static int encode_superframe(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
WMACodecContext *s = avctx->priv_data;
int i, total_gain, ret, error;
s->block_len_bits = s->frame_len_bits; // required by non variable block len
s->block_len = 1 << s->block_len_bits;
ret = apply_window_and_mdct(avctx, frame);
if (ret < 0)
return ret;
if (s->ms_stereo) {
float a, b;
int i;
for (i = 0; i < s->block_len; i++) {
a = s->coefs[0][i] * 0.5;
b = s->coefs[1][i] * 0.5;
s->coefs[0][i] = a + b;
s->coefs[1][i] = a - b;
}
}
if ((ret = ff_alloc_packet2(avctx, avpkt, 2 * MAX_CODED_SUPERFRAME_SIZE, 0)) < 0)
return ret;
total_gain = 128;
for (i = 64; i; i >>= 1) {
error = encode_frame(s, s->coefs, avpkt->data, avpkt->size,
total_gain - i);
if (error <= 0)
total_gain -= i;
}
while(total_gain <= 128 && error > 0)
error = encode_frame(s, s->coefs, avpkt->data, avpkt->size, total_gain++);
if (error > 0) {
av_log(avctx, AV_LOG_ERROR, "Invalid input data or requested bitrate too low, cannot encode\n");
avpkt->size = 0;
return AVERROR(EINVAL);
}
av_assert0((put_bits_count(&s->pb) & 7) == 0);
i= avctx->block_align - (put_bits_count(&s->pb)+7)/8;
av_assert0(i>=0);
while(i--)
put_bits(&s->pb, 8, 'N');
flush_put_bits(&s->pb);
av_assert0(put_bits_ptr(&s->pb) - s->pb.buf == avctx->block_align);
if (frame->pts != AV_NOPTS_VALUE)
avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
avpkt->size = avctx->block_align;
*got_packet_ptr = 1;
return 0;
}
int ff_copy_pce_data(PutBitContext *pb, GetBitContext *gb)
{
int five_bit_ch, four_bit_ch, comment_size, bits;
int offset = put_bits_count(pb);
copy_bits(pb, gb, 10); //Tag, Object Type, Frequency
five_bit_ch = copy_bits(pb, gb, 4); //Front
five_bit_ch += copy_bits(pb, gb, 4); //Side
five_bit_ch += copy_bits(pb, gb, 4); //Back
four_bit_ch = copy_bits(pb, gb, 2); //LFE
four_bit_ch += copy_bits(pb, gb, 3); //Data
five_bit_ch += copy_bits(pb, gb, 4); //Coupling
if (copy_bits(pb, gb, 1)) //Mono Mixdown
copy_bits(pb, gb, 4);
if (copy_bits(pb, gb, 1)) //Stereo Mixdown
copy_bits(pb, gb, 4);
if (copy_bits(pb, gb, 1)) //Matrix Mixdown
copy_bits(pb, gb, 3);
for (bits = five_bit_ch*5+four_bit_ch*4; bits > 16; bits -= 16)
copy_bits(pb, gb, 16);
if (bits)
copy_bits(pb, gb, bits);
align_put_bits(pb);
align_get_bits(gb);
comment_size = copy_bits(pb, gb, 8);
for (; comment_size > 0; comment_size--)
copy_bits(pb, gb, 8);
return put_bits_count(pb) - offset;
}
static int encode_superframe(AVCodecContext *avctx,
unsigned char *buf, int buf_size, void *data){
WMACodecContext *s = avctx->priv_data;
short *samples = data;
int i, total_gain;
s->block_len_bits= s->frame_len_bits; //required by non variable block len
s->block_len = 1 << s->block_len_bits;
apply_window_and_mdct(avctx, samples, avctx->frame_size);
if (s->ms_stereo) {
float a, b;
int i;
for(i = 0; i < s->block_len; i++) {
a = s->coefs[0][i]*0.5;
b = s->coefs[1][i]*0.5;
s->coefs[0][i] = a + b;
s->coefs[1][i] = a - b;
}
}
#if 1
total_gain= 128;
for(i=64; i; i>>=1){
int error= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
if(error<0)
total_gain-= i;
}
#else
total_gain= 90;
best= encode_frame(s, s->coefs, buf, buf_size, total_gain);
for(i=32; i; i>>=1){
int scoreL= encode_frame(s, s->coefs, buf, buf_size, total_gain-i);
int scoreR= encode_frame(s, s->coefs, buf, buf_size, total_gain+i);
av_log(NULL, AV_LOG_ERROR, "%d %d %d (%d)\n", scoreL, best, scoreR, total_gain);
if(scoreL < FFMIN(best, scoreR)){
best = scoreL;
total_gain -= i;
}else if(scoreR < best){
best = scoreR;
total_gain += i;
}
}
#endif
encode_frame(s, s->coefs, buf, buf_size, total_gain);
assert((put_bits_count(&s->pb) & 7) == 0);
i= s->block_align - (put_bits_count(&s->pb)+7)/8;
assert(i>=0);
while(i--)
put_bits(&s->pb, 8, 'N');
flush_put_bits(&s->pb);
return put_bits_ptr(&s->pb) - s->pb.buf;
}
static int cbs_jpeg_write_unit(CodedBitstreamContext *ctx,
CodedBitstreamUnit *unit)
{
CodedBitstreamJPEGContext *priv = ctx->priv_data;
PutBitContext pbc;
int err;
if (!priv->write_buffer) {
// Initial write buffer size is 1MB.
priv->write_buffer_size = 1024 * 1024;
reallocate_and_try_again:
err = av_reallocp(&priv->write_buffer, priv->write_buffer_size);
if (err < 0) {
av_log(ctx->log_ctx, AV_LOG_ERROR, "Unable to allocate a "
"sufficiently large write buffer (last attempt "
"%"SIZE_SPECIFIER" bytes).\n", priv->write_buffer_size);
return err;
}
}
init_put_bits(&pbc, priv->write_buffer, priv->write_buffer_size);
if (unit->type == JPEG_MARKER_SOS)
err = cbs_jpeg_write_scan(ctx, unit, &pbc);
else
err = cbs_jpeg_write_segment(ctx, unit, &pbc);
if (err == AVERROR(ENOSPC)) {
// Overflow.
priv->write_buffer_size *= 2;
goto reallocate_and_try_again;
}
if (err < 0) {
// Write failed for some other reason.
