libopenh264enc.c 12.5 KB
Newer Older
1 2 3 4
/*
 * OpenH264 video encoder
 * Copyright (C) 2014 Martin Storsjo
 *
5
 * This file is part of FFmpeg.
6
 *
7
 * FFmpeg is free software; you can redistribute it and/or
8 9 10 11
 * 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.
 *
12
 * FFmpeg is distributed in the hope that it will be useful,
13 14 15 16 17
 * 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
18
 * License along with FFmpeg; if not, write to the Free Software
19 20 21 22 23 24 25 26 27
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include <wels/codec_api.h>
#include <wels/codec_ver.h>

#include "libavutil/attributes.h"
#include "libavutil/common.h"
#include "libavutil/opt.h"
28
#include "libavutil/internal.h"
29 30 31 32 33 34 35 36 37 38 39 40
#include "libavutil/intreadwrite.h"
#include "libavutil/mathematics.h"

#include "avcodec.h"
#include "internal.h"

typedef struct SVCContext {
    const AVClass *av_class;
    ISVCEncoder *encoder;
    int slice_mode;
    int loopfilter;
    char *profile;
41
    int max_nal_size;
42
    int skip_frames;
43
    int skipped;
44
    int cabac;
45 46
} SVCContext;

47 48 49 50
#define OPENH264_VER_AT_LEAST(maj, min) \
    ((OPENH264_MAJOR  > (maj)) || \
     (OPENH264_MAJOR == (maj) && OPENH264_MINOR >= (min)))

51 52 53
#define OFFSET(x) offsetof(SVCContext, x)
#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
static const AVOption options[] = {
54 55 56 57
    { "slice_mode", "set slice mode", OFFSET(slice_mode), AV_OPT_TYPE_INT, { .i64 = SM_AUTO_SLICE }, SM_SINGLE_SLICE, SM_RESERVED, VE, "slice_mode" },
        { "fixed", "a fixed number of slices", 0, AV_OPT_TYPE_CONST, { .i64 = SM_FIXEDSLCNUM_SLICE }, 0, 0, VE, "slice_mode" },
        { "rowmb", "one slice per row of macroblocks", 0, AV_OPT_TYPE_CONST, { .i64 = SM_ROWMB_SLICE }, 0, 0, VE, "slice_mode" },
        { "auto", "automatic number of slices according to number of threads", 0, AV_OPT_TYPE_CONST, { .i64 = SM_AUTO_SLICE }, 0, 0, VE, "slice_mode" },
58
        { "dyn", "Dynamic slicing", 0, AV_OPT_TYPE_CONST, { .i64 = SM_DYN_SLICE }, 0, 0, VE, "slice_mode" },
59
    { "loopfilter", "enable loop filter", OFFSET(loopfilter), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, VE },
60
    { "profile", "set profile restrictions", OFFSET(profile), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, VE },
61 62
    { "max_nal_size", "set maximum NAL size in bytes", OFFSET(max_nal_size), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, VE },
    { "allow_skip_frames", "allow skipping frames to hit the target bitrate", OFFSET(skip_frames), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE },
63
    { "cabac", "Enable cabac", OFFSET(cabac), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, VE },
64 65 66 67 68 69 70
    { NULL }
};

static const AVClass class = {
    "libopenh264enc", av_default_item_name, options, LIBAVUTIL_VERSION_INT
};

71
// Convert libopenh264 log level to equivalent ffmpeg log level.
72
static int libopenh264_to_ffmpeg_log_level(int libopenh264_log_level)
73
{
74
    if      (libopenh264_log_level >= WELS_LOG_DETAIL)  return AV_LOG_TRACE;
75
    else if (libopenh264_log_level >= WELS_LOG_DEBUG)   return AV_LOG_DEBUG;
76
    else if (libopenh264_log_level >= WELS_LOG_INFO)    return AV_LOG_VERBOSE;
77 78 79
    else if (libopenh264_log_level >= WELS_LOG_WARNING) return AV_LOG_WARNING;
    else if (libopenh264_log_level >= WELS_LOG_ERROR)   return AV_LOG_ERROR;
    else                                                return AV_LOG_QUIET;
80 81 82 83 84 85 86 87
}

// This function will be provided to the libopenh264 library.  The function will be called
// when libopenh264 wants to log a message (error, warning, info, etc.).  The signature for
// this function (defined in .../codec/api/svc/codec_api.h) is:
//
//        typedef void (*WelsTraceCallback) (void* ctx, int level, const char* string);

