflow (pps)n");
goto fail;
}
pps->weighted_pred= get_bits1(&s->gb);
pps->weighted_bipred_idc= get_bits(&s->gb, 2);
pps->init_qp= get_se_golomb(&s->gb) + 26;
pps->init_qs= get_se_golomb(&s->gb) + 26;
pps->chroma_qp_index_offset[0]= get_se_golomb(&s->gb);
pps->deblocking_filter_parameters_present= get_bits1(&s->gb);
pps->constrained_intra_pred= get_bits1(&s->gb);
pps->redundant_pic_cnt_present = get_bits1(&s->gb);
pps->transform_8x8_mode= 0;
h->dequant_coeff_pps= -1; //contents of sps/pps can change even if id doesn't, so reinit
memcpy(pps->scaling_matrix4, h->sps_buffers[pps->sps_id]->scaling_matrix4, sizeof(pps->scaling_matrix4));
memcpy(pps->scaling_matrix8, h->sps_buffers[pps->sps_id]->scaling_matrix8, sizeof(pps->scaling_matrix8));
if(get_bits_count(&s->gb) < bit_length){
pps->transform_8x8_mode= get_bits1(&s->gb);
decode_scaling_matrices(h, h->sps_buffers[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8);
pps->chroma_qp_index_offset[1]= get_se_golomb(&s->gb); //second_chroma_qp_index_offset
} else {
pps->chroma_qp_index_offset[1]= pps->chroma_qp_index_offset[0];
}
build_qp_table(pps, 0, pps->chroma_qp_index_offset[0]);
build_qp_table(pps, 1, pps->chroma_qp_index_offset[1]);
if(pps->chroma_qp_index_offset[0] != pps->chroma_qp_index_offset[1])
h->pps.chroma_qp_diff= 1;
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
av_log(h->s.avctx, AV_LOG_DEBUG, "pps:%u sps:%u %s slice_groups:%d ref:%d/%d %s qp:%d/%d/%d/%d %s %s %s %sn",
pps_id, pps->sps_id,
pps->cabac ? "CABAC" : "CAVLC",
pps->slice_group_count,
pps->ref_count[0], pps->ref_count[1],
pps->weighted_pred ? "weighted" : "",
pps->init_qp, pps->init_qs, pps->chroma_qp_index_offset[0], pps->chroma_qp_index_offset[1],
pps->deblocking_filter_parameters_present ? "LPAR" : "",
pps->constrained_intra_pred ? "CONSTR" : "",
pps->redundant_pic_cnt_present ? "REDU" : "",
pps->transform_8x8_mode ? "8x8DCT" : ""
);
}
av_free(h->pps_buffers[pps_id]);
h->pps_buffers[pps_id]= pps;
return 0;
fail:
av_free(pps);
return -1;
}
/**
* Call decode_slice() for each context.
*
* @param h h264 master context
* @param context_count number of contexts to execute
*/
static void execute_decode_slices(H264Context *h, int context_count){
MpegEncContext * const s = &h->s;
AVCodecContext * const avctx= s->avctx;
H264Context *hx;
int i;
if(avctx->codec_id == CODEC_ID_H264_VDPAU)
return;
if(context_count == 1)
{
decode_slice(avctx, &h);
}
else
{
for(i = 1; i < context_count; i++)
{
hx = h->thread_context[i];
hx->s.error_recognition = avctx->error_recognition;
hx->s.error_count = 0;
}
avctx->execute(avctx, decode_slice, (void **)h->thread_context, NULL, context_count, sizeof(void*));
/* pull back stuff from slices to master context */
hx = h->thread_context[context_count - 1];
s->mb_x = hx->s.mb_x;
s->mb_y = hx->s.mb_y;
s->dropable = hx->s.dropable;
s->picture_structure = hx->s.picture_structure;
for(i = 1; i < context_count; i++)
h->s.error_count += h->thread_context[i]->s.error_count;
}
}
static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size){
MpegEncContext * const s = &h->s;
AVCodecContext * const avctx= s->avctx;
int buf_index=0;
H264Context *hx; ///< thread context
int context_count = 0;
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)
s->current_picture_ptr= NULL;
}
for(;;){
int consumed;
int dst_length;
int bit_length;
const uint8_t *ptr;
int i, nalsize = 0;
int err;
if(h->is_avc) {
if(buf_index >= buf_size) break;
nalsize = 0;
for(i = 0; i < h->nal_length_size; i++)
nalsize = (nalsize << 8) | buf[buf_index++];
if(nalsize <= 1 || (nalsize+buf_index > buf_size)){
if(nalsize == 1){
buf_index++;
continue;
}else{
av_log(h->s.avctx, AV_LOG_ERROR, "AVC: nal size %dn", nalsize);
break;
}
}
} else {
// start code prefix search
for(; buf_index + 3 < buf_size; buf_index++){
// This should always succeed in the first iteration.