return err;
}
if (put_bits_count(&pbc) % 8)
unit->data_bit_padding = 8 - put_bits_count(&pbc) % 8;
else
unit->data_bit_padding = 0;
unit->data_size = (put_bits_count(&pbc) + 7) / 8;
flush_put_bits(&pbc);
err = ff_cbs_alloc_unit_data(ctx, unit, unit->data_size);
if (err < 0)
return err;
memcpy(unit->data, priv->write_buffer, unit->data_size);
return 0;
}
static void vaapi_encode_h264_write_sei(PutBitContext *pbc,
VAAPIEncodeContext *ctx,
VAAPIEncodePicture *pic)
{
VAAPIEncodeH264Context *priv = ctx->priv_data;
PutBitContext payload_bits;
char payload[256];
int payload_type, payload_size, i;
void (*write_payload)(PutBitContext *pbc,
VAAPIEncodeContext *ctx,
VAAPIEncodePicture *pic) = NULL;
vaapi_encode_h264_write_nal_header(pbc, NAL_SEI, 0);
for (payload_type = 0; payload_type < 64; payload_type++) {
switch (payload_type) {
case SEI_TYPE_BUFFERING_PERIOD:
if (!priv->send_timing_sei ||
pic->type != PICTURE_TYPE_IDR)
continue;
write_payload = &vaapi_encode_h264_write_buffering_period;
break;
case SEI_TYPE_PIC_TIMING:
if (!priv->send_timing_sei)
continue;
write_payload = &vaapi_encode_h264_write_pic_timing;
break;
case SEI_TYPE_USER_DATA_UNREGISTERED:
if (pic->encode_order != 0)
continue;
write_payload = &vaapi_encode_h264_write_identifier;
break;
default:
continue;
}
init_put_bits(&payload_bits, payload, sizeof(payload));
write_payload(&payload_bits, ctx, pic);
if (put_bits_count(&payload_bits) & 7) {
write_u(&payload_bits, 1, 1, bit_equal_to_one);
while (put_bits_count(&payload_bits) & 7)
write_u(&payload_bits, 1, 0, bit_equal_to_zero);
}
payload_size = put_bits_count(&payload_bits) / 8;
flush_put_bits(&payload_bits);
u(8, payload_type, last_payload_type_byte);
u(8, payload_size, last_payload_size_byte);
for (i = 0; i < payload_size; i++)
u(8, payload[i] & 0xff, sei_payload);
}
vaapi_encode_h264_write_trailing_rbsp(pbc);
}
static int encode_frame(AVCodecContext *avctx, unsigned char *buf, int buf_size, void *data){
ASV1Context * const a = avctx->priv_data;
AVFrame *pict = data;
AVFrame * const p= (AVFrame*)&a->picture;
int size;
int mb_x, mb_y;
init_put_bits(&a->pb, buf, buf_size);
*p = *pict;
p->pict_type= I_TYPE;
p->key_frame= 1;
for(mb_y=0; mb_y<a->mb_height2; mb_y++){
for(mb_x=0; mb_x<a->mb_width2; mb_x++){
dct_get(a, mb_x, mb_y);
encode_mb(a, a->block);
}
}
if(a->mb_width2 != a->mb_width){
mb_x= a->mb_width2;
for(mb_y=0; mb_y<a->mb_height2; mb_y++){
dct_get(a, mb_x, mb_y);
encode_mb(a, a->block);
}
}
if(a->mb_height2 != a->mb_height){
mb_y= a->mb_height2;
for(mb_x=0; mb_x<a->mb_width; mb_x++){
dct_get(a, mb_x, mb_y);
encode_mb(a, a->block);
}
}
emms_c();
align_put_bits(&a->pb);
while(put_bits_count(&a->pb)&31)
put_bits(&a->pb, 8, 0);
size= put_bits_count(&a->pb)/32;
if(avctx->codec_id == CODEC_ID_ASV1)
a->dsp.bswap_buf((uint32_t*)buf, (uint32_t*)buf, size);
else{
int i;
for(i=0; i<4*size; i++)
buf[i]= reverse[ buf[i] ];
}
return size*4;
}
static int append_picture_header(const MPEG4PictureInfo *pic_info)
{
MPEG4SliceInfo slice_info;
uint8_t buf[32], buf_size;
const uint8_t *slice_data;
unsigned int slice_data_size;
PutBitContext pb;
int r;
if (mpeg4_get_slice_info(0, &slice_info) < 0)
return -1;
int time_incr = 1 + ilog2(pic_info->vop_time_increment_resolution - 1);
if (time_incr < 1)
time_incr = 1;
/* Reconstruct the VOP header */
init_put_bits(&pb, buf, sizeof(buf));
put_bits(&pb, 16, 0); /* vop header */
put_bits(&pb, 16, VOP_STARTCODE); /* vop header */
put_bits(&pb, 2, pic_info->vop_fields.bits.vop_coding_type);
put_bits(&pb, 1, 0);
put_bits(&pb, 1, 1); /* marker */
put_bits(&pb, time_incr, 0); /* time increment */
put_bits(&pb, 1, 1); /* marker */
put_bits(&pb, 1, 1); /* vop coded */
if (pic_info->vop_fields.bits.vop_coding_type == VOP_P_TYPE)
put_bits(&pb, 1, pic_info->vop_fields.bits.vop_rounding_type);
put_bits(&pb, 3, pic_info->vop_fields.bits.intra_dc_vlc_thr);
if (pic_info->vol_fields.bits.interlaced) {
put_bits(&pb, 1, pic_info->vop_fields.bits.top_field_first);
put_bits(&pb, 1, pic_info->vop_fields.bits.alternate_vertical_scan_flag);
}
put_bits(&pb, pic_info->quant_precision, slice_info.quant_scale);
if (pic_info->vop_fields.bits.vop_coding_type != VOP_I_TYPE)
put_bits(&pb, 3, pic_info->vop_fcode_forward);
if (pic_info->vop_fields.bits.vop_coding_type == VOP_B_TYPE)
put_bits(&pb, 3, pic_info->vop_fcode_backward);
/* Merge in bits from the first byte of the slice */
if ((put_bits_count(&pb) % 8) != slice_info.macroblock_offset)
return -1;
if (mpeg4_get_slice_data(0, &slice_data, &slice_data_size) < 0)
return -1;
r = 8 - (put_bits_count(&pb) % 8);
put_bits(&pb, r, slice_data[0] & ((1U << r) - 1));
flush_put_bits(&pb);
buf_size = put_bits_count(&pb) / 8;
return vdpau_append_slice_data(buf, buf_size);
}
int avpriv_dca_convert_bitstream(const uint8_t *src, int src_size, uint8_t *dst,
int max_size)
{
uint32_t mrk;
int i, tmp;
PutBitContext pb;
if ((unsigned) src_size > (unsigned) max_size)
src_size = max_size;
mrk = AV_RB32(src);
switch (mrk) {
case DCA_SYNCWORD_CORE_BE:
memcpy(dst, src, src_size);
return src_size;
case DCA_SYNCWORD_CORE_LE:
for (i = 0; i < (src_size + 1) >> 1; i++) {
AV_WB16(dst, AV_RL16(src));
src += 2;
dst += 2;
}
return src_size;
case DCA_SYNCWORD_CORE_14B_BE:
case DCA_SYNCWORD_CORE_14B_LE:
init_put_bits(&pb, dst, max_size);
for (i = 0; i < (src_size + 1) >> 1; i++, src += 2) {
tmp = ((mrk == DCA_SYNCWORD_CORE_14B_BE) ? AV_RB16(src) : AV_RL16(src)) & 0x3FFF;
put_bits(&pb, 14, tmp);
}
flush_put_bits(&pb);
return (put_bits_count(&pb) + 7) >> 3;
default:
return AVERROR_INVALIDDATA;
}
}
size_t xau_pcm8m_encode( const sint8* pcm, size_t samples, void* buf, size_t bufsize )
{
// check preconditions
xau_codec_t codec;
init_put_bits( &codec.pb, (uint8*)buf, bufsize );
rice_init( codec.rice, PCM8M_K1, PCM8M_K2 );
ppfilter_init( &codec.flt, PCM8M_SH );
const sint8* p = pcm;
for( size_t n = 0; n != samples; ++n ) {
// a) skip sample diffing part as it seems hurting performance a lot on 8bit mode
// *p -= prev / 2
// b) apply polyphase filter
sint32 tmp = ppfilter_encode( &codec.flt, *p++ );
// c) skip median predictor part - also hurts performance on 8bps
// *p -= PREDICTOR1( last, 4 );
// d) encode the 'unsignificated' code
rice_enc( &codec.pb, codec.rice, xbs_sign_pack( tmp ) );
}
align_put_bits( &codec.pb );
uint32 written = put_bits_count( &codec.pb ) / 8;
// check for overruns
return written;
}
