2 * OpenH264 video encoder
3 * Copyright (C) 2014 Martin Storsjo
5 * This file is part of FFmpeg.
7 * FFmpeg 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.
12 * FFmpeg 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.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22 #include <wels/codec_api.h>
23 #include <wels/codec_ver.h>
25 #include "libavutil/attributes.h"
26 #include "libavutil/common.h"
27 #include "libavutil/opt.h"
28 #include "libavutil/internal.h"
29 #include "libavutil/intreadwrite.h"
30 #include "libavutil/mathematics.h"
34 #include "libopenh264.h"
36 #if !OPENH264_VER_AT_LEAST(1, 6)
37 #define SM_SIZELIMITED_SLICE SM_DYN_SLICE
40 #define TARGET_BITRATE_DEFAULT 2*1000*1000
42 typedef struct SVCContext {
43 const AVClass *av_class;
57 #define OFFSET(x) offsetof(SVCContext, x)
58 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
59 #define DEPRECATED AV_OPT_FLAG_DEPRECATED
60 static const AVOption options[] = {
61 #if FF_API_OPENH264_SLICE_MODE
62 #if OPENH264_VER_AT_LEAST(1, 6)
63 { "slice_mode", "set slice mode, use slices/max_nal_size", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_FIXEDSLCNUM_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE|DEPRECATED, "slice_mode" },
65 { "slice_mode", "set slice mode, use slices/max_nal_size", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_AUTO_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE|DEPRECATED, "slice_mode" },
67 { "fixed", "a fixed number of slices", 0, AV_OPT_TYPE_CONST, { .i64 = SM_FIXEDSLCNUM_SLICE }, 0, 0, VE, "slice_mode" },
68 #if OPENH264_VER_AT_LEAST(1, 6)
69 { "dyn", "Size limited (compatibility name)", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
70 { "sizelimited", "Size limited", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
72 { "rowmb", "one slice per row of macroblocks", 0, AV_OPT_TYPE_CONST, { .i64 = SM_ROWMB_SLICE }, 0, 0, VE, "slice_mode" },
73 { "auto", "automatic number of slices according to number of threads", 0, AV_OPT_TYPE_CONST, { .i64 = SM_AUTO_SLICE }, 0, 0, VE, "slice_mode" },
74 { "dyn", "Dynamic slicing", 0, AV_OPT_TYPE_CONST, { .i64 = SM_DYN_SLICE }, 0, 0, VE, "slice_mode" },
77 { "loopfilter", "enable loop filter", OFFSET(loopfilter), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE },
78 { "profile", "set profile restrictions", OFFSET(profile), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
79 { "max_nal_size", "set maximum NAL size in bytes", OFFSET(max_nal_size), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE },
80 { "allow_skip_frames", "allow skipping frames to hit the target bitrate", OFFSET(skip_frames), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
81 { "cabac", "Enable cabac", OFFSET(cabac), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
83 { "rc_mode", "Select rate control mode", OFFSET(rc_mode), AV_OPT_TYPE_INT, { .i64 = RC_QUALITY_MODE }, RC_OFF_MODE, RC_TIMESTAMP_MODE, VE, "rc_mode" },
84 { "off", "bit rate control off", 0, AV_OPT_TYPE_CONST, { .i64 = RC_OFF_MODE }, 0, 0, VE, "rc_mode" },
85 { "quality", "quality mode", 0, AV_OPT_TYPE_CONST, { .i64 = RC_QUALITY_MODE }, 0, 0, VE, "rc_mode" },
86 { "bitrate", "bitrate mode", 0, AV_OPT_TYPE_CONST, { .i64 = RC_BITRATE_MODE }, 0, 0, VE, "rc_mode" },
87 { "buffer", "using buffer status to adjust the video quality (no bitrate control)", 0, AV_OPT_TYPE_CONST, { .i64 = RC_BUFFERBASED_MODE }, 0, 0, VE, "rc_mode" },
88 #if OPENH264_VER_AT_LEAST(1, 4)