if(buf[buf_index] == 0 && buf[buf_index+1] == 0 && buf[buf_index+2] == 1)
break;
}
if(buf_index+3 >= buf_size) break;
buf_index+=3;
}
hx = h->thread_context[context_count];
ptr= decode_nal(hx, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index);
if (ptr==NULL || dst_length < 0){
return -1;
}
while(ptr[dst_length - 1] == 0 && dst_length > 0)
dst_length--;
bit_length= !dst_length ? 0 : (8*dst_length - decode_rbsp_trailing(h, ptr + dst_length - 1));
if(s->avctx->debug&FF_DEBUG_STARTCODE){
av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d at %d/%d length %dn", hx->nal_unit_type, buf_index, buf_size, dst_length);
}
if (h->is_avc && (nalsize != consumed)){
av_log(h->s.avctx, AV_LOG_ERROR, "AVC: Consumed only %d bytes instead of %dn", consumed, nalsize);
consumed= nalsize;
}
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))
continue;
again:
err = 0;
switch(hx->nal_unit_type){
case NAL_IDR_SLICE:
if (h->nal_unit_type != NAL_IDR_SLICE) {
av_log(h->s.avctx, AV_LOG_ERROR, "Invalid mix of idr and non-idr slices");
return -1;
}
idr(h); //FIXME ensure we don't loose some frames if there is reordering
case NAL_SLICE:
init_get_bits(&hx->s.gb, ptr, bit_length);
hx->intra_gb_ptr=
hx->inter_gb_ptr= &hx->s.gb;
hx->s.data_partitioning = 0;
if((err = decode_slice_header(hx, h)))
break;
s->current_picture_ptr->key_frame|= (hx->nal_unit_type == NAL_IDR_SLICE);
if(hx->redundant_pic_count==0 && hx->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_ALL)
{
/*if(ENABLE_H264_VDPAU_DECODER && avctx->codec_id == CODEC_ID_H264_VDPAU)
{
static const uint8_t start_code[] = {0x00, 0x00, 0x01};
ff_vdpau_add_data_chunk(s, start_code, sizeof(start_code));
ff_vdpau_add_data_chunk(s, &buf[buf_index - consumed], consumed );
}
else*/
context_count++;
}
break;
case NAL_DPA:
init_get_bits(&hx->s.gb, ptr, bit_length);
hx->intra_gb_ptr=
hx->inter_gb_ptr= NULL;
hx->s.data_partitioning = 1;
err = decode_slice_header(hx, h);
break;
case NAL_DPB:
init_get_bits(&hx->intra_gb, ptr, bit_length);
hx->intra_gb_ptr= &hx->intra_gb;
break;
case NAL_DPC:
init_get_bits(&hx->inter_gb, ptr, bit_length);
hx->inter_gb_ptr= &hx->inter_gb;
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_ALL)
context_count++;
break;
case NAL_SEI:
init_get_bits(&s->gb, ptr, bit_length);
decode_sei(h);
break;
case NAL_SPS:
init_get_bits(&s->gb, ptr, bit_length);
decode_seq_parameter_set(h);
if(s->flags& CODEC_FLAG_LOW_DELAY)
s->low_delay=1;
if(avctx->has_b_frames < 2)
avctx->has_b_frames= !s->low_delay;
break;
case NAL_PPS:
init_get_bits(&s->gb, ptr, bit_length);
decode_picture_parameter_set(h, bit_length);
break;
case NAL_AUD:
case NAL_END_SEQUENCE:
case NAL_END_STREAM:
case NAL_FILLER_DATA:
case NAL_SPS_EXT:
case NAL_AUXILIARY_SLICE:
break;