static int cbs_write_ue_golomb(CodedBitstreamContext *ctx, PutBitContext *pbc,
const char *name, uint32_t value,
uint32_t range_min, uint32_t range_max)
{
int len;
if (value < range_min || value > range_max) {
av_log(ctx->log_ctx, AV_LOG_ERROR, "%s out of range: "
"%"PRIu32", but must be in [%"PRIu32",%"PRIu32"].\n",
name, value, range_min, range_max);
return AVERROR_INVALIDDATA;
}
av_assert0(value != UINT32_MAX);
len = av_log2(value + 1);
if (put_bits_left(pbc) < 2 * len + 1)
return AVERROR(ENOSPC);
if (ctx->trace_enable) {
char bits[65];
int i;
for (i = 0; i < len; i++)
bits[i] = '0';
bits[len] = '1';
for (i = 0; i < len; i++)
bits[len + i + 1] = (value + 1) >> (len - i - 1) & 1 ? '1' : '0';
bits[len + len + 1] = 0;
ff_cbs_trace_syntax_element(ctx, put_bits_count(pbc), name, bits, value);
}
static int vaapi_encode_h264_write_extra_header(AVCodecContext *avctx,
VAAPIEncodePicture *pic,
int index, int *type,
char *data, size_t *data_len)
{
VAAPIEncodeContext *ctx = avctx->priv_data;
PutBitContext pbc;
char tmp[256];
size_t header_len;
if (index == 0 && ctx->va_rc_mode == VA_RC_CBR) {
*type = VAEncPackedHeaderH264_SEI;
init_put_bits(&pbc, tmp, sizeof(tmp));
vaapi_encode_h264_write_sei(&pbc, ctx, pic);
header_len = put_bits_count(&pbc);
flush_put_bits(&pbc);
return ff_vaapi_encode_h26x_nal_unit_to_byte_stream(data, data_len,
tmp, header_len);
} else {
return AVERROR_EOF;
}
}
static int encode_frame(AVCodecContext *avctx, unsigned char *buf,
int buf_size, void *data)
{
PutBitContext pb;
AVFrame *p = data;
int x, y;
p->pict_type = AV_PICTURE_TYPE_I;
p->key_frame = 1;
init_put_bits(&pb, buf, buf_size / 8);
for (y = 0; y < avctx->height; y++) {
uint8_t *luma = &p->data[0][y * p->linesize[0]];
uint8_t *cb = &p->data[1][y * p->linesize[1]];
uint8_t *cr = &p->data[2][y * p->linesize[2]];
for (x = 0; x < avctx->width; x += 4) {
put_bits(&pb, 5, luma[3] >> 3);
put_bits(&pb, 5, luma[2] >> 3);
put_bits(&pb, 5, luma[1] >> 3);
put_bits(&pb, 5, luma[0] >> 3);
luma += 4;
put_bits(&pb, 6, *(cb++) >> 2);
put_bits(&pb, 6, *(cr++) >> 2);
}
}
flush_put_bits(&pb);
return put_bits_count(&pb) / 8;
}
int ff_dca_convert_bitstream(const uint8_t *src, int src_size, uint8_t *dst,
int max_size)
{
uint32_t mrk;
int i, tmp;
const uint16_t *ssrc = (const uint16_t *) src;
uint16_t *sdst = (uint16_t *) dst;
PutBitContext pb;
if ((unsigned) src_size > (unsigned) max_size)
src_size = max_size;
mrk = AV_RB32(src);
switch (mrk) {
case DCA_MARKER_RAW_BE:
memcpy(dst, src, src_size);
return src_size;
case DCA_MARKER_RAW_LE:
for (i = 0; i < (src_size + 1) >> 1; i++)
*sdst++ = av_bswap16(*ssrc++);
return src_size;
case DCA_MARKER_14B_BE:
case DCA_MARKER_14B_LE:
init_put_bits(&pb, dst, max_size);
for (i = 0; i < (src_size + 1) >> 1; i++, src += 2) {
tmp = ((mrk == DCA_MARKER_14B_BE) ? AV_RB16(src) : AV_RL16(src)) & 0x3FFF;
put_bits(&pb, 14, tmp);
}
flush_put_bits(&pb);
return (put_bits_count(&pb) + 7) >> 3;
default:
return AVERROR_INVALIDDATA;
}
}
static int latm_read_audio_specific_config(GetBitContext *gb,
PutBitContext *pb)
{
int num_bits=0;
int audio_object_type;
MPEG4AudioConfig b, *c;
c = &b;
c->sbr = -1;
audio_object_type = copy_bits(pb, gb, 5);
if (audio_object_type == AOT_ESCAPE) {
audio_object_type = AOT_ESCAPE + copy_bits(pb, gb, 6) + 1;
}
c->object_type = audio_object_type;
c->sampling_index = copy_bits(pb, gb, 4);
c->sample_rate = ff_mpeg4audio_sample_rates[c->sampling_index];
if (c->sampling_index == 0x0f) {
c->sample_rate = copy_bits(pb, gb, 24);
}
c->chan_config = copy_bits(pb, gb, 4);
if (c->chan_config < FF_ARRAY_ELEMS(ff_mpeg4audio_channels))
c->channels = ff_mpeg4audio_channels[c->chan_config];
if (audio_object_type == AOT_AAC_MAIN ||
audio_object_type == AOT_AAC_LC ||
audio_object_type == AOT_AAC_SSR ||
audio_object_type == AOT_AAC_LTP ||
audio_object_type == AOT_AAC_SCALABLE ||
audio_object_type == AOT_TWINVQ) {
latm_read_ga_specific_config(audio_object_type, c, gb, pb);
} else if (audio_object_type == AOT_SBR) {
c->sbr = 1;
c->ext_sampling_index = copy_bits(pb, gb, 4);
c->ext_sample_rate = ff_mpeg4audio_sample_rates[c->ext_sampling_index];
if (c->ext_sampling_index == 0x0f) {
c->ext_sample_rate = copy_bits(pb, gb, 24);
}
c->object_type = copy_bits(pb, gb, 5);
} else if (audio_object_type >= AOT_ER_AAC_LC) {
latm_read_ga_specific_config(audio_object_type, c, gb, pb);
copy_bits(pb, gb, 2); // epConfig
}
if (c->sbr == -1 && c->sample_rate <= 24000)
c->sample_rate *= 2;
// count the extradata
num_bits = put_bits_count(pb);
flush_put_bits(pb);
return num_bits;
}
void ff_copy_bits(PutBitContext *pb, const uint8_t *src, int length)
{
int words= length>>4;
int bits= length&15;
int i;
if(length==0) return;
if(words < 16 || put_bits_count(pb)&7){
for(i=0; i<words; i++) put_bits(pb, 16, AV_RB16(src + 2*i));
}else{
for(i=0; put_bits_count(pb)&31; i++)
put_bits(pb, 8, src[i]);
flush_put_bits(pb);
memcpy(put_bits_ptr(pb), src+i, 2*words-i);
skip_put_bytes(pb, 2*words-i);
}
put_bits(pb, bits, AV_RB16(src + 2*words)>>(16-bits));
}
static int vaapi_encode_h264_write_sequence_header(AVCodecContext *avctx,
char *data, size_t *data_len)
{
VAAPIEncodeContext *ctx = avctx->priv_data;
PutBitContext pbc;
char tmp[256];
int err;
size_t nal_len, bit_len, bit_pos, next_len;
bit_len = *data_len;
bit_pos = 0;
init_put_bits(&pbc, tmp, sizeof(tmp));
vaapi_encode_h264_write_sps(&pbc, ctx);
nal_len = put_bits_count(&pbc);
flush_put_bits(&pbc);
next_len = bit_len - bit_pos;
err = ff_vaapi_encode_h26x_nal_unit_to_byte_stream(data + bit_pos / 8,
&next_len,
tmp, nal_len);
if (err < 0)
return err;
bit_pos += next_len;
init_put_bits(&pbc, tmp, sizeof(tmp));
vaapi_encode_h264_write_pps(&pbc, ctx);
nal_len = put_bits_count(&pbc);
flush_put_bits(&pbc);
next_len = bit_len - bit_pos;
err = ff_vaapi_encode_h26x_nal_unit_to_byte_stream(data + bit_pos / 8,
&next_len,
tmp, nal_len);
if (err < 0)
return err;
bit_pos += next_len;
*data_len = bit_pos;
return 0;
}
static int encode_frame(WMACodecContext *s, float (*src_coefs)[BLOCK_MAX_SIZE],
uint8_t *buf, int buf_size, int total_gain)
{
init_put_bits(&s->pb, buf, buf_size);
if (s->use_bit_reservoir)
av_assert0(0); // FIXME not implemented
else if (encode_block(s, src_coefs, total_gain) < 0)
return INT_MAX;
avpriv_align_put_bits(&s->pb);
return put_bits_count(&s->pb) / 8 - s->avctx->block_align;
}
/**
* Write some auxiliary information about the created AAC file.
