2022-01-28 01:14:42 +01:00
/*
* Generic DCT based hybrid video encoder
* Copyright ( c ) 2000 , 2001 , 2002 Fabrice Bellard
* Copyright ( c ) 2002 - 2004 Michael Niedermayer
*
* This file is part of FFmpeg .
*
* FFmpeg is free software ; you can redistribute it and / or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation ; either
* version 2.1 of the License , or ( at your option ) any later version .
*
* FFmpeg is distributed in the hope that it will be useful ,
* but WITHOUT ANY WARRANTY ; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE . See the GNU
* Lesser General Public License for more details .
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg ; if not , write to the Free Software
* Foundation , Inc . , 51 Franklin Street , Fifth Floor , Boston , MA 02110 - 1301 USA
*/
/**
* @ file
* mpegvideo header .
*/
# ifndef AVCODEC_MPEGVIDEOENC_H
# define AVCODEC_MPEGVIDEOENC_H
2022-05-16 18:15:37 +02:00
# include <float.h>
2025-03-15 07:16:33 +01:00
# include "libavutil/avassert.h"
2022-01-28 01:14:42 +01:00
# include "libavutil/opt.h"
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
# include "fdctdsp.h"
2022-01-28 01:14:42 +01:00
# include "mpegvideo.h"
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
# include "mpegvideoencdsp.h"
# include "pixblockdsp.h"
# include "put_bits.h"
2022-01-28 20:43:54 +01:00
# include "ratecontrol.h"
2022-01-28 01:14:42 +01:00
2022-01-28 20:24:18 +01:00
# define MPVENC_MAX_B_FRAMES 16
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
typedef struct MPVEncContext {
MpegEncContext c ; ///< the common base context
/** bit output */
PutBitContext pb ;
2025-03-19 12:50:06 +01:00
unsigned int lambda ; ///< Lagrange multiplier used in rate distortion
unsigned int lambda2 ; ///< (lambda*lambda) >> FF_LAMBDA_SHIFT
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
int * lambda_table ;
int adaptive_quant ; ///< use adaptive quantization
int dquant ; ///< qscale difference to prev qscale
int skipdct ; ///< skip dct and code zero residual
int quantizer_noise_shaping ;
int luma_elim_threshold ;
int chroma_elim_threshold ;
int mpv_flags ; ///< flags set by private options
/// Bitfield containing information which frames to reconstruct.
int frame_reconstruction_bitfield ;
/**
* Reference to the source picture .
*/
AVFrame * new_pic ;
FDCTDSPContext fdsp ;
MpegvideoEncDSPContext mpvencdsp ;
PixblockDSPContext pdsp ;
int16_t ( * p_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) P-frame
int16_t ( * b_forw_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) forward mode B-frame
int16_t ( * b_back_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) backward mode B-frame
int16_t ( * b_bidir_forw_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) bidir mode B-frame
int16_t ( * b_bidir_back_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) bidir mode B-frame
int16_t ( * b_direct_mv_table ) [ 2 ] ; ///< MV table (1MV per MB) direct mode B-frame
int16_t ( * b_field_mv_table [ 2 ] [ 2 ] [ 2 ] ) [ 2 ] ; ///< MV table (4MV per MB) interlaced B-frame
uint8_t ( * p_field_select_table [ 2 ] ) ; ///< Only the first element is allocated
uint8_t ( * b_field_select_table [ 2 ] [ 2 ] ) ; ///< allocated jointly with p_field_select_table
uint16_t * mb_type ; ///< Table for candidate MB types