default:
av_log(avctx, AV_LOG_DEBUG, "Unknown NAL code: %d (%d bits)n", h->nal_unit_type, bit_length);
}
if(context_count == h->max_contexts) {
execute_decode_slices(h, context_count);
context_count = 0;
}
if (err < 0)
av_log(h->s.avctx, AV_LOG_ERROR, "decode_slice_header errorn");
else if(err == 1) {
/* Slice could not be decoded in parallel mode, copy down
* NAL unit stuff to context 0 and restart. Note that
* rbsp_buffer is not transferred, but since we no longer
* run in parallel mode this should not be an issue. */
h->nal_unit_type = hx->nal_unit_type;
h->nal_ref_idc = hx->nal_ref_idc;
hx = h;
goto again;
}
}
if(context_count)
execute_decode_slices(h, context_count);
return buf_index;
}
/**
* returns the number of bytes consumed for building the current frame
*/
static int get_consumed_bytes(MpegEncContext *s, int pos, int buf_size){
if(pos==0) pos=1; //avoid infinite loops (i doubt that is needed but ...)
if(pos+10>buf_size) pos=buf_size; // oops ;)
return pos;
}
static int decode_frame(AVCodecContext *avctx,
void *data, int *data_size,
const uint8_t *buf, int buf_size)
{
H264Context *h = avctx->priv_data;
MpegEncContext *s = &h->s;
AVFrame *pict = data;
int buf_index;
s->flags= avctx->flags;
s->flags2= avctx->flags2;
/* end of stream, output what is still in the buffers */
if (buf_size == 0) {
Picture *out;
int i, out_idx;
//FIXME factorize this with the output code below
out = h->delayed_pic[0];
out_idx = 0;
for(i=1; h->delayed_pic[i] && (h->delayed_pic[i]->poc && !h->delayed_pic[i]->key_frame); i++)
if(h->delayed_pic[i]->poc < out->poc){
out = h->delayed_pic[i];
out_idx = i;
}
for(i=out_idx; h->delayed_pic[i]; i++)
h->delayed_pic[i] = h->delayed_pic[i+1];
if(out){
*data_size = sizeof(AVFrame);
*pict= *(AVFrame*)out;
}
return 0;
}
if(h->is_avc && !h->got_avcC) {
int i, cnt, nalsize;
unsigned char *p = avctx->extradata;
if(avctx->extradata_size < 7) {
av_log(avctx, AV_LOG_ERROR, "avcC too shortn");
return -1;
}
if(*p != 1) {
av_log(avctx, AV_LOG_ERROR, "Unknown avcC version %dn", *p);
return -1;
}
/* sps and pps in the avcC always have length coded with 2 bytes,
so put a fake nal_length_size = 2 while parsing them */
h->nal_length_size = 2;
// Decode sps from avcC
cnt = *(p+5) & 0x1f; // Number of sps
p += 6;
for (i = 0; i < cnt; i++) {
nalsize = AV_RB16(p) + 2;
if(decode_nal_units(h, p, nalsize) < 0) {
av_log(avctx, AV_LOG_ERROR, "Decoding sps %d from avcC failedn", i);
return -1;
}
p += nalsize;
}
// Decode pps from avcC
cnt = *(p++); // Number of pps
for (i = 0; i < cnt; i++) {
nalsize = AV_RB16(p) + 2;
if(decode_nal_units(h, p, nalsize) != nalsize) {
av_log(avctx, AV_LOG_ERROR, "Decoding pps %d from avcC failedn", i);
return -1;
}
p += nalsize;
}