88
static void libopenh264_trace_callback(void *ctx, int level, const char *msg)
89 90
{
    // The message will be logged only if the requested EQUIVALENT ffmpeg log level is
91
    // less than or equal to the current ffmpeg log level.
92 93
    int equiv_ffmpeg_log_level = libopenh264_to_ffmpeg_log_level(level);
    av_log(ctx, equiv_ffmpeg_log_level, "%s\n", msg);
94 95
}

96 97 98 99 100 101
static av_cold int svc_encode_close(AVCodecContext *avctx)
{
    SVCContext *s = avctx->priv_data;

    if (s->encoder)
        WelsDestroySVCEncoder(s->encoder);
102 103
    if (s->skipped > 0)
        av_log(avctx, AV_LOG_WARNING, "%d frames skipped\n", s->skipped);
104 105 106 107 108 109 110 111
    return 0;
}

static av_cold int svc_encode_init(AVCodecContext *avctx)
{
    SVCContext *s = avctx->priv_data;
    SEncParamExt param = { 0 };
    int err = AVERROR_UNKNOWN;
112
    int log_level;
113
    WelsTraceCallback callback_function;
114
    AVCPBProperties *props;
115 116 117 118 119

    // Mingw GCC < 4.7 on x86_32 uses an incorrect/buggy ABI for the WelsGetCodecVersion
    // function (for functions returning larger structs), thus skip the check in those
    // configurations.
#if !defined(_WIN32) || !defined(__GNUC__) || !ARCH_X86_32 || AV_GCC_VERSION_AT_LEAST(4, 7)
120
    OpenH264Version libver = WelsGetCodecVersion();
121 122 123 124 125 126 127 128 129 130 131
    if (memcmp(&libver, &g_stCodecVersion, sizeof(libver))) {
        av_log(avctx, AV_LOG_ERROR, "Incorrect library version loaded\n");
        return AVERROR(EINVAL);
    }
#endif

    if (WelsCreateSVCEncoder(&s->encoder)) {
        av_log(avctx, AV_LOG_ERROR, "Unable to create encoder\n");
        return AVERROR_UNKNOWN;
    }

132 133 134
    // Pass all libopenh264 messages to our callback, to allow ourselves to filter them.
    log_level = WELS_LOG_DETAIL;
    (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_LEVEL, &log_level);
135 136 137

    // Set the logging callback function to one that uses av_log() (see implementation above).
    callback_function = (WelsTraceCallback) libopenh264_trace_callback;
138
    (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK, (void *)&callback_function);
139 140

    // Set the AVCodecContext as the libopenh264 callback context so that it can be passed to av_log().
141
    (*s->encoder)->SetOption(s->encoder, ENCODER_OPTION_TRACE_CALLBACK_CONTEXT, (void *)&avctx);
142

143 144
    (*s->encoder)->GetDefaultParams(s->encoder, &param);

145 146 147 148 149 150 151
#if FF_API_CODER_TYPE
FF_DISABLE_DEPRECATION_WARNINGS
    if (!s->cabac)
        s->cabac = avctx->coder_type == FF_CODER_TYPE_AC;
FF_ENABLE_DEPRECATION_WARNINGS
#endif

152
    param.fMaxFrameRate              = 1/av_q2d(avctx->time_base);
153 154 155 156 157 158 159 160 161 162
    param.iPicWidth                  = avctx->width;
    param.iPicHeight                 = avctx->height;
    param.iTargetBitrate             = avctx->bit_rate;
    param.iMaxBitrate                = FFMAX(avctx->rc_max_rate, avctx->bit_rate);
    param.iRCMode                    = RC_QUALITY_MODE;
    param.iTemporalLayerNum          = 1;
    param.iSpatialLayerNum           = 1;
    param.bEnableDenoise             = 0;
    param.bEnableBackgroundDetection = 1;
    param.bEnableAdaptiveQuant       = 1;
163
    param.bEnableFrameSkip           = s->skip_frames;
164 165 166
    param.bEnableLongTermReference   = 0;
    param.iLtrMarkPeriod             = 30;
    param.uiIntraPeriod              = avctx->gop_size;
167 168 169
#if OPENH264_VER_AT_LEAST(1, 4)
    param.eSpsPpsIdStrategy          = CONSTANT_ID;
#else
170
    param.bEnableSpsPpsIdAddition    = 0;
171
#endif
172 173 174 175 176 177
    param.bPrefixNalAddingCtrl       = 0;
    param.iLoopFilterDisableIdc      = !s->loopfilter;
    param.iEntropyCodingModeFlag     = 0;
    param.iMultipleThreadIdc         = avctx->thread_count;
    if (s->profile && !strcmp(s->profile, "main"))
        param.iEntropyCodingModeFlag = 1;
178
    else if (!s->profile && s->cabac)
179 180 181 182 183 184 185 186
        param.iEntropyCodingModeFlag = 1;

    param.sSpatialLayers[0].iVideoWidth         = param.iPicWidth;
    param.sSpatialLayers[0].iVideoHeight        = param.iPicHeight;
    param.sSpatialLayers[0].fFrameRate          = param.fMaxFrameRate;
    param.sSpatialLayers[0].iSpatialBitrate     = param.iTargetBitrate;
    param.sSpatialLayers[0].iMaxSpatialBitrate  = param.iMaxBitrate;