uint16_t * mb_var ; ///< Table for MB variances
uint16_t * mc_mb_var ; ///< Table for motion compensated MB variances
uint8_t * mb_mean ; ///< Table for MB luminance
uint64_t encoding_error [ MPV_MAX_PLANES ] ;
int intra_quant_bias ; ///< bias for the quantizer
int inter_quant_bias ; ///< bias for the quantizer
int min_qcoeff ; ///< minimum encodable coefficient
int max_qcoeff ; ///< maximum encodable coefficient
int ac_esc_length ; ///< num of bits needed to encode the longest esc
2025-03-19 13:03:22 +01:00
const uint8_t * intra_ac_vlc_length ;
const uint8_t * intra_ac_vlc_last_length ;
const uint8_t * intra_chroma_ac_vlc_length ;
const uint8_t * intra_chroma_ac_vlc_last_length ;
const uint8_t * inter_ac_vlc_length ;
const uint8_t * inter_ac_vlc_last_length ;
const uint8_t * luma_dc_vlc_length ;
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
int coded_score [ 12 ] ;
/** precomputed matrix (combine qscale and DCT renorm) */
int ( * q_intra_matrix ) [ 64 ] ;
int ( * q_chroma_intra_matrix ) [ 64 ] ;
int ( * q_inter_matrix ) [ 64 ] ;
/** identical to the above but for MMX & these are not permutated, second 64 entries are bias*/
uint16_t ( * q_intra_matrix16 ) [ 2 ] [ 64 ] ;
uint16_t ( * q_chroma_intra_matrix16 ) [ 2 ] [ 64 ] ;
uint16_t ( * q_inter_matrix16 ) [ 2 ] [ 64 ] ;
/* noise reduction */
void ( * denoise_dct ) ( struct MPVEncContext * s , int16_t * block ) ;
int ( * dct_error_sum ) [ 64 ] ;
int dct_count [ 2 ] ;
uint16_t ( * dct_offset ) [ 64 ] ;
/* statistics, used for 2-pass encoding */
int mv_bits ;
int i_tex_bits ;
int p_tex_bits ;
int i_count ;
int misc_bits ; ///< cbp, mb_type
int last_bits ; ///< temp var used for calculating the above vars
/* H.263 specific */
int mb_info ; ///< interval for outputting info about mb offsets as side data
int prev_mb_info , last_mb_info ;
int mb_info_size ;
uint8_t * mb_info_ptr ;
/* MJPEG specific */
struct MJpegContext * mjpeg_ctx ;
int esc_pos ;
/* MPEG-1 specific */
int last_mv_dir ; ///< last mv_dir, used for B-frame encoding
/* MPEG-4 specific */
PutBitContext tex_pb ; ///< used for data partitioned VOPs
PutBitContext pb2 ; ///< used for data partitioned VOPs
/* MSMPEG4 specific */
int esc3_level_length ;
/* RTP specific */
int rtp_mode ;
int rtp_payload_size ;
int error_rate ;
uint8_t * ptr_lastgob ;
void ( * encode_mb ) ( struct MPVEncContext * s , int16_t block [ ] [ 64 ] ,
int motion_x , int motion_y ) ;
int ( * dct_quantize ) ( struct MPVEncContext * s , int16_t * block /*align 16*/ , int n , int qscale , int * overflow ) ;
me_cmp_func ildct_cmp [ 2 ] ; ///< 0 = intra, 1 = non-intra
me_cmp_func n_sse_cmp [ 2 ] ; ///< either SSE or NSSE cmp func
me_cmp_func sad_cmp [ 2 ] ;
me_cmp_func sse_cmp [ 2 ] ;
int ( * sum_abs_dctelem ) ( const int16_t * block ) ;
int intra_penalty ;
} MPVEncContext ;
2025-02-28 18:56:15 +01:00
typedef struct MPVMainEncContext {
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
MPVEncContext s ; ///< The main slicecontext
2022-01-28 20:24:18 +01:00
2025-03-02 02:55:04 +01:00
int intra_only ; ///< if true, only intra pictures are generated
int gop_size ;
2025-03-18 11:20:25 +01:00
int max_b_frames ; ///< max number of B-frames
2025-03-02 02:55:04 +01:00
int picture_in_gop_number ; ///< 0-> first pic in gop, ...