*/
static void put_bitstream_info(AACEncContext *s, const char *name)
{
int i, namelen, padbits;
namelen = strlen(name) + 2;
put_bits(&s->pb, 3, TYPE_FIL);
put_bits(&s->pb, 4, FFMIN(namelen, 15));
if (namelen >= 15)
put_bits(&s->pb, 8, namelen - 14);
put_bits(&s->pb, 4, 0); //extension type - filler
padbits = -put_bits_count(&s->pb) & 7;
avpriv_align_put_bits(&s->pb);
for (i = 0; i < namelen - 2; i++)
put_bits(&s->pb, 8, name[i]);
put_bits(&s->pb, 12 - padbits, 0);
}
static int vaapi_encode_h264_write_slice_header(AVCodecContext *avctx,
VAAPIEncodePicture *pic,
VAAPIEncodeSlice *slice,
char *data, size_t *data_len)
{
VAAPIEncodeContext *ctx = avctx->priv_data;
PutBitContext pbc;
char tmp[256];
size_t header_len;
init_put_bits(&pbc, tmp, sizeof(tmp));
vaapi_encode_h264_write_slice_header2(&pbc, ctx, pic, slice);
header_len = put_bits_count(&pbc);
flush_put_bits(&pbc);
return ff_vaapi_encode_h26x_nal_unit_to_byte_stream(data, data_len,
tmp, header_len);
}
static int encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
DCAEncContext *c = avctx->priv_data;
const int32_t *samples;
int ret, i;
if ((ret = ff_alloc_packet2(avctx, avpkt, c->frame_size, 0)) < 0)
return ret;
samples = (const int32_t *)frame->data[0];
subband_transform(c, samples);
if (c->lfe_channel)
lfe_downsample(c, samples);
calc_masking(c, samples);
find_peaks(c);
assign_bits(c);
calc_scales(c);
quantize_all(c);
shift_history(c, samples);
init_put_bits(&c->pb, avpkt->data, avpkt->size);
put_frame_header(c);
put_primary_audio_header(c);
for (i = 0; i < SUBFRAMES; i++)
put_subframe(c, i);
for (i = put_bits_count(&c->pb); i < 8*c->frame_size; i++)
put_bits(&c->pb, 1, 0);
flush_put_bits(&c->pb);
avpkt->pts = frame->pts;
avpkt->duration = ff_samples_to_time_base(avctx, frame->nb_samples);
avpkt->size = put_bits_count(&c->pb) >> 3;
*got_packet_ptr = 1;
return 0;
}
static int encode_frame(AVCodecContext *avctx, AVPacket *pkt,
const AVFrame *p, int *got_packet)
{
PutBitContext pb;
int x, y, ret;
if ((ret = ff_alloc_packet(pkt, 32*avctx->height*avctx->width/4)) < 0) {
av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
return ret;
}
avctx->coded_frame->pict_type = AV_PICTURE_TYPE_I;
avctx->coded_frame->key_frame = 1;
init_put_bits(&pb, pkt->data, pkt->size);
for (y = 0; y < avctx->height; y++) {
uint8_t *luma = &p->data[0][y * p->linesize[0]];
uint8_t *cb = &p->data[1][y * p->linesize[1]];
uint8_t *cr = &p->data[2][y * p->linesize[2]];
for (x = 0; x < avctx->width; x += 4) {
put_bits(&pb, 5, luma[3] >> 3);
put_bits(&pb, 5, luma[2] >> 3);
put_bits(&pb, 5, luma[1] >> 3);
put_bits(&pb, 5, luma[0] >> 3);
luma += 4;
put_bits(&pb, 6, *(cb++) >> 2);
put_bits(&pb, 6, *(cr++) >> 2);
}
}
flush_put_bits(&pb);
pkt->size = put_bits_count(&pb) / 8;
pkt->flags |= AV_PKT_FLAG_KEY;
*got_packet = 1;
return 0;
}
/**
* Convert bitstream to one representation based on sync marker
*/
static int dca_convert_bitstream(const uint8_t * src, int src_size, uint8_t * dst,
int max_size)
{
uint32_t mrk;
int i, tmp;
const uint16_t *ssrc = (const uint16_t *) src;
uint16_t *sdst = (uint16_t *) dst;
PutBitContext pb;
if((unsigned)src_size > (unsigned)max_size) {
// av_log(NULL, AV_LOG_ERROR, "Input frame size larger then DCA_MAX_FRAME_SIZE!\n");
// return -1;
src_size = max_size;
}
mrk = AV_RB32(src);
switch (mrk) {
case DCA_MARKER_RAW_BE:
memcpy(dst, src, src_size);
return src_size;
case DCA_MARKER_RAW_LE:
for (i = 0; i < (src_size + 1) >> 1; i++)
*sdst++ = bswap_16(*ssrc++);
return src_size;
case DCA_MARKER_14B_BE:
case DCA_MARKER_14B_LE:
init_put_bits(&pb, dst, max_size);
for (i = 0; i < (src_size + 1) >> 1; i++, src += 2) {
tmp = ((mrk == DCA_MARKER_14B_BE) ? AV_RB16(src) : AV_RL16(src)) & 0x3FFF;
put_bits(&pb, 14, tmp);
}
flush_put_bits(&pb);
return (put_bits_count(&pb) + 7) >> 3;
default:
return -1;
}
}
/**
*
* @return the number of bytes written
*/
static int put_cabac_terminate(CABACContext *c, int bit) {
c->range -= 2;
if(!bit) {
renorm_cabac_encoder(c);
} else {
c->low += c->range;
c->range= 2;
renorm_cabac_encoder(c);
assert(c->low <= 0x1FF);
put_cabac_bit(c, c->low>>9);
put_bits(&c->pb, 2, ((c->low>>7)&3)|1);
flush_put_bits(&c->pb); //FIXME FIXME FIXME XXX wrong
}
#ifdef STRICT_LIMITS
c->symCount++;
#endif
return (put_bits_count(&c->pb)+7)>>3;
}
static int jpeg_create_header(uint8_t *buf, int size, uint32_t type, uint32_t w,
uint32_t h, const uint8_t *qtable, int nb_qtable)
{
PutBitContext pbc;
init_put_bits(&pbc, buf, size);
/* Convert from blocks to pixels. */
w <<= 3;
h <<= 3;
/* SOI */
put_marker(&pbc, SOI);
/* JFIF header */
put_marker(&pbc, APP0);
put_bits(&pbc, 16, 16);
avpriv_put_string(&pbc, "JFIF", 1);
put_bits(&pbc, 16, 0x0201);
put_bits(&pbc, 8, 0);
put_bits(&pbc, 16, 1);
put_bits(&pbc, 16, 1);
put_bits(&pbc, 8, 0);
put_bits(&pbc, 8, 0);
/* DQT */
put_marker(&pbc, DQT);
if (nb_qtable == 2) {
put_bits(&pbc, 16, 2 + 2 * (1 + 64));
} else {
put_bits(&pbc, 16, 2 + 1 * (1 + 64));
}
put_bits(&pbc, 8, 0);
/* Each table is an array of 64 values given in zig-zag
* order, identical to the format used in a JFIF DQT
* marker segment. */
avpriv_copy_bits(&pbc, qtable, 64 * 8);
if (nb_qtable == 2) {
put_bits(&pbc, 8, 1);
avpriv_copy_bits(&pbc, qtable + 64, 64 * 8);
}
/* DHT */
put_marker(&pbc, DHT);
jpeg_create_huffman_table(&pbc, 0, 0, avpriv_mjpeg_bits_dc_luminance,
avpriv_mjpeg_val_dc);
jpeg_create_huffman_table(&pbc, 0, 1, avpriv_mjpeg_bits_dc_chrominance,
avpriv_mjpeg_val_dc);
jpeg_create_huffman_table(&pbc, 1, 0, avpriv_mjpeg_bits_ac_luminance,
avpriv_mjpeg_val_ac_luminance);
jpeg_create_huffman_table(&pbc, 1, 1, avpriv_mjpeg_bits_ac_chrominance,
avpriv_mjpeg_val_ac_chrominance);
/* SOF0 */
put_marker(&pbc, SOF0);
put_bits(&pbc, 16, 17);
put_bits(&pbc, 8, 8);
put_bits(&pbc, 8, h >> 8);
put_bits(&pbc, 8, h);
put_bits(&pbc, 8, w >> 8);
put_bits(&pbc, 8, w);
put_bits(&pbc, 8, 3);
put_bits(&pbc, 8, 1);
put_bits(&pbc, 8, type ? 34 : 33);
put_bits(&pbc, 8, 0);
put_bits(&pbc, 8, 2);
put_bits(&pbc, 8, 17);
put_bits(&pbc, 8, nb_qtable == 2 ? 1 : 0);
put_bits(&pbc, 8, 3);
put_bits(&pbc, 8, 17);
put_bits(&pbc, 8, nb_qtable == 2 ? 1 : 0);
/* SOS */
put_marker(&pbc, SOS);
put_bits(&pbc, 16, 12);
put_bits(&pbc, 8, 3);
put_bits(&pbc, 8, 1);
put_bits(&pbc, 8, 0);
put_bits(&pbc, 8, 2);
put_bits(&pbc, 8, 17);
put_bits(&pbc, 8, 3);
put_bits(&pbc, 8, 17);
put_bits(&pbc, 8, 0);
put_bits(&pbc, 8, 63);
put_bits(&pbc, 8, 0);
/* Fill the buffer. */
flush_put_bits(&pbc);
/* Return the length in bytes of the JPEG header. */
return put_bits_count(&pbc) / 8;
}
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
AACEncContext *s = avctx->priv_data;
float **samples = s->planar_samples, *samples2, *la, *overlap;
ChannelElement *cpe;
SingleChannelElement *sce;
IndividualChannelStream *ics;
int i, its, ch, w, chans, tag, start_ch, ret, frame_bits;
int target_bits, rate_bits, too_many_bits, too_few_bits;
int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
int chan_el_counter[4];
FFPsyWindowInfo windows[AAC_MAX_CHANNELS];
if (s->last_frame == 2)
return 0;
/* add current frame to queue */
if (frame) {
if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
return ret;
}
copy_input_samples(s, frame);
if (s->psypp)
ff_psy_preprocess(s->psypp, s->planar_samples, s->channels);
if (!avctx->frame_number)
return 0;
start_ch = 0;
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
for (ch = 0; ch < chans; ch++) {
float clip_avoidance_factor;
sce = &cpe->ch[ch];
ics = &sce->ics;
s->cur_channel = start_ch + ch;
overlap = &samples[s->cur_channel][0];
samples2 = overlap + 1024;
la = samples2 + (448+64);
if (!frame)
la = NULL;
if (tag == TYPE_LFE) {
wi[ch].window_type[0] = ONLY_LONG_SEQUENCE;
wi[ch].window_shape = 0;
wi[ch].num_windows = 1;
wi[ch].grouping[0] = 1;
/* Only the lowest 12 coefficients are used in a LFE channel.