// Now store right nal length size, that will be use to parse all other nals
h->nal_length_size = ((*(((char*)(avctx->extradata))+4))&0x03)+1;
// Do not reparse avcC
h->got_avcC = 1;
}
if(!h->got_avcC && !h->is_avc && s->avctx->extradata_size){
if(decode_nal_units(h, s->avctx->extradata, s->avctx->extradata_size) < 0)
return -1;
h->got_avcC = 1;
}
buf_index=decode_nal_units(h, buf, buf_size);
if(buf_index < 0)
return -1;
if(!(s->flags2 & CODEC_FLAG2_CHUNKS) && !s->current_picture_ptr){
if (avctx->skip_frame >= AVDISCARD_NONREF || s->hurry_up) return 0;
av_log(avctx, AV_LOG_ERROR, "no frame!n");
return -1;
}
if(!(s->flags2 & CODEC_FLAG2_CHUNKS) || (s->mb_y >= s->mb_height && s->mb_height)){
Picture *out = s->current_picture_ptr;
Picture *cur = s->current_picture_ptr;
int i, pics, cross_idr, out_of_order, out_idx;
s->mb_y= 0;
s->current_picture_ptr->qscale_type= FF_QSCALE_TYPE_H264;
s->current_picture_ptr->pict_type= s->pict_type;
//if (ENABLE_H264_VDPAU_DECODER && avctx->codec_id == CODEC_ID_H264_VDPAU)
// ff_vdpau_h264_set_reference_frames(s);
if(!s->dropable) {
execute_ref_pic_marking(h, h->mmco, h->mmco_index);
h->prev_poc_msb= h->poc_msb;
h->prev_poc_lsb= h->poc_lsb;
}
h->prev_frame_num_offset= h->frame_num_offset;
h->prev_frame_num= h->frame_num;
//if (ENABLE_H264_VDPAU_DECODER && avctx->codec_id == CODEC_ID_H264_VDPAU)
// ff_vdpau_h264_picture_complete(s);
/*
* FIXME: Error handling code does not seem to support interlaced
* when slices span multiple rows
* The ff_er_add_slice calls don't work right for bottom
* fields; they cause massive erroneous error concealing
* Error marking covers both fields (top and bottom).
* This causes a mismatched s->error_count
* and a bad error table. Further, the error count goes to
* INT_MAX when called for bottom field, because mb_y is
* past end by one (callers fault) and resync_mb_y != 0
* causes problems for the first MB line, too.
*/
if (!FIELD_PICTURE)
ff_er_frame_end(s);
MPV_frame_end(s);
if (cur->field_poc[0]==INT_MAX || cur->field_poc[1]==INT_MAX) {
/* Wait for second field. */
*data_size = 0;
} else {
cur->repeat_pict = 0;
/* Signal interlacing information externally. */
/* Prioritize picture timing SEI information over used decoding process if it exists. */
if(h->sps.pic_struct_present_flag){
switch (h->sei_pic_struct)
{
case SEI_PIC_STRUCT_FRAME:
cur->interlaced_frame = 0;
break;
case SEI_PIC_STRUCT_TOP_FIELD:
case SEI_PIC_STRUCT_BOTTOM_FIELD:
case SEI_PIC_STRUCT_TOP_BOTTOM:
case SEI_PIC_STRUCT_BOTTOM_TOP:
cur->interlaced_frame = 1;
break;
case SEI_PIC_STRUCT_TOP_BOTTOM_TOP:
case SEI_PIC_STRUCT_BOTTOM_TOP_BOTTOM:
// Signal the possibility of telecined film externally (pic_struct 5,6)
// From these hints, let the applications decide if they apply deinterlacing.