int input_picture_number ; ///< used to set pic->display_picture_number
int coded_picture_number ; ///< used to set pic->coded_picture_number
2025-03-18 11:47:58 +01:00
MPVPicture * input_picture [ MPVENC_MAX_B_FRAMES + 1 ] ; ///< next pictures in display order
MPVPicture * reordered_input_picture [ MPVENC_MAX_B_FRAMES + 1 ] ; ///< next pictures in coded order
2025-03-02 02:55:04 +01:00
int64_t user_specified_pts ; ///< last non-zero pts from user-supplied AVFrame
/**
* pts difference between the first and second input frame , used for
* calculating dts of the first frame when there ' s a delay */
int64_t dts_delta ;
/**
* reordered pts to be used as dts for the next output frame when there ' s
* a delay */
int64_t reordered_pts ;
2022-01-28 20:24:18 +01:00
/// temporary frames used by b_frame_strategy = 2
AVFrame * tmp_frames [ MPVENC_MAX_B_FRAMES + 2 ] ;
int b_frame_strategy ;
int b_sensitivity ;
int brd_scale ;
2022-01-28 20:43:54 +01:00
2025-03-02 04:01:24 +01:00
int scenechange_threshold ;
2025-03-16 10:30:48 +01:00
int noise_reduction ;
2025-03-02 04:01:24 +01:00
float border_masking ;
int lmin , lmax ;
int vbv_ignore_qmax ;
2025-03-18 11:52:20 +01:00
/* MPEG-1/2 specific */
int vbv_delay_pos ; ///< offset of vbv_delay in the bitstream
2022-02-07 10:24:38 +01:00
const uint8_t * fcode_tab ; ///< smallest fcode needed for each MV
2025-03-01 02:04:41 +01:00
/* frame skip options */
int frame_skip_threshold ;
int frame_skip_factor ;
int frame_skip_exp ;
int frame_skip_cmp ;
me_cmp_func frame_skip_cmp_fn ;
2025-03-03 23:35:01 +01:00
int ( * encode_picture_header ) ( struct MPVMainEncContext * m ) ;
2022-01-28 20:43:54 +01:00
/* bit rate control */
2025-03-15 07:25:15 +01:00
int64_t bit_rate ;
2022-01-28 20:43:54 +01:00
int64_t total_bits ;
int frame_bits ; ///< bits used for the current frame
2022-01-29 05:34:14 +01:00
int header_bits ;
2022-01-28 20:43:54 +01:00
int stuffing_bits ; ///< bits used for stuffing
int next_lambda ; ///< next lambda used for retrying to encode a frame
2025-03-02 04:01:24 +01:00
int fixed_qscale ; ///< fixed qscale if non zero
2022-01-29 05:25:25 +01:00
int last_lambda_for [ 5 ] ; ///< last lambda for a specific pict type
int last_pict_type ; //FIXME removes
int last_non_b_pict_type ; ///< used for MPEG-4 gmc B-frames & ratecontrol
2022-01-28 20:43:54 +01:00
RateControlContext rc_context ; ///< contains stuff only accessed in ratecontrol.c
2022-10-13 02:15:46 +02:00
2025-03-19 00:39:31 +01:00
int me_penalty_compensation ;
int me_pre ; ///< prepass for motion estimation
2022-10-13 02:15:46 +02:00
int64_t mb_var_sum ; ///< sum of MB variance for current frame
int64_t mc_mb_var_sum ; ///< motion compensated MB variance for current frame
2022-10-25 18:27:17 +02:00
char * me_map_base ; ///< backs MotionEstContext.(map|score_map)
char * dct_error_sum_base ; ///< backs dct_error_sum
2025-03-16 09:02:05 +01:00
int16_t ( * mv_table_base ) [ 2 ] ;
2025-02-28 18:56:15 +01:00
} MPVMainEncContext ;
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
static inline const MPVMainEncContext * slice_to_mainenc ( const MPVEncContext * s )
2025-03-15 07:16:33 +01:00
{
# ifdef NO_SLICE_THREADING_HERE
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
av_assert2 ( s - > c . slice_context_count < = 1 & &
! ( s - > c . avctx - > codec - > capabilities & AV_CODEC_CAP_SLICE_THREADS ) ) ;
2025-03-15 07:16:33 +01:00
return ( const MPVMainEncContext * ) s ;
# else
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
return s - > c . encparent ;
2025-03-15 07:16:33 +01:00
# endif
}
2025-03-03 21:01:31 +01:00
# define MAX_FCODE 7
2022-01-28 01:14:42 +01:00
# define UNI_AC_ENC_INDEX(run,level) ((run)*128 + (level))
2023-10-15 23:27:24 +02:00
# define INPLACE_OFFSET 16
/* MB types for encoding */