* The expression below results in only the bottom 8 coefficients
* being used for 11.025kHz to 16kHz sample rates.
*/
ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
} else {
wi[ch] = s->psy.model->window(&s->psy, samples2, la, s->cur_channel,
ics->window_sequence[0]);
}
ics->window_sequence[1] = ics->window_sequence[0];
ics->window_sequence[0] = wi[ch].window_type[0];
ics->use_kb_window[1] = ics->use_kb_window[0];
ics->use_kb_window[0] = wi[ch].window_shape;
ics->num_windows = wi[ch].num_windows;
ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
ics->max_sfb = FFMIN(ics->max_sfb, ics->num_swb);
ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
ff_swb_offset_128 [s->samplerate_index]:
ff_swb_offset_1024[s->samplerate_index];
ics->tns_max_bands = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
ff_tns_max_bands_128 [s->samplerate_index]:
ff_tns_max_bands_1024[s->samplerate_index];
clip_avoidance_factor = 0.0f;
for (w = 0; w < ics->num_windows; w++)
ics->group_len[w] = wi[ch].grouping[w];
for (w = 0; w < ics->num_windows; w++) {
if (wi[ch].clipping[w] > CLIP_AVOIDANCE_FACTOR) {
ics->window_clipping[w] = 1;
clip_avoidance_factor = FFMAX(clip_avoidance_factor, wi[ch].clipping[w]);
} else {
ics->window_clipping[w] = 0;
}
}
if (clip_avoidance_factor > CLIP_AVOIDANCE_FACTOR) {
ics->clip_avoidance_factor = CLIP_AVOIDANCE_FACTOR / clip_avoidance_factor;
} else {
ics->clip_avoidance_factor = 1.0f;
}
apply_window_and_mdct(s, sce, overlap);
if (s->options.ltp && s->coder->update_ltp) {
s->coder->update_ltp(s, sce);
apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
}
if (!(isfinite(cpe->ch[ch].coeffs[ 0]) &&
isfinite(cpe->ch[ch].coeffs[ 128]) &&
isfinite(cpe->ch[ch].coeffs[2*128]) &&
isfinite(cpe->ch[ch].coeffs[3*128]) &&
isfinite(cpe->ch[ch].coeffs[4*128]) &&
isfinite(cpe->ch[ch].coeffs[5*128]) &&
isfinite(cpe->ch[ch].coeffs[6*128]) &&
isfinite(cpe->ch[ch].coeffs[7*128]))
) {
av_log(avctx, AV_LOG_ERROR, "Input contains NaN/+-Inf\n");
return AVERROR(EINVAL);
}
avoid_clipping(s, sce);
}
start_ch += chans;
}
if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels, 0)) < 0)
return ret;
frame_bits = its = 0;
do {
init_put_bits(&s->pb, avpkt->data, avpkt->size);
if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & AV_CODEC_FLAG_BITEXACT))
put_bitstream_info(s, LIBAVCODEC_IDENT);
start_ch = 0;
target_bits = 0;
memset(chan_el_counter, 0, sizeof(chan_el_counter));
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
const float *coeffs[2];
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
cpe->common_window = 0;
memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
put_bits(&s->pb, 3, tag);
put_bits(&s->pb, 4, chan_el_counter[tag]++);
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
coeffs[ch] = sce->coeffs;
sce->ics.predictor_present = 0;
sce->ics.ltp.present = 0;
memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
for (w = 0; w < 128; w++)
if (sce->band_type[w] > RESERVED_BT)
sce->band_type[w] = 0;
}
s->psy.bitres.alloc = -1;
s->psy.bitres.bits = s->last_frame_pb_count / s->channels;
s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
if (s->psy.bitres.alloc > 0) {
/* Lambda unused here on purpose, we need to take psy's unscaled allocation */
target_bits += s->psy.bitres.alloc
* (s->lambda / (avctx->global_quality ? avctx->global_quality : 120));
s->psy.bitres.alloc /= chans;
}
s->cur_type = tag;
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch + ch;
if (s->options.pns && s->coder->mark_pns)
s->coder->mark_pns(s, avctx, &cpe->ch[ch]);
s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
}
if (chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape) {
cpe->common_window = 1;
for (w = 0; w < wi[0].num_windows; w++) {
if (wi[0].grouping[w] != wi[1].grouping[w]) {
cpe->common_window = 0;
break;
}
}
}
for (ch = 0; ch < chans; ch++) { /* TNS and PNS */
sce = &cpe->ch[ch];
s->cur_channel = start_ch + ch;
if (s->options.tns && s->coder->search_for_tns)
s->coder->search_for_tns(s, sce);
if (s->options.tns && s->coder->apply_tns_filt)
s->coder->apply_tns_filt(s, sce);
if (sce->tns.present)
tns_mode = 1;
if (s->options.pns && s->coder->search_for_pns)
s->coder->search_for_pns(s, avctx, sce);
}
s->cur_channel = start_ch;
if (s->options.intensity_stereo) { /* Intensity Stereo */
if (s->coder->search_for_is)
s->coder->search_for_is(s, avctx, cpe);
if (cpe->is_mode) is_mode = 1;
apply_intensity_stereo(cpe);
}
if (s->options.pred) { /* Prediction */
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
s->cur_channel = start_ch + ch;
if (s->options.pred && s->coder->search_for_pred)
s->coder->search_for_pred(s, sce);
if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
}
if (s->coder->adjust_common_pred)
s->coder->adjust_common_pred(s, cpe);
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
s->cur_channel = start_ch + ch;
if (s->options.pred && s->coder->apply_main_pred)
s->coder->apply_main_pred(s, sce);
}
s->cur_channel = start_ch;
}
if (s->options.mid_side) { /* Mid/Side stereo */
if (s->options.mid_side == -1 && s->coder->search_for_ms)
s->coder->search_for_ms(s, cpe);
else if (cpe->common_window)
memset(cpe->ms_mask, 1, sizeof(cpe->ms_mask));
apply_mid_side_stereo(cpe);
}
adjust_frame_information(cpe, chans);
if (s->options.ltp) { /* LTP */
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
s->cur_channel = start_ch + ch;
if (s->coder->search_for_ltp)
s->coder->search_for_ltp(s, sce, cpe->common_window);
if (sce->ics.ltp.present) pred_mode = 1;
}
s->cur_channel = start_ch;
if (s->coder->adjust_common_ltp)
s->coder->adjust_common_ltp(s, cpe);
}
if (chans == 2) {
put_bits(&s->pb, 1, cpe->common_window);
if (cpe->common_window) {
put_ics_info(s, &cpe->ch[0].ics);
if (s->coder->encode_main_pred)
s->coder->encode_main_pred(s, &cpe->ch[0]);
if (s->coder->encode_ltp_info)
s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
encode_ms_info(&s->pb, cpe);
if (cpe->ms_mode) ms_mode = 1;
}
}
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch + ch;
encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
}
start_ch += chans;
}
if (avctx->flags & CODEC_FLAG_QSCALE) {
/* When using a constant Q-scale, don't mess with lambda */
break;
}
/* rate control stuff
* allow between the nominal bitrate, and what psy's bit reservoir says to target
* but drift towards the nominal bitrate always
*/
frame_bits = put_bits_count(&s->pb);
rate_bits = avctx->bit_rate * 1024 / avctx->sample_rate;
rate_bits = FFMIN(rate_bits, 6144 * s->channels - 3);
too_many_bits = FFMAX(target_bits, rate_bits);
too_many_bits = FFMIN(too_many_bits, 6144 * s->channels - 3);
too_few_bits = FFMIN(FFMAX(rate_bits - rate_bits/4, target_bits), too_many_bits);
/* When using ABR, be strict (but only for increasing) */
too_few_bits = too_few_bits - too_few_bits/8;
too_many_bits = too_many_bits + too_many_bits/2;
if ( its == 0 /* for steady-state Q-scale tracking */
|| (its < 5 && (frame_bits < too_few_bits || frame_bits > too_many_bits))
|| frame_bits >= 6144 * s->channels - 3 )
{
