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 typedef struct SVCContext {
41 const AVClass *av_class;
52 #define OFFSET(x) offsetof(SVCContext, x)
53 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
54 static const AVOption options[] = {
55 #if OPENH264_VER_AT_LEAST(1, 6)
56 { "slice_mode", "set slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_FIXEDSLCNUM_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
58 { "slice_mode", "set slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_AUTO_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
60 { "fixed", "a fixed number of slices", 0, AV_OPT_TYPE_CONST, { .i64 = SM_FIXEDSLCNUM_SLICE }, 0, 0, VE, "slice_mode" },
61 #if OPENH264_VER_AT_LEAST(1, 6)
62 { "dyn", "Size limited (compatibility name)", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
63 { "sizelimited", "Size limited", 0, AV_OPT_TYPE_CONST, { .i64 = SM_SIZELIMITED_SLICE }, 0, 0, VE, "slice_mode" },
65 { "rowmb", "one slice per row of macroblocks", 0, AV_OPT_TYPE_CONST, { .i64 = SM_ROWMB_SLICE }, 0, 0, VE, "slice_mode" },
66 { "auto", "automatic number of slices according to number of threads", 0, AV_OPT_TYPE_CONST, { .i64 = SM_AUTO_SLICE }, 0, 0, VE, "slice_mode" },
67 { "dyn", "Dynamic slicing", 0, AV_OPT_TYPE_CONST, { .i64 = SM_DYN_SLICE }, 0, 0, VE, "slice_mode" },
69 { "loopfilter", "enable loop filter", OFFSET(loopfilter), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE },
70 { "profile", "set profile restrictions", OFFSET(profile), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
71 { "max_nal_size", "set maximum NAL size in bytes", OFFSET(max_nal_size), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE },
72 { "allow_skip_frames", "allow skipping frames to hit the target bitrate", OFFSET(skip_frames), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
73 { "cabac", "Enable cabac", OFFSET(cabac), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
77 static const AVClass class = {
78 "libopenh264enc", av_default_item_name, options, LIBAVUTIL_VERSION_INT
81 static av_cold int svc_encode_close(AVCodecContext *avctx)
83 SVCContext *s = avctx->priv_data;
86 WelsDestroySVCEncoder(s->encoder);
88 av_log(avctx, AV_LOG_WARNING, "%d frames skipped\n", s->skipped);
92 static av_cold int svc_encode_init(AVCodecContext *avctx)
94 SVCContext *s = avctx->priv_data;
95 SEncParamExt param = { 0 };
98 WelsTraceCallback callback_function;
99 AVCPBProperties *props;
101 if ((err = ff_libopenh264_check_version(avctx)) < 0)
103 // Use a default error for multiple error paths below
104 err = AVERROR_UNKNOWN;
106 if (WelsCreateSVCEncoder(&s->encoder)) {
107 av_log(avctx, AV_LOG_ERROR, "Unable to create encoder\n");
108 return AVERROR_UNKNOWN;
111 // Pass all libopenh264 messages to our callback, to allow ourselves to filter them.
112 log_level = WELS_LOG_DETAIL;
113 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_LEVEL, &log_level);
115 // Set the logging callback function to one that uses av_log() (see implementation above).
116 callback_function = (WelsTraceCallback) ff_libopenh264_trace_callback;
117 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK, (void *)&callback_function);
119 // Set the AVCodecContext as the libopenh264 callback context so that it can be passed to av_log().
120 (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK_CONTEXT, (void *)&avctx);
122 (*s->encoder)->GetDefaultParams(s->encoder, ¶m);
124 #if FF_API_CODER_TYPE
125 FF_DISABLE_DEPRECATION_WARNINGS
127 s->cabac = avctx->coder_type == FF_CODER_TYPE_AC;
128 FF_ENABLE_DEPRECATION_WARNINGS
131 param.fMaxFrameRate = 1/av_q2d(avctx->time_base);
132 param.iPicWidth = avctx->width;
133 param.iPicHeight = avctx->height;
134 param.iTargetBitrate = avctx->bit_rate;
135 param.iMaxBitrate = FFMAX(avctx->rc_max_rate, avctx->bit_rate);
136 param.iRCMode = RC_QUALITY_MODE;
137 param.iTemporalLayerNum = 1;
138 param.iSpatialLayerNum = 1;
139 param.bEnableDenoise = 0;
140 param.bEnableBackgroundDetection = 1;
141 param.bEnableAdaptiveQuant = 1;
142 param.bEnableFrameSkip = s->skip_frames;
143 param.bEnableLongTermReference = 0;
144 param.iLtrMarkPeriod = 30;
145 param.uiIntraPeriod = avctx->gop_size;
146 #if OPENH264_VER_AT_LEAST(1, 4)
147 param.eSpsPpsIdStrategy = CONSTANT_ID;
149 param.bEnableSpsPpsIdAddition = 0;
151 param.bPrefixNalAddingCtrl = 0;
152 param.iLoopFilterDisableIdc = !s->loopfilter;
153 param.iEntropyCodingModeFlag = 0;
154 param.iMultipleThreadIdc = avctx->thread_count;
155 if (s->profile && !strcmp(s->profile, "main"))
156 param.iEntropyCodingModeFlag = 1;