# define CANDIDATE_MB_TYPE_INTRA (1 << 0)
# define CANDIDATE_MB_TYPE_INTER (1 << 1)
# define CANDIDATE_MB_TYPE_INTER4V (1 << 2)
# define CANDIDATE_MB_TYPE_SKIPPED (1 << 3)
# define CANDIDATE_MB_TYPE_DIRECT (1 << 4)
# define CANDIDATE_MB_TYPE_FORWARD (1 << 5)
# define CANDIDATE_MB_TYPE_BACKWARD (1 << 6)
# define CANDIDATE_MB_TYPE_BIDIR (1 << 7)
# define CANDIDATE_MB_TYPE_INTER_I (1 << 8)
# define CANDIDATE_MB_TYPE_FORWARD_I (1 << 9)
# define CANDIDATE_MB_TYPE_BACKWARD_I (1 << 10)
# define CANDIDATE_MB_TYPE_BIDIR_I (1 << 11)
# define CANDIDATE_MB_TYPE_DIRECT0 (1 << 12)
2022-01-28 01:14:42 +01:00
/* mpegvideo_enc common options */
# define FF_MPV_FLAG_SKIP_RD 0x0001
# define FF_MPV_FLAG_STRICT_GOP 0x0002
# define FF_MPV_FLAG_QP_RD 0x0004
# define FF_MPV_FLAG_CBP_RD 0x0008
# define FF_MPV_FLAG_NAQ 0x0010
# define FF_MPV_FLAG_MV0 0x0020
# define FF_MPV_OPT_CMP_FUNC \
2024-02-11 15:41:05 +01:00
{ " sad " , " Sum of absolute differences, fast " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_SAD } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " sse " , " Sum of squared errors " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_SSE } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " satd " , " Sum of absolute Hadamard transformed differences " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_SATD } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " dct " , " Sum of absolute DCT transformed differences " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_DCT } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " psnr " , " Sum of squared quantization errors, low quality " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_PSNR } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " bit " , " Number of bits needed for the block " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_BIT } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " rd " , " Rate distortion optimal, slow " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_RD } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " zero " , " Zero " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_ZERO } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " vsad " , " Sum of absolute vertical differences " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_VSAD } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " vsse " , " Sum of squared vertical differences " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_VSSE } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " nsse " , " Noise preserving sum of squared differences " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_NSSE } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " dct264 " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_DCT264 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " dctmax " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_DCTMAX } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " chroma " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_CHROMA } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
{ " msad " , " Sum of absolute differences, median predicted " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_CMP_MEDIAN_SAD } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " }
2022-01-28 01:14:42 +01:00
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
# define FF_MPV_OFFSET(x) offsetof(MPVEncContext, x)
2022-01-28 20:24:18 +01:00
# define FF_MPV_MAIN_OFFSET(x) offsetof(MPVMainEncContext, x)
2022-01-28 20:43:54 +01:00
# define FF_RC_OFFSET(x) offsetof(MPVMainEncContext, rc_context.x)
2022-01-28 01:14:42 +01:00
# define FF_MPV_OPT_FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
# define FF_MPV_COMMON_OPTS \
FF_MPV_OPT_CMP_FUNC , \
2024-02-11 15:41:05 +01:00