cur->repeat_pict = 1;
cur->interlaced_frame = FIELD_OR_MBAFF_PICTURE;
break;
case SEI_PIC_STRUCT_FRAME_DOUBLING:
// Force progressive here, as doubling interlaced frame is a bad idea.
cur->interlaced_frame = 0;
cur->repeat_pict = 2;
break;
case SEI_PIC_STRUCT_FRAME_TRIPLING:
cur->interlaced_frame = 0;
cur->repeat_pict = 4;
break;
}
}else{
/* Derive interlacing flag from used decoding process. */
cur->interlaced_frame = FIELD_OR_MBAFF_PICTURE;
}
if (cur->field_poc[0] != cur->field_poc[1]){
/* Derive top_field_first from field pocs. */
cur->top_field_first = cur->field_poc[0] < cur->field_poc[1];
}else{
if(cur->interlaced_frame || h->sps.pic_struct_present_flag){
/* Use picture timing SEI information. Even if it is a information of a past frame, better than nothing. */
if(h->sei_pic_struct == SEI_PIC_STRUCT_TOP_BOTTOM
|| h->sei_pic_struct == SEI_PIC_STRUCT_TOP_BOTTOM_TOP)
cur->top_field_first = 1;
else
cur->top_field_first = 0;
}else{
/* Most likely progressive */
cur->top_field_first = 0;
}
}
//FIXME do something with unavailable reference frames
/* Sort B-frames into display order */
if(h->sps.bitstream_restriction_flag
&& s->avctx->has_b_frames < h->sps.num_reorder_frames){
s->avctx->has_b_frames = h->sps.num_reorder_frames;
s->low_delay = 0;
}
if( s->avctx->strict_std_compliance >= FF_COMPLIANCE_STRICT
&& !h->sps.bitstream_restriction_flag){
s->avctx->has_b_frames= MAX_DELAYED_PIC_COUNT;
s->low_delay= 0;
}
pics = 0;
while(h->delayed_pic[pics]) pics++;
assert(pics <= MAX_DELAYED_PIC_COUNT);
h->delayed_pic[pics++] = cur;
if(cur->reference == 0)
cur->reference = DELAYED_PIC_REF;
out = h->delayed_pic[0];
out_idx = 0;
for(i=1; h->delayed_pic[i] && (h->delayed_pic[i]->poc && !h->delayed_pic[i]->key_frame); i++)
if(h->delayed_pic[i]->poc < out->poc){
out = h->delayed_pic[i];
out_idx = i;
}
cross_idr = !h->delayed_pic[0]->poc || !!h->delayed_pic[i] || h->delayed_pic[0]->key_frame;
out_of_order = !cross_idr && out->poc < h->outputed_poc;
if(h->sps.bitstream_restriction_flag && s->avctx->has_b_frames >= h->sps.num_reorder_frames)
{ }
else if((out_of_order && pics-1 == s->avctx->has_b_frames && s->avctx->has_b_frames < MAX_DELAYED_PIC_COUNT)
|| (s->low_delay &&
((!cross_idr && out->poc > h->outputed_poc + 2)
|| cur->pict_type == FF_B_TYPE)))
{
s->low_delay = 0;
s->avctx->has_b_frames++;
}
if(out_of_order || pics > s->avctx->has_b_frames){
out->reference &= ~DELAYED_PIC_REF;
for(i=out_idx; h->delayed_pic[i]; i++)
h->delayed_pic[i] = h->delayed_pic[i+1];
}
if(!out_of_order && pics > s->avctx->has_b_frames){
*data_size = sizeof(AVFrame);
h->outputed_poc = out->poc;
*pict= *(AVFrame*)out;
}else{
av_log(avctx, AV_LOG_DEBUG, "no picturen");
}
}
}
assert(pict->data[0] || !*data_size);
//ff_print_debug_info(s, pict);
//printf("out %dn", (int)pict->data[0]);
#if 0 //?