157 else if (!s->profile && s->cabac)
158 param.iEntropyCodingModeFlag = 1;
160 param.sSpatialLayers[0].iVideoWidth = param.iPicWidth;
161 param.sSpatialLayers[0].iVideoHeight = param.iPicHeight;
162 param.sSpatialLayers[0].fFrameRate = param.fMaxFrameRate;
163 param.sSpatialLayers[0].iSpatialBitrate = param.iTargetBitrate;
164 param.sSpatialLayers[0].iMaxSpatialBitrate = param.iMaxBitrate;
166 if ((avctx->slices > 1) && (s->max_nal_size)) {
167 av_log(avctx, AV_LOG_ERROR,
168 "Invalid combination -slices %d and -max_nal_size %d.\n",
169 avctx->slices, s->max_nal_size);
173 if (avctx->slices > 1)
174 s->slice_mode = SM_FIXEDSLCNUM_SLICE;
177 s->slice_mode = SM_SIZELIMITED_SLICE;
179 #if OPENH264_VER_AT_LEAST(1, 6)
180 param.sSpatialLayers[0].sSliceArgument.uiSliceMode = s->slice_mode;
181 param.sSpatialLayers[0].sSliceArgument.uiSliceNum = avctx->slices;
183 param.sSpatialLayers[0].sSliceCfg.uiSliceMode = s->slice_mode;
184 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceNum = avctx->slices;
187 if (s->slice_mode == SM_SIZELIMITED_SLICE) {
188 if (s->max_nal_size){
189 param.uiMaxNalSize = s->max_nal_size;
190 #if OPENH264_VER_AT_LEAST(1, 6)
191 param.sSpatialLayers[0].sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
193 param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
196 av_log(avctx, AV_LOG_ERROR, "Invalid -max_nal_size, "
197 "specify a valid max_nal_size to use -slice_mode dyn\n");
202 if ((*s->encoder)->InitializeExt(s->encoder, ¶m) != cmResultSuccess) {
203 av_log(avctx, AV_LOG_ERROR, "Initialize failed\n");
207 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
208 SFrameBSInfo fbi = { 0 };
210 (*s->encoder)->EncodeParameterSets(s->encoder, &fbi);
211 for (i = 0; i < fbi.sLayerInfo[0].iNalCount; i++)
212 size += fbi.sLayerInfo[0].pNalLengthInByte[i];
213 avctx->extradata = av_mallocz(size + AV_INPUT_BUFFER_PADDING_SIZE);
214 if (!avctx->extradata) {
215 err = AVERROR(ENOMEM);
218 avctx->extradata_size = size;
219 memcpy(avctx->extradata, fbi.sLayerInfo[0].pBsBuf, size);
222 props = ff_add_cpb_side_data(avctx);
224 err = AVERROR(ENOMEM);
227 props->max_bitrate = param.iMaxBitrate;
228 props->avg_bitrate = param.iTargetBitrate;
233 svc_encode_close(avctx);
237 static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
238 const AVFrame *frame, int *got_packet)
240 SVCContext *s = avctx->priv_data;
241 SFrameBSInfo fbi = { 0 };
244 SSourcePicture sp = { 0 };
245 int size = 0, layer, first_layer = 0;
246 int layer_size[MAX_LAYER_NUM_OF_FRAME] = { 0 };
248 sp.iColorFormat = videoFormatI420;
249 for (i = 0; i < 3; i++) {
250 sp.iStride[i] = frame->linesize[i];
251 sp.pData[i] = frame->data[i];
253 sp.iPicWidth = avctx->width;
254 sp.iPicHeight = avctx->height;
256 encoded = (*s->encoder)->EncodeFrame(s->encoder, &sp, &fbi);
257 if (encoded != cmResultSuccess) {
258 av_log(avctx, AV_LOG_ERROR, "EncodeFrame failed\n");
259 return AVERROR_UNKNOWN;
261 if (fbi.eFrameType == videoFrameTypeSkip) {
263 av_log(avctx, AV_LOG_DEBUG, "frame skipped\n");
267 // Normal frames are returned with one single layer, while IDR
268 // frames have two layers, where the first layer contains the SPS/PPS.
269 // If using global headers, don't include the SPS/PPS in the returned
270 // packet - thus, only return one layer.
271 if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
272 first_layer = fbi.iLayerNum - 1;
274 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
275 for (i = 0; i < fbi.sLayerInfo[layer].iNalCount; i++)
276 layer_size[layer] += fbi.sLayerInfo[layer].pNalLengthInByte[i];
277 size += layer_size[layer];
279 av_log(avctx, AV_LOG_DEBUG, "%d slices\n", fbi.sLayerInfo[fbi.iLayerNum - 1].iNalCount);
281 if ((ret = ff_alloc_packet2(avctx, avpkt, size, size))) {
282 av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
286 for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
287 memcpy(avpkt->data + size, fbi.sLayerInfo[layer].pBsBuf, layer_size[layer]);
288 size += layer_size[layer];
290 avpkt->pts = frame->pts;
291 if (fbi.eFrameType == videoFrameTypeIDR)
292 avpkt->flags |= AV_PKT_FLAG_KEY;
297 AVCodec ff_libopenh264_encoder = {
298 .name = "libopenh264",
299 .long_name = NULL_IF_CONFIG_SMALL("OpenH264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
300 .type = AVMEDIA_TYPE_VIDEO,
301 .id = AV_CODEC_ID_H264,
302 .priv_data_size = sizeof(SVCContext),
303 .init = svc_encode_init,
304 .encode2 = svc_encode_frame,
305 .close = svc_encode_close,
306 .capabilities = AV_CODEC_CAP_AUTO_THREADS,
307 .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P,
309 .priv_class = &class,