FFmpeg
qsvenc_h264.c
Go to the documentation of this file.
1 /*
2  * Intel MediaSDK QSV based H.264 encoder
3  *
4  * copyright (c) 2013 Yukinori Yamazoe
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * FFmpeg is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with FFmpeg; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 
24 #include <stdint.h>
25 #include <sys/types.h>
26 
27 #include <mfx/mfxvideo.h>
28 
29 #include "libavutil/common.h"
30 #include "libavutil/opt.h"
31 
32 #include "avcodec.h"
33 #include "internal.h"
34 #include "qsv.h"
35 #include "qsv_internal.h"
36 #include "qsvenc.h"
37 #include "atsc_a53.h"
38 
39 typedef struct QSVH264EncContext {
40  AVClass *class;
43 
45  const AVFrame *frame, mfxEncodeCtrl* enc_ctrl)
46 {
47  QSVH264EncContext *qh264 = avctx->priv_data;
48  QSVEncContext *q = &qh264->qsv;
49 
50  if (q->a53_cc && frame) {
51  mfxPayload* payload;
52  mfxU8* sei_data;
53  size_t sei_size;
54  int res;
55 
56  res = ff_alloc_a53_sei(frame, sizeof(mfxPayload) + 2, (void**)&payload, &sei_size);
57  if (res < 0 || !payload)
58  return res;
59 
60  sei_data = (mfxU8*)(payload + 1);
61  // SEI header
62  sei_data[0] = 4;
63  sei_data[1] = (mfxU8)sei_size; // size of SEI data
64  // SEI data filled in by ff_alloc_a53_sei
65 
66  payload->BufSize = sei_size + 2;
67  payload->NumBit = payload->BufSize * 8;
68  payload->Type = 4;
69  payload->Data = sei_data;
70 
71  enc_ctrl->NumExtParam = 0;
72  enc_ctrl->NumPayload = 1;
73  enc_ctrl->Payload[0] = payload;
74  }
75  return 0;
76 }
77 
79 {
80  QSVH264EncContext *q = avctx->priv_data;
81 
83  return ff_qsv_enc_init(avctx, &q->qsv);
84 }
85 
87  const AVFrame *frame, int *got_packet)
88 {
89  QSVH264EncContext *q = avctx->priv_data;
90 
91  return ff_qsv_encode(avctx, &q->qsv, pkt, frame, got_packet);
92 }
93 
95 {
96  QSVH264EncContext *q = avctx->priv_data;
97 
98  return ff_qsv_enc_close(avctx, &q->qsv);
99 }
100 
101 #define OFFSET(x) offsetof(QSVH264EncContext, x)
102 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
103 static const AVOption options[] = {
105 
106  { "cavlc", "Enable CAVLC", OFFSET(qsv.cavlc), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
107 #if QSV_HAVE_VCM
108  { "vcm", "Use the video conferencing mode ratecontrol", OFFSET(qsv.vcm), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
109 #endif
110  { "idr_interval", "Distance (in I-frames) between IDR frames", OFFSET(qsv.idr_interval), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE },
111  { "pic_timing_sei", "Insert picture timing SEI with pic_struct_syntax element", OFFSET(qsv.pic_timing_sei), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, VE },
112  { "single_sei_nal_unit", "Put all the SEI messages into one NALU", OFFSET(qsv.single_sei_nal_unit), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
113  { "max_dec_frame_buffering", "Maximum number of frames buffered in the DPB", OFFSET(qsv.max_dec_frame_buffering), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, UINT16_MAX, VE },
114 
115 #if QSV_HAVE_LA
116  { "look_ahead", "Use VBR algorithm with look ahead", OFFSET(qsv.look_ahead), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
117  { "look_ahead_depth", "Depth of look ahead in number frames", OFFSET(qsv.look_ahead_depth), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 100, VE },
118 #endif
119 #if QSV_HAVE_LA_DS
120  { "look_ahead_downsampling", "Downscaling factor for the frames saved for the lookahead analysis", OFFSET(qsv.look_ahead_downsampling),
121  AV_OPT_TYPE_INT, { .i64 = MFX_LOOKAHEAD_DS_UNKNOWN }, MFX_LOOKAHEAD_DS_UNKNOWN, MFX_LOOKAHEAD_DS_4x, VE, "look_ahead_downsampling" },
122  { "unknown" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_LOOKAHEAD_DS_UNKNOWN }, INT_MIN, INT_MAX, VE, "look_ahead_downsampling" },
123  { "auto" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_LOOKAHEAD_DS_UNKNOWN }, INT_MIN, INT_MAX, VE, "look_ahead_downsampling" },
124  { "off" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_LOOKAHEAD_DS_OFF }, INT_MIN, INT_MAX, VE, "look_ahead_downsampling" },
125  { "2x" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_LOOKAHEAD_DS_2x }, INT_MIN, INT_MAX, VE, "look_ahead_downsampling" },
126  { "4x" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_LOOKAHEAD_DS_4x }, INT_MIN, INT_MAX, VE, "look_ahead_downsampling" },