89 { "timestamp", "bit rate control based on timestamp", 0, AV_OPT_TYPE_CONST, { .i64 = RC_TIMESTAMP_MODE }, 0, 0, VE, "rc_mode" },
95 static const AVClass class = {
96 .class_name = "libopenh264enc",
97 .item_name = av_default_item_name,
99 .version = LIBAVUTIL_VERSION_INT,
102 static av_cold int svc_encode_close(AVCodecContext *avctx)
104 SVCContext *s = avctx->priv_data;
107 WelsDestroySVCEncoder(s->encoder);
109 av_log(avctx, AV_LOG_WARNING, "%d frames skipped\n", s->skipped);
113 static av_cold int svc_encode_init(AVCodecContext *avctx)
115 SVCContext *s = avctx->priv_data;
116 SEncParamExt param = { 0 };
119 WelsTraceCallback callback_function;
120 AVCPBProperties *props;
122 if ((err = ff_libopenh264_check_version(avctx)) < 0)
125 if (WelsCreateSVCEncoder(&s->encoder)) {
126 av_log(avctx, AV_LOG_ERROR, "Unable to create encoder\n");
127 return AVERROR_UNKNOWN;
130 // Pass all libopenh264 messages to our callback, to allow ourselves to filter them.
131 log_level = WELS_LOG_DETAIL;
132 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_LEVEL, &log_level);
134 // Set the logging callback function to one that uses av_log() (see implementation above).
135 callback_function = (WelsTraceCallback) ff_libopenh264_trace_callback;
136 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK, &callback_function);
138 // Set the AVCodecContext as the libopenh264 callback context so that it can be passed to av_log().
139 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK_CONTEXT, &avctx);
141 (*s->encoder)->GetDefaultParams(s->encoder, ¶m);
143 #if FF_API_CODER_TYPE
144 FF_DISABLE_DEPRECATION_WARNINGS
146 s->cabac = avctx->coder_type == FF_CODER_TYPE_AC;
147 FF_ENABLE_DEPRECATION_WARNINGS
150 param.fMaxFrameRate = 1/av_q2d(avctx->time_base);
151 param.iPicWidth = avctx->width;
152 param.iPicHeight = avctx->height;
153 param.iTargetBitrate = avctx->bit_rate > 0 ? avctx->bit_rate : TARGET_BITRATE_DEFAULT;
154 param.iMaxBitrate = FFMAX(avctx->rc_max_rate, avctx->bit_rate);
155 param.iRCMode = s->rc_mode;
156 if (avctx->qmax >= 0)
157 param.iMaxQp = av_clip(avctx->qmax, 1, 51);
158 if (avctx->qmin >= 0)
159 param.iMinQp = av_clip(avctx->qmin, 1, param.iMaxQp);
160 param.iTemporalLayerNum = 1;
161 param.iSpatialLayerNum = 1;
162 param.bEnableDenoise = 0;
163 param.bEnableBackgroundDetection = 1;
164 param.bEnableAdaptiveQuant = 1;
165 param.bEnableFrameSkip = s->skip_frames;
166 param.bEnableLongTermReference = 0;
167 param.iLtrMarkPeriod = 30;
168 if (avctx->gop_size >= 0)
169 param.uiIntraPeriod = avctx->gop_size;
170 #if OPENH264_VER_AT_LEAST(1, 4)
171 param.eSpsPpsIdStrategy = CONSTANT_ID;
173 param.bEnableSpsPpsIdAddition = 0;
175 param.bPrefixNalAddingCtrl = 0;
176 param.iLoopFilterDisableIdc = !s->loopfilter;
177 param.iEntropyCodingModeFlag = 0;
178 param.iMultipleThreadIdc = avctx->thread_count;
179 if (s->profile && !strcmp(s->profile, "main"))
180 param.iEntropyCodingModeFlag = 1;
181 else if (!s->profile && s->cabac)
182 param.iEntropyCodingModeFlag = 1;
184 param.sSpatialLayers[0].iVideoWidth = param.iPicWidth;
185 param.sSpatialLayers[0].iVideoHeight = param.iPicHeight;
186 param.sSpatialLayers[0].fFrameRate = param.fMaxFrameRate;
187 param.sSpatialLayers[0].iSpatialBitrate = param.iTargetBitrate;
188 param.sSpatialLayers[0].iMaxSpatialBitrate = param.iMaxBitrate;
190 #if OPENH264_VER_AT_LEAST(1, 7)
191 if (avctx->sample_aspect_ratio.num && avctx->sample_aspect_ratio.den) {
193 static const AVRational sar_idc[] = {
194 { 0, 0 }, // Unspecified (never written here).