/* Return the Picture timestamp as the frame number */
/* we subtract 1 because it is added on utils.c */
avctx->frame_number = s->picture_number - 1;
#endif
return get_consumed_bytes(s, buf_index, buf_size);
}
#if 0
static inline void fill_mb_avail(H264Context *h){
MpegEncContext * const s = &h->s;
const int mb_xy= s->mb_x + s->mb_y*s->mb_stride;
if(s->mb_y){
h->mb_avail[0]= s->mb_x && h->slice_table[mb_xy - s->mb_stride - 1] == h->slice_num;
h->mb_avail[1]= h->slice_table[mb_xy - s->mb_stride ] == h->slice_num;
h->mb_avail[2]= s->mb_x+1 < s->mb_width && h->slice_table[mb_xy - s->mb_stride + 1] == h->slice_num;
}else{
h->mb_avail[0]=
h->mb_avail[1]=
h->mb_avail[2]= 0;
}
h->mb_avail[3]= s->mb_x && h->slice_table[mb_xy - 1] == h->slice_num;
h->mb_avail[4]= 1; //FIXME move out
h->mb_avail[5]= 0; //FIXME move out
}
#endif
#ifdef TEST
#undef printf
#undef random
#define COUNT 8000
#define SIZE (COUNT*40)
int main(void){
int i;
uint8_t temp[SIZE];
PutBitContext pb;
GetBitContext gb;
// int int_temp[10000];
DSPContext dsp;
AVCodecContext avctx;
dsputil_init(&dsp, &avctx);
init_put_bits(&pb, temp, SIZE);
printf("testing unsigned exp golombn");
for(i=0; i<COUNT; i++){
START_TIMER
set_ue_golomb(&pb, i);
STOP_TIMER("set_ue_golomb");
}
flush_put_bits(&pb);
init_get_bits(&gb, temp, 8*SIZE);
for(i=0; i<COUNT; i++){
int j, s;
s= show_bits(&gb, 24);
START_TIMER
j= get_ue_golomb(&gb);
if(j != i){
printf("mismatch! at %d (%d should be %d) bits:%6Xn", i, j, i, s);
// return -1;
}
STOP_TIMER("get_ue_golomb");
}
init_put_bits(&pb, temp, SIZE);
printf("testing signed exp golombn");
for(i=0; i<COUNT; i++){
START_TIMER
set_se_golomb(&pb, i - COUNT/2);
STOP_TIMER("set_se_golomb");
}
flush_put_bits(&pb);
init_get_bits(&gb, temp, 8*SIZE);
for(i=0; i<COUNT; i++){
int j, s;
s= show_bits(&gb, 24);
START_TIMER
j= get_se_golomb(&gb);
if(j != i - COUNT/2){
printf("mismatch! at %d (%d should be %d) bits:%6Xn", i, j, i, s);
// return -1;
}
STOP_TIMER("get_se_golomb");
}
#if 0
printf("testing 4x4 (I)DCTn");
DCTELEM block[16];
uint8_t src[16], ref[16];
uint64_t error= 0, max_error=0;
for(i=0; i<COUNT; i++){
int j;
// printf("%d %d %dn", r1, r2, (r2-r1)*16);
for(j=0; j<16; j++){
ref[j]= random()%255;
src[j]= random()%255;
}
h264_diff_dct_c(block, src, ref, 4);
//normalize
for(j=0; j<16; j++){
// printf("%d ", block[j]);
block[j]= block[j]*4;
if(j&1) block[j]= (block[j]*4 + 2)/5;
if(j&4) block[j]= (block[j]*4 + 2)/5;
}
// printf("n");
s->dsp.h264_idct_add(ref, block, 4);
/* for(j=0; j<16; j++){
printf("%d ", ref[j]);
}
printf("n");*/
for(j=0; j<16; j++){
int diff= FFABS(src[j] - ref[j]);
error+= diff*diff;
max_error= FFMAX(max_error, diff);
}
}
printf("error=%f max_error=%dn", ((float)error)/COUNT/16, (int)max_error );
printf("testing quantizern");
for(qp=0; qp<52; qp++){
for(i=0; i<16; i++)