float ratio = ((float)rate_bits) / frame_bits;
if (frame_bits >= too_few_bits && frame_bits <= too_many_bits) {
/*
* This path is for steady-state Q-scale tracking
* When frame bits fall within the stable range, we still need to adjust
* lambda to maintain it like so in a stable fashion (large jumps in lambda
* create artifacts and should be avoided), but slowly
*/
ratio = sqrtf(sqrtf(ratio));
ratio = av_clipf(ratio, 0.9f, 1.1f);
} else {
/* Not so fast though */
ratio = sqrtf(ratio);
}
s->lambda = FFMIN(s->lambda * ratio, 65536.f);
/* Keep iterating if we must reduce and lambda is in the sky */
if (ratio > 0.9f && ratio < 1.1f) {
break;
} else {
if (is_mode || ms_mode || tns_mode || pred_mode) {
for (i = 0; i < s->chan_map[0]; i++) {
// Must restore coeffs
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
for (ch = 0; ch < chans; ch++)
memcpy(cpe->ch[ch].coeffs, cpe->ch[ch].pcoeffs, sizeof(cpe->ch[ch].coeffs));
}
}
its++;
}
} else {
break;
}
} while (1);
if (s->options.ltp && s->coder->ltp_insert_new_frame)
s->coder->ltp_insert_new_frame(s);
put_bits(&s->pb, 3, TYPE_END);
flush_put_bits(&s->pb);
s->last_frame_pb_count = put_bits_count(&s->pb);
s->lambda_sum += s->lambda;
s->lambda_count++;
if (!frame)
s->last_frame++;
ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
&avpkt->duration);
avpkt->size = put_bits_count(&s->pb) >> 3;
*got_packet_ptr = 1;
return 0;
}
static int aac_encode_frame(AVCodecContext *avctx,
uint8_t *frame, int buf_size, void *data)
{
AACEncContext *s = avctx->priv_data;
int16_t *samples = s->samples, *samples2, *la;
ChannelElement *cpe;
int i, ch, w, g, chans, tag, start_ch;
int chan_el_counter[4];
FFPsyWindowInfo windows[AAC_MAX_CHANNELS];
if (s->last_frame)
return 0;
if (data) {
if (!s->psypp) {
if (avctx->channels <= 2) {
memcpy(s->samples + 1024 * avctx->channels, data,
1024 * avctx->channels * sizeof(s->samples[0]));
} else {
for (i = 0; i < 1024; i++)
for (ch = 0; ch < avctx->channels; ch++)
s->samples[(i + 1024) * avctx->channels + ch] =
((int16_t*)data)[i * avctx->channels +
channel_maps[avctx->channels-1][ch]];
}
} else {
start_ch = 0;
samples2 = s->samples + 1024 * avctx->channels;
for (i = 0; i < s->chan_map[0]; i++) {
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
ff_psy_preprocess(s->psypp,
(uint16_t*)data + channel_maps[avctx->channels-1][start_ch],
samples2 + start_ch, start_ch, chans);
start_ch += chans;
}
}
}
if (!avctx->frame_number) {
memcpy(s->samples, s->samples + 1024 * avctx->channels,
1024 * avctx->channels * sizeof(s->samples[0]));
return 0;
}
start_ch = 0;
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
for (ch = 0; ch < chans; ch++) {
IndividualChannelStream *ics = &cpe->ch[ch].ics;
int cur_channel = start_ch + ch;
samples2 = samples + cur_channel;
la = samples2 + (448+64) * avctx->channels;
if (!data)
la = NULL;
if (tag == TYPE_LFE) {
wi[ch].window_type[0] = ONLY_LONG_SEQUENCE;
wi[ch].window_shape = 0;
wi[ch].num_windows = 1;
wi[ch].grouping[0] = 1;
} else {
wi[ch] = s->psy.model->window(&s->psy, samples2, la, cur_channel,
ics->window_sequence[0]);
}
ics->window_sequence[1] = ics->window_sequence[0];
ics->window_sequence[0] = wi[ch].window_type[0];
ics->use_kb_window[1] = ics->use_kb_window[0];
ics->use_kb_window[0] = wi[ch].window_shape;
ics->num_windows = wi[ch].num_windows;
ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
ics->num_swb = tag == TYPE_LFE ? 12 : s->psy.num_bands[ics->num_windows == 8];
for (w = 0; w < ics->num_windows; w++)
ics->group_len[w] = wi[ch].grouping[w];
apply_window_and_mdct(avctx, s, &cpe->ch[ch], samples2);
}
start_ch += chans;
}
do {
int frame_bits;
init_put_bits(&s->pb, frame, buf_size*8);
if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT))
put_bitstream_info(avctx, s, LIBAVCODEC_IDENT);
start_ch = 0;
memset(chan_el_counter, 0, sizeof(chan_el_counter));
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
const float *coeffs[2];
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
put_bits(&s->pb, 3, tag);
put_bits(&s->pb, 4, chan_el_counter[tag]++);
for (ch = 0; ch < chans; ch++)
coeffs[ch] = cpe->ch[ch].coeffs;
s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch * 2 + ch;
s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
}
cpe->common_window = 0;
if (chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape) {
cpe->common_window = 1;
for (w = 0; w < wi[0].num_windows; w++) {
if (wi[0].grouping[w] != wi[1].grouping[w]) {
cpe->common_window = 0;
break;
}
}
}
s->cur_channel = start_ch * 2;
if (s->options.stereo_mode && cpe->common_window) {
if (s->options.stereo_mode > 0) {
IndividualChannelStream *ics = &cpe->ch[0].ics;
for (w = 0; w < ics->num_windows; w += ics->group_len[w])
for (g = 0; g < ics->num_swb; g++)
cpe->ms_mask[w*16+g] = 1;
} else if (s->coder->search_for_ms) {
s->coder->search_for_ms(s, cpe, s->lambda);
}
}
adjust_frame_information(s, cpe, chans);
if (chans == 2) {
put_bits(&s->pb, 1, cpe->common_window);
if (cpe->common_window) {
put_ics_info(s, &cpe->ch[0].ics);
encode_ms_info(&s->pb, cpe);
}
}
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch + ch;
encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
}
start_ch += chans;
}
frame_bits = put_bits_count(&s->pb);
if (frame_bits <= 6144 * avctx->channels - 3) {
s->psy.bitres.bits = frame_bits / avctx->channels;
break;
}
s->lambda *= avctx->bit_rate * 1024.0f / avctx->sample_rate / frame_bits;
} while (1);
put_bits(&s->pb, 3, TYPE_END);
flush_put_bits(&s->pb);
avctx->frame_bits = put_bits_count(&s->pb);
// rate control stuff
if (!(avctx->flags & CODEC_FLAG_QSCALE)) {
float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
s->lambda *= ratio;
s->lambda = FFMIN(s->lambda, 65536.f);
}
if (!data)
s->last_frame = 1;
memcpy(s->samples, s->samples + 1024 * avctx->channels,
1024 * avctx->channels * sizeof(s->samples[0]));
return put_bits_count(&s->pb)>>3;
}
static inline int put_bits_left(PutBitContext* s)
{
return (s->buf_end - s->buf) * 8 - put_bits_count(s);
}
static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
AACEncContext *s = avctx->priv_data;
float **samples = s->planar_samples, *samples2, *la, *overlap;
ChannelElement *cpe;
int i, ch, w, g, chans, tag, start_ch, ret;
int chan_el_counter[4];
FFPsyWindowInfo windows[AAC_MAX_CHANNELS];
if (s->last_frame == 2)
return 0;
/* add current frame to queue */
if (frame) {
if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
return ret;
}
copy_input_samples(s, frame);
if (s->psypp)
ff_psy_preprocess(s->psypp, s->planar_samples, s->channels);
if (!avctx->frame_number)
return 0;
start_ch = 0;
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
for (ch = 0; ch < chans; ch++) {
IndividualChannelStream *ics = &cpe->ch[ch].ics;
int cur_channel = start_ch + ch;
overlap = &samples[cur_channel][0];
samples2 = overlap + 1024;
la = samples2 + (448+64);
if (!frame)
la = NULL;
if (tag == TYPE_LFE) {
wi[ch].window_type[0] = ONLY_LONG_SEQUENCE;
wi[ch].window_shape = 0;
wi[ch].num_windows = 1;
wi[ch].grouping[0] = 1;
/* Only the lowest 12 coefficients are used in a LFE channel.