195 { 1, 1 }, { 12, 11 }, { 10, 11 }, { 16, 11 },
196 { 40, 33 }, { 24, 11 }, { 20, 11 }, { 32, 11 },
197 { 80, 33 }, { 18, 11 }, { 15, 11 }, { 64, 33 },
198 { 160, 99 }, // Last 3 are unknown to openh264: { 4, 3 }, { 3, 2 }, { 2, 1 },
200 static const ESampleAspectRatio asp_idc[] = {
202 ASP_1x1, ASP_12x11, ASP_10x11, ASP_16x11,
203 ASP_40x33, ASP_24x11, ASP_20x11, ASP_32x11,
204 ASP_80x33, ASP_18x11, ASP_15x11, ASP_64x33,
209 av_reduce(&num, &den, avctx->sample_aspect_ratio.num,
210 avctx->sample_aspect_ratio.den, 65535);
212 for (i = 1; i < FF_ARRAY_ELEMS(sar_idc); i++) {
213 if (num == sar_idc[i].num &&
214 den == sar_idc[i].den)
217 if (i == FF_ARRAY_ELEMS(sar_idc)) {
218 param.sSpatialLayers[0].eAspectRatio = ASP_EXT_SAR;
219 param.sSpatialLayers[0].sAspectRatioExtWidth = num;
220 param.sSpatialLayers[0].sAspectRatioExtHeight = den;
222 param.sSpatialLayers[0].eAspectRatio = asp_idc[i];
224 param.sSpatialLayers[0].bAspectRatioPresent = true;
226 param.sSpatialLayers[0].bAspectRatioPresent = false;
230 if ((avctx->slices > 1) && (s->max_nal_size)) {
231 av_log(avctx, AV_LOG_ERROR,
232 "Invalid combination -slices %d and -max_nal_size %d.\n",
233 avctx->slices, s->max_nal_size);
234 return AVERROR(EINVAL);
237 if (avctx->slices > 1)
238 s->slice_mode = SM_FIXEDSLCNUM_SLICE;
241 s->slice_mode = SM_SIZELIMITED_SLICE;
243 #if OPENH264_VER_AT_LEAST(1, 6)
244 param.sSpatialLayers[0].sSliceArgument.uiSliceMode = s->slice_mode;
245 param.sSpatialLayers[0].sSliceArgument.uiSliceNum = avctx->slices;
247 param.sSpatialLayers[0].sSliceCfg.uiSliceMode = s->slice_mode;
248 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceNum = avctx->slices;
250 if (avctx->slices == 0 && s->slice_mode == SM_FIXEDSLCNUM_SLICE)
251 av_log(avctx, AV_LOG_WARNING, "Slice count will be set automatically\n");
253 if (s->slice_mode == SM_SIZELIMITED_SLICE) {
254 if (s->max_nal_size) {
255 param.uiMaxNalSize = s->max_nal_size;
256 #if OPENH264_VER_AT_LEAST(1, 6)
257 param.sSpatialLayers[0].sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
259 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
262 av_log(avctx, AV_LOG_ERROR, "Invalid -max_nal_size, "
263 "specify a valid max_nal_size to use -slice_mode dyn\n");
264 return AVERROR(EINVAL);
268 if ((*s->encoder)->InitializeExt(s->encoder, ¶m) != cmResultSuccess) {
269 av_log(avctx, AV_LOG_ERROR, "Initialize failed\n");
270 return AVERROR_UNKNOWN;
273 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
274 SFrameBSInfo fbi = { 0 };
276 (*s->encoder)->EncodeParameterSets(s->encoder, &fbi);
277 for (i = 0; i < fbi.sLayerInfo[0].iNalCount; i++)
278 size += fbi.sLayerInfo[0].pNalLengthInByte[i];
279 avctx->extradata = av_mallocz(size + AV_INPUT_BUFFER_PADDING_SIZE);
280 if (!avctx->extradata)
281 return AVERROR(ENOMEM);
282 avctx->extradata_size = size;