src1_block[i]= src2_block[i]= random()%255;
}
printf("Testing NAL layern");
uint8_t bitstream[COUNT];
uint8_t nal[COUNT*2];
H264Context h;
memset(&h, 0, sizeof(H264Context));
for(i=0; i<COUNT; i++){
int zeros= i;
int nal_length;
int consumed;
int out_length;
uint8_t *out;
int j;
for(j=0; j<COUNT; j++){
bitstream[j]= (random() % 255) + 1;
}
for(j=0; j<zeros; j++){
int pos= random() % COUNT;
while(bitstream[pos] == 0){
pos++;
pos %= COUNT;
}
bitstream[pos]=0;
}
START_TIMER
nal_length= encode_nal(&h, nal, bitstream, COUNT, COUNT*2);
if(nal_length<0){
printf("encoding failedn");
return -1;
}
out= decode_nal(&h, nal, &out_length, &consumed, nal_length);
STOP_TIMER("NAL")
if(out_length != COUNT){
printf("incorrect length %d %dn", out_length, COUNT);
return -1;
}
if(consumed != nal_length){
printf("incorrect consumed length %d %dn", nal_length, consumed);
return -1;
}
if(memcmp(bitstream, out, COUNT)){
printf("mismatchn");
return -1;
}
}
#endif
printf("Testing RBSPn");
return 0;
}
#endif /* TEST */
static av_cold int decode_end(AVCodecContext *avctx)
{
H264Context *h = avctx->priv_data;
MpegEncContext *s = &h->s;
int i;
av_freep(&h->rbsp_buffer[0]);
av_freep(&h->rbsp_buffer[1]);
free_tables(h); //FIXME cleanup init stuff perhaps
for(i = 0; i < MAX_SPS_COUNT; i++)
av_freep(h->sps_buffers + i);
for(i = 0; i < MAX_PPS_COUNT; i++)
av_freep(h->pps_buffers + i);
MPV_common_end(s);
// memset(h, 0, sizeof(H264Context));
return 0;
}
AVCodec h264_decoder = {
"h264",
CODEC_TYPE_VIDEO,
CODEC_ID_H264,
sizeof(H264Context),
decode_init,
NULL,
decode_end,
decode_frame,
/*CODEC_CAP_DRAW_HORIZ_BAND |*/ CODEC_CAP_DR1 | CODEC_CAP_DELAY,
NULL,
flush_dpb
};
#ifdef CONFIG_H264_VDPAU_DECODER
AVCodec h264_vdpau_decoder = {
"h264_vdpau",
CODEC_TYPE_VIDEO,
CODEC_ID_H264_VDPAU,
sizeof(H264Context),
decode_init,
NULL,
decode_end,
decode_frame,
CODEC_CAP_DR1 | CODEC_CAP_DELAY | CODEC_CAP_HWACCEL_VDPAU,
.flush= flush_dpb,
.long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (VDPAU acceleration)"),
};
#endif
#ifdef CONFIG_SVQ3_DECODER
#include "svq3.c"
#endif
///
///////////////////////////////////////////////////////////////////////
static void h264_mmx_init()
{
static int inited = 0;
if (inited != 0)
return;
inited = 1;
__asm
{
push eax
push ebx
push ecx
push edx
mov eax,1
cpuid
test edx,00800000h
jz ending_mmx
mov mmx_detect,1
ending_mmx:
pop edx
pop ecx
pop ebx
pop eax
}
}
static int decoder_create(void* param)
{
h264_dec_create_t* dec_param;
FFDECODER* dec_handle;
AVCodec* codec;
H264Context* h;
MpegEncContext* s;
dec_param = (h264_dec_create_t*) param;
dec_handle = (FFDECODER*) av_malloc(sizeof(FFDECODER));
codec = &h264_decoder;
h264_mmx_init();
init_dither_tab();
avcodec_init();
dec_handle->context = avcodec_alloc_context();
dec_handle->picture = avcodec_alloc_frame();