127 #endif
128 
129  { "int_ref_type", "Intra refresh type", OFFSET(qsv.int_ref_type), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, UINT16_MAX, VE, "int_ref_type" },
130  { "none", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = 0 }, .flags = VE, "int_ref_type" },
131  { "vertical", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = 1 }, .flags = VE, "int_ref_type" },
132  { "int_ref_cycle_size", "Number of frames in the intra refresh cycle", OFFSET(qsv.int_ref_cycle_size), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, UINT16_MAX, VE },
133  { "int_ref_qp_delta", "QP difference for the refresh MBs", OFFSET(qsv.int_ref_qp_delta), AV_OPT_TYPE_INT, { .i64 = INT16_MIN }, INT16_MIN, INT16_MAX, VE },
134  { "recovery_point_sei", "Insert recovery point SEI messages", OFFSET(qsv.recovery_point_sei), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE },
135 
136  { "profile", NULL, OFFSET(qsv.profile), AV_OPT_TYPE_INT, { .i64 = MFX_PROFILE_UNKNOWN }, 0, INT_MAX, VE, "profile" },
137  { "unknown" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_UNKNOWN }, INT_MIN, INT_MAX, VE, "profile" },
138  { "baseline", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_AVC_BASELINE }, INT_MIN, INT_MAX, VE, "profile" },
139  { "main" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_AVC_MAIN }, INT_MIN, INT_MAX, VE, "profile" },
140  { "high" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_AVC_HIGH }, INT_MIN, INT_MAX, VE, "profile" },
141 
142  { "a53cc" , "Use A53 Closed Captions (if available)", OFFSET(qsv.a53_cc), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, VE},
143 
144  { "aud", "Insert the Access Unit Delimiter NAL", OFFSET(qsv.aud), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE},
145 
146 #if QSV_HAVE_MF
147  { "mfmode", "Multi-Frame Mode", OFFSET(qsv.mfmode), AV_OPT_TYPE_INT, { .i64 = MFX_MF_AUTO }, MFX_MF_DEFAULT, MFX_MF_AUTO, VE, "mfmode"},
148  { "off" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_MF_DISABLED }, INT_MIN, INT_MAX, VE, "mfmode" },
149  { "auto" , NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_MF_AUTO }, INT_MIN, INT_MAX, VE, "mfmode" },
150 #endif
151 
152  { "repeat_pps", "repeat pps for every frame", OFFSET(qsv.repeat_pps), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
153 
154  { NULL },
155 };
156 
157 static const AVClass class = {
158  .class_name = "h264_qsv encoder",
159  .item_name = av_default_item_name,
160  .option = options,
162 };
163 
165  { "b", "1M" },
166  { "refs", "0" },
167  // same as the x264 default
168  { "g", "250" },
169  { "bf", "3" },
170  { "qmin", "-1" },
171  { "qmax", "-1" },
172  { "trellis", "-1" },
173  { "flags", "+cgop" },
174  { NULL },
175 };
176 
178  .name = "h264_qsv",
179  .long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10 (Intel Quick Sync Video acceleration)"),
180  .priv_data_size = sizeof(QSVH264EncContext),
182  .id = AV_CODEC_ID_H264,
183  .init = qsv_enc_init,
184  .encode2 = qsv_enc_frame,
185  .close = qsv_enc_close,
186  .capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HYBRID,
187  .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_NV12,
190  AV_PIX_FMT_NONE },
191  .priv_class = &class,
192  .defaults = qsv_enc_defaults,
193  .caps_internal = FF_CODEC_CAP_INIT_CLEANUP,
194  .wrapper_name = "qsv",
195  .hw_configs = ff_qsv_enc_hw_configs,
196 };
qsv_enc_frame
static int qsv_enc_frame(AVCodecContext *avctx, AVPacket *pkt, const AVFrame *frame, int *got_packet)
Definition: qsvenc_h264.c:86
AVCodec
AVCodec.
Definition: codec.h:202
ff_alloc_a53_sei
int ff_alloc_a53_sei(const AVFrame *frame, size_t prefix_len, void **data, size_t *sei_size)
Check AVFrame for A53 side data and allocate and fill SEI message with A53 info.
Definition: atsc_a53.c:25
qsv_h264_set_encode_ctrl
static int qsv_h264_set_encode_ctrl(AVCodecContext *avctx, const AVFrame *frame, mfxEncodeCtrl *enc_ctrl)
Definition: qsvenc_h264.c:44
AVPixelFormat
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
opt.h
qsv_enc_init
static av_cold int qsv_enc_init(AVCodecContext *avctx)
Definition: qsvenc_h264.c:78
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:317
internal.h
MFX_LOOKAHEAD_DS_2x
#define MFX_LOOKAHEAD_DS_2x
Definition: qsvenc.h:73
AVOption
AVOption.
