libopenh264enc: Add an option for controlling the frame skipping
[libav.git] / libavcodec / libopenh264enc.c
CommitLineData
8a3d9ca6
MS
1/*
2 * OpenH264 video encoder
3 * Copyright (C) 2014 Martin Storsjo
4 *
5 * This file is part of Libav.
6 *
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include <wels/codec_api.h>
23#include <wels/codec_ver.h>
24
25#include "libavutil/attributes.h"
26#include "libavutil/common.h"
27#include "libavutil/opt.h"
28#include "libavutil/intreadwrite.h"
29#include "libavutil/mathematics.h"
30
31#include "avcodec.h"
32#include "internal.h"
33
34typedef struct SVCContext {
35 const AVClass *av_class;
36 ISVCEncoder *encoder;
37 int slice_mode;
38 int loopfilter;
39 char *profile;
c3e5c47b 40 int max_nal_size;
9e14a992 41 int skip_frames;
8a3d9ca6
MS
42} SVCContext;
43
25c29d32
MS
44#define OPENH264_VER_AT_LEAST(maj, min) \
45 ((OPENH264_MAJOR > (maj)) || \
46 (OPENH264_MAJOR == (maj) && OPENH264_MINOR >= (min)))
47
8a3d9ca6
MS
48#define OFFSET(x) offsetof(SVCContext, x)
49#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
50static const AVOption options[] = {
51 { "slice_mode", "Slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_AUTO_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
52 { "fixed", "A fixed number of slices", 0, AV_OPT_TYPE_CONST, { .i64 = SM_FIXEDSLCNUM_SLICE }, 0, 0, VE, "slice_mode" },
53 { "rowmb", "One slice per row of macroblocks", 0, AV_OPT_TYPE_CONST, { .i64 = SM_ROWMB_SLICE }, 0, 0, VE, "slice_mode" },
54 { "auto", "Automatic number of slices according to number of threads", 0, AV_OPT_TYPE_CONST, { .i64 = SM_AUTO_SLICE }, 0, 0, VE, "slice_mode" },
c3e5c47b 55 { "dyn", "Dynamic slicing", 0, AV_OPT_TYPE_CONST, { .i64 = SM_DYN_SLICE }, 0, 0, VE, "slice_mode" },
8a3d9ca6
MS
56 { "loopfilter", "Enable loop filter", OFFSET(loopfilter), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE },
57 { "profile", "Set profile restrictions", OFFSET(profile), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
c3e5c47b 58 { "max_nal_size", "Set maximum NAL size in bytes", OFFSET(max_nal_size), AV_OPT_TYPE_INT, { 0 }, 0, INT_MAX, VE },
9e14a992 59 { "allow_skip_frames", "Allow skipping frames to hit the target bitrate", OFFSET(skip_frames), AV_OPT_TYPE_INT, { 0 }, 0, 1, VE },
8a3d9ca6
MS
60 { NULL }
61};
62
63static const AVClass class = {
64 "libopenh264enc", av_default_item_name, options, LIBAVUTIL_VERSION_INT
65};
66
1a4c5fe5
GW
67// Convert libopenh264 log level to equivalent libav log level.
68static int libopenh264_to_libav_log_level(int libopenh264_log_level)
69{
70 if (libopenh264_log_level >= WELS_LOG_DETAIL) return AV_LOG_TRACE;
71 else if (libopenh264_log_level >= WELS_LOG_DEBUG) return AV_LOG_DEBUG;
72 else if (libopenh264_log_level >= WELS_LOG_INFO) return AV_LOG_VERBOSE;
73 else if (libopenh264_log_level >= WELS_LOG_WARNING) return AV_LOG_WARNING;
74 else if (libopenh264_log_level >= WELS_LOG_ERROR) return AV_LOG_ERROR;
75 else return AV_LOG_QUIET;
76}
77
78// This function will be provided to the libopenh264 library. The function will be called
79// when libopenh264 wants to log a message (error, warning, info, etc.). The signature for
80// this function (defined in .../codec/api/svc/codec_api.h) is:
81//
82// typedef void (*WelsTraceCallback) (void* ctx, int level, const char* string);
83
84static void libopenh264_trace_callback(void *ctx, int level, char const *msg)
85{
86 // The message will be logged only if the requested EQUIVALENT libav log level is
87 // less than or equal to the current libav log level.
88 int equiv_libav_log_level = libopenh264_to_libav_log_level(level);
89 av_log(ctx, equiv_libav_log_level, "%s\n", msg);
90}
91
8a3d9ca6
MS
92static av_cold int svc_encode_close(AVCodecContext *avctx)
93{
94 SVCContext *s = avctx->priv_data;
95
96 if (s->encoder)
97 WelsDestroySVCEncoder(s->encoder);
98 return 0;
99}
100
101static av_cold int svc_encode_init(AVCodecContext *avctx)
102{
103 SVCContext *s = avctx->priv_data;
104 SEncParamExt param = { 0 };
105 int err = AVERROR_UNKNOWN;
1a4c5fe5
GW
106 int log_level;
107 WelsTraceCallback callback_function;
8a3d9ca6
MS
108
109 // Mingw GCC < 4.7 on x86_32 uses an incorrect/buggy ABI for the WelsGetCodecVersion
110 // function (for functions returning larger structs), thus skip the check in those
111 // configurations.