* The expression below results in only the bottom 8 coefficients
* being used for 11.025kHz to 16kHz sample rates.
*/
ics->num_swb = s->samplerate_index >= 8 ? 1 : 3;
} else {
wi[ch] = s->psy.model->window(&s->psy, samples2, la, cur_channel,
ics->window_sequence[0]);
}
ics->window_sequence[1] = ics->window_sequence[0];
ics->window_sequence[0] = wi[ch].window_type[0];
ics->use_kb_window[1] = ics->use_kb_window[0];
ics->use_kb_window[0] = wi[ch].window_shape;
ics->num_windows = wi[ch].num_windows;
ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
ics->num_swb = tag == TYPE_LFE ? ics->num_swb : s->psy.num_bands[ics->num_windows == 8];
for (w = 0; w < ics->num_windows; w++)
ics->group_len[w] = wi[ch].grouping[w];
apply_window_and_mdct(s, &cpe->ch[ch], overlap);
}
start_ch += chans;
}
if ((ret = ff_alloc_packet2(avctx, avpkt, 8192 * s->channels)) < 0)
return ret;
do {
int frame_bits;
init_put_bits(&s->pb, avpkt->data, avpkt->size);
if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT))
put_bitstream_info(s, LIBAVCODEC_IDENT);
start_ch = 0;
memset(chan_el_counter, 0, sizeof(chan_el_counter));
for (i = 0; i < s->chan_map[0]; i++) {
FFPsyWindowInfo* wi = windows + start_ch;
const float *coeffs[2];
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
put_bits(&s->pb, 3, tag);
put_bits(&s->pb, 4, chan_el_counter[tag]++);
for (ch = 0; ch < chans; ch++)
coeffs[ch] = cpe->ch[ch].coeffs;
s->psy.model->analyze(&s->psy, start_ch, coeffs, wi);
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch * 2 + ch;
s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
}
cpe->common_window = 0;
if (chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape) {
cpe->common_window = 1;
for (w = 0; w < wi[0].num_windows; w++) {
if (wi[0].grouping[w] != wi[1].grouping[w]) {
cpe->common_window = 0;
break;
}
}
}
s->cur_channel = start_ch * 2;
if (s->options.stereo_mode && cpe->common_window) {
if (s->options.stereo_mode > 0) {
IndividualChannelStream *ics = &cpe->ch[0].ics;
for (w = 0; w < ics->num_windows; w += ics->group_len[w])
for (g = 0; g < ics->num_swb; g++)
cpe->ms_mask[w*16+g] = 1;
} else if (s->coder->search_for_ms) {
s->coder->search_for_ms(s, cpe, s->lambda);
}
}
adjust_frame_information(cpe, chans);
if (chans == 2) {
put_bits(&s->pb, 1, cpe->common_window);
if (cpe->common_window) {
put_ics_info(s, &cpe->ch[0].ics);
encode_ms_info(&s->pb, cpe);
}
}
for (ch = 0; ch < chans; ch++) {
s->cur_channel = start_ch + ch;
encode_individual_channel(avctx, s, &cpe->ch[ch], cpe->common_window);
}
start_ch += chans;
}
frame_bits = put_bits_count(&s->pb);
if (frame_bits <= 6144 * s->channels - 3) {
s->psy.bitres.bits = frame_bits / s->channels;
break;
}
s->lambda *= avctx->bit_rate * 1024.0f / avctx->sample_rate / frame_bits;
} while (1);
put_bits(&s->pb, 3, TYPE_END);
flush_put_bits(&s->pb);
avctx->frame_bits = put_bits_count(&s->pb);
// rate control stuff
if (!(avctx->flags & CODEC_FLAG_QSCALE)) {
float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
s->lambda *= ratio;
s->lambda = FFMIN(s->lambda, 65536.f);
}
if (!frame)
s->last_frame++;
ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts,
&avpkt->duration);
avpkt->size = put_bits_count(&s->pb) >> 3;
*got_packet_ptr = 1;
return 0;
}
static int aac_encode_frame(AVCodecContext *avctx,
uint8_t *frame, int buf_size, void *data)
{
AACEncContext *s = avctx->priv_data;
int16_t *samples = s->samples, *samples2, *la;
ChannelElement *cpe;
int i, j, chans, tag, start_ch;
const uint8_t *chan_map = aac_chan_configs[avctx->channels-1];
int chan_el_counter[4];
if(s->last_frame)
return 0;
if(data){
if(!s->psypp){
memcpy(s->samples + 1024 * avctx->channels, data, 1024 * avctx->channels * sizeof(s->samples[0]));
}else{
start_ch = 0;
samples2 = s->samples + 1024 * avctx->channels;
for(i = 0; i < chan_map[0]; i++){
tag = chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
ff_psy_preprocess(s->psypp, (uint16_t*)data + start_ch, samples2 + start_ch, start_ch, chans);
start_ch += chans;
}
}
}
if(!avctx->frame_number){
memcpy(s->samples, s->samples + 1024 * avctx->channels, 1024 * avctx->channels * sizeof(s->samples[0]));
return 0;
}
init_put_bits(&s->pb, frame, buf_size*8);
if((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT)){
put_bitstream_info(avctx, s, LIBAVCODEC_IDENT);
}
start_ch = 0;
memset(chan_el_counter, 0, sizeof(chan_el_counter));
for(i = 0; i < chan_map[0]; i++){
FFPsyWindowInfo wi[2];
tag = chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
samples2 = samples + start_ch;
la = samples2 + 1024 * avctx->channels + start_ch;
if(!data) la = NULL;
for(j = 0; j < chans; j++){
IndividualChannelStream *ics = &cpe->ch[j].ics;
int k;
wi[j] = ff_psy_suggest_window(&s->psy, samples2, la, start_ch + j, ics->window_sequence[0]);
ics->window_sequence[1] = ics->window_sequence[0];
ics->window_sequence[0] = wi[j].window_type[0];
ics->use_kb_window[1] = ics->use_kb_window[0];
ics->use_kb_window[0] = wi[j].window_shape;
ics->num_windows = wi[j].num_windows;
ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
ics->num_swb = s->psy.num_bands[ics->num_windows == 8];
for(k = 0; k < ics->num_windows; k++)
ics->group_len[k] = wi[j].grouping[k];
s->cur_channel = start_ch + j;
apply_window_and_mdct(avctx, s, &cpe->ch[j], samples2, j);
s->coder->search_for_quantizers(avctx, s, &cpe->ch[j], s->lambda);
}
cpe->common_window = 0;
if(chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape){
cpe->common_window = 1;
for(j = 0; j < wi[0].num_windows; j++){
if(wi[0].grouping[j] != wi[1].grouping[j]){
cpe->common_window = 0;
break;
}
}
}
if(cpe->common_window && s->coder->search_for_ms)
s->coder->search_for_ms(s, cpe, s->lambda);
adjust_frame_information(s, cpe, chans);