Definition: opt.h:247
VE
#define VE
Definition: qsvenc_h264.c:102
init
static int init
Definition: av_tx.c:47
OFFSET
#define OFFSET(x)
Definition: qsvenc_h264.c:101
type
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf type
Definition: writing_filters.txt:86
MFX_LOOKAHEAD_DS_UNKNOWN
#define MFX_LOOKAHEAD_DS_UNKNOWN
Definition: qsvenc.h:71
qsv_internal.h
ff_qsv_enc_hw_configs
const AVCodecHWConfigInternal *const ff_qsv_enc_hw_configs[]
Definition: qsvenc.c:1798
pkt
AVPacket * pkt
Definition: movenc.c:59
av_cold
#define av_cold
Definition: attributes.h:90
QSVEncContext
Definition: qsvenc.h:105
qsvenc.h
pix_fmts
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:296
AV_CODEC_ID_H264
@ AV_CODEC_ID_H264
Definition: codec_id.h:77
AVCodecDefault
Definition: internal.h:215
LIBAVUTIL_VERSION_INT
#define LIBAVUTIL_VERSION_INT
Definition: version.h:85
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
NULL
#define NULL
Definition: coverity.c:32
QSVH264EncContext::qsv
QSVEncContext qsv
Definition: qsvenc_h264.c:41
qsv.h
QSV_COMMON_OPTS
#define QSV_COMMON_OPTS
Definition: qsvenc.h:77
av_default_item_name
const char * av_default_item_name(void *ptr)
Return the context name.
Definition: log.c:235
QSVH264EncContext
Definition: qsvenc_h264.c:39
MFX_LOOKAHEAD_DS_4x
#define MFX_LOOKAHEAD_DS_4x
Definition: qsvenc.h:74
AV_PIX_FMT_QSV
@ AV_PIX_FMT_QSV
HW acceleration through QSV, data[3] contains a pointer to the mfxFrameSurface1 structure.
Definition: pixfmt.h:212
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:117
qsv_enc_close
static av_cold int qsv_enc_close(AVCodecContext *avctx)
Definition: qsvenc_h264.c:94
QSVEncContext::a53_cc
int a53_cc
Definition: qsvenc.h:198
options
static const AVOption options[]
Definition: qsvenc_h264.c:103
ff_h264_qsv_encoder
const AVCodec ff_h264_qsv_encoder
Definition: qsvenc_h264.c:177
ff_qsv_enc_close
int ff_qsv_enc_close(AVCodecContext *avctx, QSVEncContext *q)
Definition: qsvenc.c:1744
QSVEncContext::set_encode_ctrl_cb
SetEncodeCtrlCB * set_encode_ctrl_cb
Definition: qsvenc.h:204
FF_CODEC_CAP_INIT_CLEANUP
#define FF_CODEC_CAP_INIT_CLEANUP
The codec allows calling the close function for deallocation even if the init function returned a fai...
Definition: internal.h:50
common.h
AVCodec::name
const char * name
Name of the codec implementation.
Definition: codec.h:209
MFX_LOOKAHEAD_DS_OFF
#define MFX_LOOKAHEAD_DS_OFF
Definition: qsvenc.h:72
avcodec.h
AV_PIX_FMT_NV12
@ AV_PIX_FMT_NV12
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:89
AVClass::class_name
const char * class_name
The name of the class; usually it is the same name as the context structure type to which the AVClass...
Definition: log.h:71
frame
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
Definition: filter_design.txt:264
atsc_a53.h
AVCodecContext
main external API structure.
Definition: avcodec.h:383
AV_PIX_FMT_NONE
@ AV_PIX_FMT_NONE
Definition: pixfmt.h:65
AV_OPT_TYPE_INT
@ AV_OPT_TYPE_INT
Definition: opt.h:224
AV_CODEC_CAP_DELAY
#define AV_CODEC_CAP_DELAY
Encoder or decoder requires flushing with NULL input at the end in order to give the complete and cor...
Definition: codec.h:82
AV_PIX_FMT_P010
#define AV_PIX_FMT_P010
Definition: pixfmt.h:453
AVMEDIA_TYPE_VIDEO
@ AVMEDIA_TYPE_VIDEO
Definition: avutil.h:201
ff_qsv_enc_init
int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q)
Definition: qsvenc.c:1267
AV_CODEC_CAP_HYBRID
#define AV_CODEC_CAP_HYBRID
Codec is potentially backed by a hardware implementation, but not necessarily.
Definition: codec.h:169
qsv_enc_defaults
static const AVCodecDefault qsv_enc_defaults[]
Definition: qsvenc_h264.c:164
AVPacket
This structure stores compressed data.
Definition: packet.h:350
AVCodecContext::priv_data
void * priv_data
Definition: avcodec.h:410
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Definition: opt.h:241
AV_OPT_TYPE_CONST
@ AV_OPT_TYPE_CONST
Definition: opt.h:233
ff_qsv_encode
int ff_qsv_encode(AVCodecContext *avctx, QSVEncContext *q, AVPacket *pkt, const AVFrame *frame, int *got_packet)
Definition: qsvenc.c:1674