112#if !defined(_WIN32) || !defined(__GNUC__) || !ARCH_X86_32 || AV_GCC_VERSION_AT_LEAST(4, 7)
440119b1 113 OpenH264Version libver = WelsGetCodecVersion();
8a3d9ca6
MS
114 if (memcmp(&libver, &g_stCodecVersion, sizeof(libver))) {
115 av_log(avctx, AV_LOG_ERROR, "Incorrect library version loaded\n");
116 return AVERROR(EINVAL);
117 }
118#endif
119
120 if (WelsCreateSVCEncoder(&s->encoder)) {
121 av_log(avctx, AV_LOG_ERROR, "Unable to create encoder\n");
122 return AVERROR_UNKNOWN;
123 }
124
1a4c5fe5
GW
125 // Pass all libopenh264 messages to our callback, to allow ourselves to filter them.
126 log_level = WELS_LOG_DETAIL;
127 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_LEVEL, &log_level);
128
129 // Set the logging callback function to one that uses av_log() (see implementation above).
130 callback_function = (WelsTraceCallback) libopenh264_trace_callback;
131 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK, (void *)&callback_function);
132
133 // Set the AVCodecContext as the libopenh264 callback context so that it can be passed to av_log().
134 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK_CONTEXT, (void *)&avctx);
135
8a3d9ca6
MS
136 (*s->encoder)->GetDefaultParams(s->encoder, &param);
137
138 param.fMaxFrameRate = avctx->time_base.den / avctx->time_base.num;
139 param.iPicWidth = avctx->width;
140 param.iPicHeight = avctx->height;
141 param.iTargetBitrate = avctx->bit_rate;
142 param.iMaxBitrate = FFMAX(avctx->rc_max_rate, avctx->bit_rate);
143 param.iRCMode = RC_QUALITY_MODE;
144 param.iTemporalLayerNum = 1;
145 param.iSpatialLayerNum = 1;
146 param.bEnableDenoise = 0;
147 param.bEnableBackgroundDetection = 1;
148 param.bEnableAdaptiveQuant = 1;
9e14a992 149 param.bEnableFrameSkip = s->skip_frames;
8a3d9ca6
MS
150 param.bEnableLongTermReference = 0;
151 param.iLtrMarkPeriod = 30;
152 param.uiIntraPeriod = avctx->gop_size;
25c29d32
MS
153#if OPENH264_VER_AT_LEAST(1, 4)
154 param.eSpsPpsIdStrategy = CONSTANT_ID;
155#else
8a3d9ca6 156 param.bEnableSpsPpsIdAddition = 0;
25c29d32 157#endif
8a3d9ca6
MS
158 param.bPrefixNalAddingCtrl = 0;
159 param.iLoopFilterDisableIdc = !s->loopfilter;
160 param.iEntropyCodingModeFlag = 0;
161 param.iMultipleThreadIdc = avctx->thread_count;
162 if (s->profile && !strcmp(s->profile, "main"))
163 param.iEntropyCodingModeFlag = 1;
164 else if (!s->profile && avctx->coder_type == FF_CODER_TYPE_AC)
165 param.iEntropyCodingModeFlag = 1;
166
167 param.sSpatialLayers[0].iVideoWidth = param.iPicWidth;
168 param.sSpatialLayers[0].iVideoHeight = param.iPicHeight;
169 param.sSpatialLayers[0].fFrameRate = param.fMaxFrameRate;
170 param.sSpatialLayers[0].iSpatialBitrate = param.iTargetBitrate;
171 param.sSpatialLayers[0].iMaxSpatialBitrate = param.iMaxBitrate;
172
c3e5c47b
MG
173 if ((avctx->slices > 1) && (s->max_nal_size)){
174 av_log(avctx,AV_LOG_ERROR,"Invalid combination -slices %d and -max_nal_size %d.\n",avctx->slices,s->max_nal_size);
175 goto fail;
176 }
177
8a3d9ca6
MS
178 if (avctx->slices > 1)
179 s->slice_mode = SM_FIXEDSLCNUM_SLICE;
c3e5c47b
MG
180
181 if (s->max_nal_size)
182 s->slice_mode = SM_DYN_SLICE;
183
8a3d9ca6
MS
184 param.sSpatialLayers[0].sSliceCfg.uiSliceMode = s->slice_mode;
185 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceNum = avctx->slices;
186
c3e5c47b
MG
187 if (s->slice_mode == SM_DYN_SLICE) {
188 if (s->max_nal_size){
189 param.uiMaxNalSize = s->max_nal_size;
190 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
191 } else {
192 if (avctx->rtp_payload_size) {
193 av_log(avctx,AV_LOG_DEBUG,"Using RTP Payload size for uiMaxNalSize");
194 param.uiMaxNalSize = avctx->rtp_payload_size;
195 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = avctx->rtp_payload_size;