{ " mpv_flags " , " Flags common for all mpegvideo-based encoders. " , FF_MPV_OFFSET ( mpv_flags ) , AV_OPT_TYPE_FLAGS , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
{ " skip_rd " , " RD optimal MB level residual skipping " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_SKIP_RD } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
{ " strict_gop " , " Strictly enforce gop size " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_STRICT_GOP } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
{ " qp_rd " , " Use rate distortion optimization for qp selection " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_QP_RD } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
{ " cbp_rd " , " use rate distortion optimization for CBP " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_CBP_RD } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
{ " naq " , " normalize adaptive quantization " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_NAQ } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
2022-01-28 01:14:42 +01:00
{ " luma_elim_threshold " , " single coefficient elimination threshold for luminance (negative values also consider dc coefficient) " , \
FF_MPV_OFFSET ( luma_elim_threshold ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " chroma_elim_threshold " , " single coefficient elimination threshold for chrominance (negative values also consider dc coefficient) " , \
FF_MPV_OFFSET ( chroma_elim_threshold ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " quantizer_noise_shaping " , NULL , FF_MPV_OFFSET ( quantizer_noise_shaping ) , AV_OPT_TYPE_INT , { . i64 = 0 } , 0 , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " error_rate " , " Simulate errors in the bitstream to test error concealment. " , \
FF_MPV_OFFSET ( error_rate ) , AV_OPT_TYPE_INT , { . i64 = 0 } , 0 , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " qsquish " , " how to keep quantizer between qmin and qmax (0 = clip, 1 = use differentiable function) " , \
2025-02-28 23:47:52 +01:00
FF_RC_OFFSET ( qsquish ) , AV_OPT_TYPE_FLOAT , { . dbl = 0 } , 0 , 99 , FF_MPV_OPT_FLAGS } , \
{ " rc_qmod_amp " , " experimental quantizer modulation " , FF_RC_OFFSET ( qmod_amp ) , AV_OPT_TYPE_FLOAT , { . dbl = 0 } , - FLT_MAX , FLT_MAX , FF_MPV_OPT_FLAGS } , \
{ " rc_qmod_freq " , " experimental quantizer modulation " , FF_RC_OFFSET ( qmod_freq ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
2022-01-28 01:14:42 +01:00
{ " rc_eq " , " Set rate control equation. When computing the expression, besides the standard functions " \
" defined in the section 'Expression Evaluation', the following functions are available: " \
" bits2qp(bits), qp2bits(qp). Also the following constants are available: iTex pTex tex mv " \
" fCode iCount mcVar var isI isP isB avgQP qComp avgIITex avgPITex avgPPTex avgBPTex avgTex. " , \
2025-02-28 23:47:52 +01:00
FF_RC_OFFSET ( rc_eq ) , AV_OPT_TYPE_STRING , . flags = FF_MPV_OPT_FLAGS } , \
{ " rc_init_cplx " , " initial complexity for 1-pass encoding " , FF_RC_OFFSET ( initial_cplx ) , AV_OPT_TYPE_FLOAT , { . dbl = 0 } , - FLT_MAX , FLT_MAX , FF_MPV_OPT_FLAGS } , \
{ " rc_buf_aggressivity " , " currently useless " , FF_RC_OFFSET ( buffer_aggressivity ) , AV_OPT_TYPE_FLOAT , { . dbl = 1.0 } , - FLT_MAX , FLT_MAX , FF_MPV_OPT_FLAGS } , \
2025-03-02 04:01:24 +01:00
{ " border_mask " , " increase the quantizer for macroblocks close to borders " , FF_MPV_MAIN_OFFSET ( border_masking ) , AV_OPT_TYPE_FLOAT , { . dbl = 0 } , - FLT_MAX , FLT_MAX , FF_MPV_OPT_FLAGS } , \
{ " lmin " , " minimum Lagrange factor (VBR) " , FF_MPV_MAIN_OFFSET ( lmin ) , AV_OPT_TYPE_INT , { . i64 = 2 * FF_QP2LAMBDA } , 0 , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " lmax " , " maximum Lagrange factor (VBR) " , FF_MPV_MAIN_OFFSET ( lmax ) , AV_OPT_TYPE_INT , { . i64 = 31 * FF_QP2LAMBDA } , 0 , INT_MAX , FF_MPV_OPT_FLAGS } , \