283 memcpy(avctx->extradata, fbi.sLayerInfo[0].pBsBuf, size);
286 props = ff_add_cpb_side_data(avctx);
288 return AVERROR(ENOMEM);
289 props->max_bitrate = param.iMaxBitrate;
290 props->avg_bitrate = param.iTargetBitrate;
295 static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
296 const AVFrame *frame, int *got_packet)
298 SVCContext *s = avctx->priv_data;
299 SFrameBSInfo fbi = { 0 };
302 SSourcePicture sp = { 0 };
303 int size = 0, layer, first_layer = 0;
304 int layer_size[MAX_LAYER_NUM_OF_FRAME] = { 0 };
306 sp.iColorFormat = videoFormatI420;
307 for (i = 0; i < 3; i++) {
308 sp.iStride[i] = frame->linesize[i];
309 sp.pData[i] = frame->data[i];
311 sp.iPicWidth = avctx->width;
312 sp.iPicHeight = avctx->height;
314 if (frame->pict_type == AV_PICTURE_TYPE_I) {
315 (*s->encoder)->ForceIntraFrame(s->encoder, true);
318 encoded = (*s->encoder)->EncodeFrame(s->encoder, &sp, &fbi);
319 if (encoded != cmResultSuccess) {
320 av_log(avctx, AV_LOG_ERROR, "EncodeFrame failed\n");
321 return AVERROR_UNKNOWN;
323 if (fbi.eFrameType == videoFrameTypeSkip) {
325 av_log(avctx, AV_LOG_DEBUG, "frame skipped\n");
329 // Normal frames are returned with one single layer, while IDR
330 // frames have two layers, where the first layer contains the SPS/PPS.
331 // If using global headers, don't include the SPS/PPS in the returned
332 // packet - thus, only return one layer.
333 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
334 first_layer = fbi.iLayerNum - 1;
336 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
337 for (i = 0; i < fbi.sLayerInfo[layer].iNalCount; i++)
338 layer_size[layer] += fbi.sLayerInfo[layer].pNalLengthInByte[i];
339 size += layer_size[layer];
341 av_log(avctx, AV_LOG_DEBUG, "%d slices\n", fbi.sLayerInfo[fbi.iLayerNum - 1].iNalCount);
343 if ((ret = ff_alloc_packet2(avctx, avpkt, size, size))) {
344 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
348 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
349 memcpy(avpkt->data + size, fbi.sLayerInfo[layer].pBsBuf, layer_size[layer]);
350 size += layer_size[layer];
352 avpkt->pts = frame->pts;
353 if (fbi.eFrameType == videoFrameTypeIDR)
354 avpkt->flags |= AV_PKT_FLAG_KEY;
359 static const AVCodecDefault svc_enc_defaults[] = {
367 AVCodec ff_libopenh264_encoder = {
368 .name = "libopenh264",
369 .long_name = NULL_IF_CONFIG_SMALL("OpenH264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
370 .type = AVMEDIA_TYPE_VIDEO,
371 .id = AV_CODEC_ID_H264,
372 .priv_data_size = sizeof(SVCContext),
373 .init = svc_encode_init,
374 .encode2 = svc_encode_frame,
375 .close = svc_encode_close,
376 .capabilities = AV_CODEC_CAP_AUTO_THREADS,
377 .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
378 .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P,
380 .defaults = svc_enc_defaults,
381 .priv_class = &class,
382 .wrapper_name = "libopenh264",