196 } else {
197 av_log(avctx,AV_LOG_ERROR,"Invalid -max_nal_size, specify a valid max_nal_size to use -slice_mode dyn\n");
198 goto fail;
199 }
200 }
201 }
202
8a3d9ca6
MS
203 if ((*s->encoder)->InitializeExt(s->encoder, &param) != cmResultSuccess) {
204 av_log(avctx, AV_LOG_ERROR, "Initialize failed\n");
205 goto fail;
206 }
207
7c6eb0a1 208 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
8a3d9ca6
MS
209 SFrameBSInfo fbi = { 0 };
210 int i, size = 0;
211 (*s->encoder)->EncodeParameterSets(s->encoder, &fbi);
212 for (i = 0; i < fbi.sLayerInfo[0].iNalCount; i++)
213 size += fbi.sLayerInfo[0].pNalLengthInByte[i];
059a9348 214 avctx->extradata = av_mallocz(size + AV_INPUT_BUFFER_PADDING_SIZE);
8a3d9ca6
MS
215 if (!avctx->extradata) {
216 err = AVERROR(ENOMEM);
217 goto fail;
218 }
219 avctx->extradata_size = size;
220 memcpy(avctx->extradata, fbi.sLayerInfo[0].pBsBuf, size);
221 }
222
223 return 0;
224
225fail:
226 svc_encode_close(avctx);
227 return err;
228}
229
230static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
231 const AVFrame *frame, int *got_packet)
232{
233 SVCContext *s = avctx->priv_data;
234 SFrameBSInfo fbi = { 0 };
235 int i, ret;
236 int encoded;
237 SSourcePicture sp = { 0 };
238 int size = 0, layer, first_layer = 0;
239 int layer_size[MAX_LAYER_NUM_OF_FRAME] = { 0 };
240
241 sp.iColorFormat = videoFormatI420;
242 for (i = 0; i < 3; i++) {
243 sp.iStride[i] = frame->linesize[i];
244 sp.pData[i] = frame->data[i];
245 }
246 sp.iPicWidth = avctx->width;
247 sp.iPicHeight = avctx->height;
248
249 encoded = (*s->encoder)->EncodeFrame(s->encoder, &sp, &fbi);
250 if (encoded != cmResultSuccess) {
251 av_log(avctx, AV_LOG_ERROR, "EncodeFrame failed\n");
252 return AVERROR_UNKNOWN;
253 }
254 if (fbi.eFrameType == videoFrameTypeSkip) {
255 av_log(avctx, AV_LOG_DEBUG, "frame skipped\n");
256 return 0;
257 }
258 first_layer = 0;
3852e2c9 259 // Normal frames are returned with one single layer, while IDR
8a3d9ca6
MS
260 // frames have two layers, where the first layer contains the SPS/PPS.
261 // If using global headers, don't include the SPS/PPS in the returned
262 // packet - thus, only return one layer.
7c6eb0a1 263 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
8a3d9ca6
MS
264 first_layer = fbi.iLayerNum - 1;
265
266 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
267 for (i = 0; i < fbi.sLayerInfo[layer].iNalCount; i++)
268 layer_size[layer] += fbi.sLayerInfo[layer].pNalLengthInByte[i];
269 size += layer_size[layer];
270 }
6996fd20 271 av_log(avctx, AV_LOG_DEBUG, "%d slices\n", fbi.sLayerInfo[fbi.iLayerNum - 1].iNalCount);
8a3d9ca6
MS
272
273 if ((ret = ff_alloc_packet(avpkt, size))) {
274 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
275 return ret;
276 }
277 size = 0;
278 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
279 memcpy(avpkt->data + size, fbi.sLayerInfo[layer].pBsBuf, layer_size[layer]);
280 size += layer_size[layer];
281 }
282 avpkt->pts = frame->pts;
283 if (fbi.eFrameType == videoFrameTypeIDR)
284 avpkt->flags |= AV_PKT_FLAG_KEY;
285 *got_packet = 1;
286 return 0;
287}
288
289AVCodec ff_libopenh264_encoder = {
290 .name = "libopenh264",
3852e2c9 291 .long_name = NULL_IF_CONFIG_SMALL("OpenH264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
8a3d9ca6
MS
292 .type = AVMEDIA_TYPE_VIDEO,
293 .id = AV_CODEC_ID_H264,
294 .priv_data_size = sizeof(SVCContext),
295 .init = svc_encode_init,
296 .encode2 = svc_encode_frame,
297 .close = svc_encode_close,
def97856 298 .capabilities = AV_CODEC_CAP_AUTO_THREADS,
58a840e2
MS
299 .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P,
300 AV_PIX_FMT_NONE },
8a3d9ca6
MS
301 .priv_class = &class,
302};