2025-03-01 02:04:41 +01:00
{ " skip_threshold " , " Frame skip threshold " , FF_MPV_MAIN_OFFSET ( frame_skip_threshold ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " skip_factor " , " Frame skip factor " , FF_MPV_MAIN_OFFSET ( frame_skip_factor ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " skip_exp " , " Frame skip exponent " , FF_MPV_MAIN_OFFSET ( frame_skip_exp ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " skip_cmp " , " Frame skip compare function " , FF_MPV_MAIN_OFFSET ( frame_skip_cmp ) , AV_OPT_TYPE_INT , { . i64 = FF_CMP_DCTMAX } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS , . unit = " cmp_func " } , \
2025-03-16 10:30:48 +01:00
{ " noise_reduction " , " Noise reduction " , FF_MPV_MAIN_OFFSET ( noise_reduction ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
2022-01-28 01:14:42 +01:00
{ " ps " , " RTP payload size in bytes " , FF_MPV_OFFSET ( rtp_payload_size ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
# define FF_MPV_COMMON_BFRAME_OPTS \
2022-01-28 20:24:18 +01:00
{ " b_strategy " , " Strategy to choose between I/P/B-frames " , FF_MPV_MAIN_OFFSET ( b_frame_strategy ) , AV_OPT_TYPE_INT , { . i64 = 0 } , 0 , 2 , FF_MPV_OPT_FLAGS } , \
{ " b_sensitivity " , " Adjust sensitivity of b_frame_strategy 1 " , FF_MPV_MAIN_OFFSET ( b_sensitivity ) , AV_OPT_TYPE_INT , { . i64 = 40 } , 1 , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " brd_scale " , " Downscale frames for dynamic B-frame decision " , FF_MPV_MAIN_OFFSET ( brd_scale ) , AV_OPT_TYPE_INT , { . i64 = 0 } , 0 , 3 , FF_MPV_OPT_FLAGS } ,
2022-01-28 01:14:42 +01:00
2022-02-07 14:50:34 +01:00
# define FF_MPV_COMMON_MOTION_EST_OPTS \
2025-03-15 06:29:35 +01:00
{ " mv0 " , " always try a mb with mv=<0,0> " , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_MPV_FLAG_MV0 } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " mpv_flags " } , \
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
{ " motion_est " , " motion estimation algorithm " , FF_MPV_OFFSET ( c . me . motion_est ) , AV_OPT_TYPE_INT , { . i64 = FF_ME_EPZS } , FF_ME_ZERO , FF_ME_XONE , FF_MPV_OPT_FLAGS , . unit = " motion_est " } , \
2024-02-11 15:41:05 +01:00
{ " zero " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_ME_ZERO } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " motion_est " } , \
{ " epzs " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_ME_EPZS } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " motion_est " } , \
{ " xone " , NULL , 0 , AV_OPT_TYPE_CONST , { . i64 = FF_ME_XONE } , 0 , 0 , FF_MPV_OPT_FLAGS , . unit = " motion_est " } , \
2025-03-19 00:39:31 +01:00
{ " mepc " , " Motion estimation bitrate penalty compensation (1.0 = 256) " , FF_MPV_MAIN_OFFSET ( me_penalty_compensation ) , AV_OPT_TYPE_INT , { . i64 = 256 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
{ " mepre " , " pre motion estimation " , FF_MPV_MAIN_OFFSET ( me_pre ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
2022-02-07 14:50:34 +01:00
{ " intra_penalty " , " Penalty for intra blocks in block decision " , FF_MPV_OFFSET ( intra_penalty ) , AV_OPT_TYPE_INT , { . i64 = 0 } , 0 , INT_MAX / 2 , FF_MPV_OPT_FLAGS } , \
2025-03-15 06:29:35 +01:00
{ " sc_threshold " , " Scene change threshold " , FF_MPV_MAIN_OFFSET ( scenechange_threshold ) , AV_OPT_TYPE_INT , { . i64 = 0 } , INT_MIN , INT_MAX , FF_MPV_OPT_FLAGS } , \
2022-02-07 14:50:34 +01:00
2022-01-28 01:14:42 +01:00
extern const AVClass ff_mpv_enc_class ;
int ff_mpv_encode_init ( AVCodecContext * avctx ) ;
int ff_mpv_encode_end ( AVCodecContext * avctx ) ;
int ff_mpv_encode_picture ( AVCodecContext * avctx , AVPacket * pkt ,
const AVFrame * frame , int * got_packet ) ;