put_bits(&s->pb, 3, tag);
put_bits(&s->pb, 4, chan_el_counter[tag]++);
if(chans == 2){
put_bits(&s->pb, 1, cpe->common_window);
if(cpe->common_window){
put_ics_info(s, &cpe->ch[0].ics);
encode_ms_info(&s->pb, cpe);
}
}
for(j = 0; j < chans; j++){
s->cur_channel = start_ch + j;
ff_psy_set_band_info(&s->psy, s->cur_channel, cpe->ch[j].coeffs, &wi[j]);
encode_individual_channel(avctx, s, &cpe->ch[j], cpe->common_window);
}
start_ch += chans;
}
put_bits(&s->pb, 3, TYPE_END);
flush_put_bits(&s->pb);
avctx->frame_bits = put_bits_count(&s->pb);
// rate control stuff
if(!(avctx->flags & CODEC_FLAG_QSCALE)){
float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
s->lambda *= ratio;
}
if (avctx->frame_bits > 6144*avctx->channels) {
av_log(avctx, AV_LOG_ERROR, "input buffer violation %d > %d.\n", avctx->frame_bits, 6144*avctx->channels);
}
if(!data)
s->last_frame = 1;
memcpy(s->samples, s->samples + 1024 * avctx->channels, 1024 * avctx->channels * sizeof(s->samples[0]));
return put_bits_count(&s->pb)>>3;
}
// Translate VASliceDataBuffer
static int
translate_VASliceDataBuffer(
vdpau_driver_data_t *driver_data,
object_context_p obj_context,
object_buffer_p obj_buffer
)
{
if (obj_context->vdp_codec == VDP_CODEC_H264) {
/* Check we have the start code */
/* XXX: check for other codecs too? */
/* XXX: this assumes we get SliceParams before SliceData */
static const uint8_t start_code_prefix[3] = { 0x00, 0x00, 0x01 };
VASliceParameterBufferH264 * const slice_params = obj_context->last_slice_params;
unsigned int i;
for (i = 0; i < obj_context->last_slice_params_count; i++) {
VASliceParameterBufferH264 * const slice_param = &slice_params[i];
uint8_t *buf = (uint8_t *)obj_buffer->buffer_data + slice_param->slice_data_offset;
if (memcmp(buf, start_code_prefix, sizeof(start_code_prefix)) != 0) {
if (append_VdpBitstreamBuffer(obj_context,
start_code_prefix,
sizeof(start_code_prefix)) < 0)
return 0;
}
if (append_VdpBitstreamBuffer(obj_context,
buf,
slice_param->slice_data_size) < 0)
return 0;
}
return 1;
}
if (obj_context->vdp_codec == VDP_CODEC_MPEG2)
return translate_VASliceDataBuffer_MPEG2(driver_data,
obj_context, obj_buffer);
#if USE_VDPAU_MPEG4
if (obj_context->vdp_codec == VDP_CODEC_MPEG4 &&
obj_context->vdp_bitstream_buffers_count == 0) {
PutBitContext pb;
uint8_t slice_header_buffer[32];
uint8_t *slice_header;
int slice_header_size;
const uint8_t *slice_data = obj_buffer->buffer_data;
uint32_t slice_data_size = obj_buffer->buffer_size;
VAPictureParameterBufferMPEG4 * const pic_param = obj_context->last_pic_param;
VASliceParameterBufferMPEG4 * const slice_param = obj_context->last_slice_params;
int time_incr = 1 + ilog2(pic_param->vop_time_increment_resolution - 1);
if (time_incr < 1)
time_incr = 1;
static const uint16_t VOP_STARTCODE = 0x01b6;
enum {
VOP_I_TYPE = 0,
VOP_P_TYPE,
VOP_B_TYPE,
VOP_S_TYPE
};
/* XXX: this is a hack to compute the length of
modulo_time_base "1" sequence. We probably should be
reconstructing through an extra VOP field in VA-API? */
int nbits = (32 + /* VOP start code */
2 + /* vop_coding_type */
1 + /* modulo_time_base "0" */
1 + /* marker_bit */
time_incr + /* vop_time_increment */
1 + /* marker_bit */
1 + /* vop_coded */
(pic_param->vop_fields.bits.vop_coding_type == VOP_P_TYPE ? 1 : 0) +
3 + /* intra_dc_vlc_thr */
(pic_param->vol_fields.bits.interlaced ? 2 : 0) +
5 + /* vop_quant */
(pic_param->vop_fields.bits.vop_coding_type != VOP_I_TYPE ? 3 : 0) +
(pic_param->vop_fields.bits.vop_coding_type == VOP_B_TYPE ? 3 : 0));
if ((nbits = slice_param->macroblock_offset - (nbits % 8)) < 0)
nbits += 8;
/* Reconstruct the VOP header */
init_put_bits(&pb, slice_header_buffer, sizeof(slice_header_buffer));
put_bits(&pb, 16, 0); /* vop header */
put_bits(&pb, 16, VOP_STARTCODE); /* vop header */
put_bits(&pb, 2, pic_param->vop_fields.bits.vop_coding_type);
while (nbits-- > 0)
put_bits(&pb, 1, 1); /* modulo_time_base "1" */
put_bits(&pb, 1, 0); /* modulo_time_base "0" */
put_bits(&pb, 1, 1); /* marker */
put_bits(&pb, time_incr, 0); /* time increment */
put_bits(&pb, 1, 1); /* marker */
put_bits(&pb, 1, 1); /* vop coded */
if (pic_param->vop_fields.bits.vop_coding_type == VOP_P_TYPE)
put_bits(&pb, 1, pic_param->vop_fields.bits.vop_rounding_type);
put_bits(&pb, 3, pic_param->vop_fields.bits.intra_dc_vlc_thr);
if (pic_param->vol_fields.bits.interlaced) {
put_bits(&pb, 1, pic_param->vop_fields.bits.top_field_first);
put_bits(&pb, 1, pic_param->vop_fields.bits.alternate_vertical_scan_flag);
}
put_bits(&pb, 5, slice_param->quant_scale);
if (pic_param->vop_fields.bits.vop_coding_type != VOP_I_TYPE)
put_bits(&pb, 3, pic_param->vop_fcode_forward);
if (pic_param->vop_fields.bits.vop_coding_type == VOP_B_TYPE)
put_bits(&pb, 3, pic_param->vop_fcode_backward);
/* Merge in bits from the first byte of the slice */
ASSERT((put_bits_count(&pb) % 8) == slice_param->macroblock_offset);
if ((put_bits_count(&pb) % 8) != slice_param->macroblock_offset)
return 0;
const int r = 8 - (put_bits_count(&pb) % 8);
if (r > 0)
put_bits(&pb, r, slice_data[0] & ((1U << r) - 1));
flush_put_bits(&pb);
ASSERT((put_bits_count(&pb) % 8) == 0);
slice_header_size = put_bits_count(&pb) / 8;
ASSERT(slice_header_size <= sizeof(slice_header_buffer));
slice_header = alloc_gen_slice_data(obj_context, slice_header_size);
if (!slice_header)
return 0;
memcpy(slice_header, slice_header_buffer, slice_header_size);
if (append_VdpBitstreamBuffer(obj_context, slice_header, slice_header_size) < 0)
return 0;
if (append_VdpBitstreamBuffer(obj_context, slice_data + 1, slice_data_size - 1) < 0)
return 0;
return 1;
}
#endif
if (append_VdpBitstreamBuffer(obj_context,
obj_buffer->buffer_data,
obj_buffer->buffer_size) < 0)
return 0;
return 1;
}