//if(codec->capabilities&CODEC_CAP_TRUNCATED) //file using, network no use
// c->flags|= CODEC_FLAG_TRUNCATED;
if (avcodec_open(dec_handle->context, codec) < 0)
return -1;
h = dec_handle->context->priv_data;
s = &h->s;
s->dsp.idct_permutation_type =1;
dsputil_init(&s->dsp, dec_handle->context);
dec_param->handle = dec_handle;
return 0;
}
static int decoder_decode(void* handle, void* param)
{ int consumed_bytes;
int got_picture;
H264Context* h;
unsigned char* in_buf = NULL;
FFDECODER* dec_handle = (FFDECODER*) handle;
h264_dec_frame_t* frame = (h264_dec_frame_t*) param;
in_buf = (unsigned char*)frame->bitstream;
consumed_bytes = avcodec_decode_video(dec_handle->context, dec_handle->picture, &got_picture, in_buf, frame->size);
if(got_picture > 0)
{
frame->width = dec_handle->context->width;
frame->height = dec_handle->context->height;
frame->strider_Y = dec_handle->picture->linesize[0];
frame->strider_U = dec_handle->picture->linesize[1];
frame->strider_V = dec_handle->picture->linesize[2];
frame->output_Y = dec_handle->picture->data[0];
frame->output_U = dec_handle->picture->data[1];
frame->output_V = dec_handle->picture->data[2];
if( mmx_detect )
{
YUV420_TO_ARGB32_MMX( dec_handle->picture->data[0],
dec_handle->picture->linesize[0],
dec_handle->picture->data[1],
dec_handle->picture->data[2],
dec_handle->picture->linesize[1],
dec_handle->picture->linesize[2],
(unsigned char*)frame->output_ARGB,
frame->width,
frame->height,
frame->width*4 );
}
else
{
YUV420ToARGB32( dec_handle->picture->data[0],
dec_handle->picture->data[1],
dec_handle->picture->data[2],
dec_handle->picture->linesize[0],
dec_handle->picture->linesize[1],
dec_handle->picture->linesize[2],
(unsigned char*)frame->output_ARGB,
frame->width,
frame->height,
frame->width*4 );
}
h = dec_handle->context->priv_data;
switch( h->slice_type )
{
case FF_I_TYPE:
case FF_SI_TYPE:
frame->type = 1;
break;
case FF_P_TYPE:
case FF_SP_TYPE:
frame->type = 2;
break;
case FF_B_TYPE:
frame->type = 3;
break;
default:
frame->type = 4;
break;
}
}
else
{
if( consumed_bytes >= 0 )
frame->type = 0;
else
frame->type = -1;
}
return consumed_bytes; }
static int decoder_destroy(void* handle)
{
FFDECODER* dec_handle = (FFDECODER*) handle;
//H264Context* h = dec_handle->context->priv_data;
//decode_destory_vlc(h);
if(dec_handle->context)
{
avcodec_close(dec_handle->context);
av_free(dec_handle->context);
dec_handle->context = NULL;
}
if(dec_handle->picture)
{
av_free(dec_handle->picture);
dec_handle->picture = NULL;
}
if( dec_handle )
{
av_free(dec_handle);
dec_handle = NULL;
}
return 0;
}
int h264_decore(int opt, void* handle, void* param)
{
switch( opt )
{
case H264_DEC_CREATE:
return decoder_create(param);
case H264_DEC_DECODE:
return decoder_decode(handle, param);
case H264_DEC_DESTROY:
return decoder_destroy(handle);
}
return -1;
}