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
int ff_mpv_reallocate_putbitbuffer ( MPVEncContext * s , size_t threshold , size_t size_increase ) ;
2022-01-28 01:14:42 +01:00
void ff_write_quant_matrix ( PutBitContext * pb , uint16_t * matrix ) ;
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
void ff_dct_encode_init ( MPVEncContext * s ) ;
void ff_mpvenc_dct_init_mips ( MPVEncContext * s ) ;
void ff_dct_encode_init_x86 ( MPVEncContext * s ) ;
2022-01-28 01:14:42 +01:00
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
void ff_convert_matrix ( MPVEncContext * s , int ( * qmat ) [ 64 ] , uint16_t ( * qmat16 ) [ 2 ] [ 64 ] ,
2022-01-28 01:14:42 +01:00
const uint16_t * quant_matrix , int bias , int qmin , int qmax , int intra ) ;
2023-10-04 21:33:18 +02:00
void ff_block_permute ( int16_t * block , const uint8_t * permutation ,
2022-01-28 01:14:42 +01:00
const uint8_t * scantable , int last ) ;
avcodec/mpegvideoenc: Add MPVEncContext
Many of the fields of MpegEncContext (which is also used by decoders)
are actually only used by encoders. Therefore this commit adds
a new encoder-only structure and moves all of the encoder-only
fields to it except for those which require more explicit
synchronisation between the main slice context and the other
slice contexts. This synchronisation is currently mainly provided
by ff_update_thread_context() which simply copies most of
the main slice context over the other slice contexts. Fields
which are moved to the new MPVEncContext no longer participate
in this (which is desired, because it is horrible and for the
fields b) below wasteful) which means that some fields can only
be moved when explicit synchronisation code is added in later commits.
More explicitly, this commit moves the following fields:
a) Fields not copied by ff_update_duplicate_context():
dct_error_sum and dct_count; the former does not need synchronisation,
the latter is synchronised in merge_context_after_encode().
b) Fields which do not change after initialisation (these fields
could also be put into MPVMainEncContext at the cost of
an indirection to access them): lambda_table, adaptive_quant,
{luma,chroma}_elim_threshold, new_pic, fdsp, mpvencdsp, pdsp,
{p,b_forw,b_back,b_bidir_forw,b_bidir_back,b_direct,b_field}_mv_table,
[pb]_field_select_table, mb_{type,var,mean}, mc_mb_var, {min,max}_qcoeff,
{inter,intra}_quant_bias, ac_esc_length, the *_vlc_length fields,
the q_{intra,inter,chroma_intra}_matrix{,16}, dct_offset, mb_info,
mjpeg_ctx, rtp_mode, rtp_payload_size, encode_mb, all function
pointers, mpv_flags, quantizer_noise_shaping,
frame_reconstruction_bitfield, error_rate and intra_penalty.
c) Fields which are already (re)set explicitly: The PutBitContexts
pb, tex_pb, pb2; dquant, skipdct, encoding_error, the statistics
fields {mv,i_tex,p_tex,misc,last}_bits and i_count; last_mv_dir,
esc_pos (reset when writing the header).
d) Fields which are only used by encoders not supporting slice
threading for which synchronisation doesn't matter: esc3_level_length
and the remaining mb_info fields.
e) coded_score: This field is only really used when FF_MPV_FLAG_CBP_RD
is set (which implies trellis) and even then it is only used for
non-intra blocks. For these blocks dct_quantize_trellis_c() either
sets coded_score[n] or returns a last_non_zero value of -1
in which case coded_score will be reset in encode_mb_internal().
Therefore no old values are ever used.
The MotionEstContext has not been moved yet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2025-03-19 08:11:01 +01:00
static inline int get_bits_diff ( MPVEncContext * s )
2022-01-28 01:14:42 +01:00
{
const int bits = put_bits_count ( & s - > pb ) ;
const int last = s - > last_bits ;
s - > last_bits = bits ;
return bits - last ;
}
# endif