187 188 189 190 191
    if ((avctx->slices > 1) && (s->max_nal_size)){
        av_log(avctx,AV_LOG_ERROR,"Invalid combination -slices %d and -max_nal_size %d.\n",avctx->slices,s->max_nal_size);
        goto fail;
    }

192 193
    if (avctx->slices > 1)
        s->slice_mode = SM_FIXEDSLCNUM_SLICE;
194 195 196 197

    if (s->max_nal_size)
        s->slice_mode = SM_DYN_SLICE;

198 199 200
    param.sSpatialLayers[0].sSliceCfg.uiSliceMode               = s->slice_mode;
    param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceNum = avctx->slices;

201 202 203 204 205
    if (s->slice_mode == SM_DYN_SLICE) {
        if (s->max_nal_size){
            param.uiMaxNalSize = s->max_nal_size;
            param.sSpatialLayers[0].sSliceCfg.sSliceArgument.uiSliceSizeConstraint = s->max_nal_size;
        } else {
206 207 208
            av_log(avctx, AV_LOG_ERROR, "Invalid -max_nal_size, "
                   "specify a valid max_nal_size to use -slice_mode dyn\n");
            goto fail;
209 210 211
        }
    }

212 213 214 215 216
    if ((*s->encoder)->InitializeExt(s->encoder, &param) != cmResultSuccess) {
        av_log(avctx, AV_LOG_ERROR, "Initialize failed\n");
        goto fail;
    }

217
    if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
218 219 220 221 222
        SFrameBSInfo fbi = { 0 };
        int i, size = 0;
        (*s->encoder)->EncodeParameterSets(s->encoder, &fbi);
        for (i = 0; i < fbi.sLayerInfo[0].iNalCount; i++)
            size += fbi.sLayerInfo[0].pNalLengthInByte[i];
223
        avctx->extradata = av_mallocz(size + AV_INPUT_BUFFER_PADDING_SIZE);
224 225 226 227 228 229 230 231
        if (!avctx->extradata) {
            err = AVERROR(ENOMEM);
            goto fail;
        }
        avctx->extradata_size = size;
        memcpy(avctx->extradata, fbi.sLayerInfo[0].pBsBuf, size);
    }

232 233 234 235 236 237 238 239
    props = ff_add_cpb_side_data(avctx);
    if (!props) {
        err = AVERROR(ENOMEM);
        goto fail;
    }
    props->max_bitrate = param.iMaxBitrate;
    props->avg_bitrate = param.iTargetBitrate;

240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
    return 0;

fail:
    svc_encode_close(avctx);
    return err;
}

static int svc_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                            const AVFrame *frame, int *got_packet)
{
    SVCContext *s = avctx->priv_data;
    SFrameBSInfo fbi = { 0 };
    int i, ret;
    int encoded;
    SSourcePicture sp = { 0 };
    int size = 0, layer, first_layer = 0;
    int layer_size[MAX_LAYER_NUM_OF_FRAME] = { 0 };

    sp.iColorFormat = videoFormatI420;
    for (i = 0; i < 3; i++) {
        sp.iStride[i] = frame->linesize[i];
        sp.pData[i]   = frame->data[i];
    }
    sp.iPicWidth  = avctx->width;
    sp.iPicHeight = avctx->height;

    encoded = (*s->encoder)->EncodeFrame(s->encoder, &sp, &fbi);
    if (encoded != cmResultSuccess) {
        av_log(avctx, AV_LOG_ERROR, "EncodeFrame failed\n");
        return AVERROR_UNKNOWN;
    }
    if (fbi.eFrameType == videoFrameTypeSkip) {
272
        s->skipped++;
273 274 275 276
        av_log(avctx, AV_LOG_DEBUG, "frame skipped\n");
        return 0;
    }
    first_layer = 0;
277
    // Normal frames are returned with one single layer, while IDR
278 279 280
    // frames have two layers, where the first layer contains the SPS/PPS.
    // If using global headers, don't include the SPS/PPS in the returned
    // packet - thus, only return one layer.
281
    if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER)
282 283 284 285 286 287 288
        first_layer = fbi.iLayerNum - 1;

    for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
        for (i = 0; i < fbi.sLayerInfo[layer].iNalCount; i++)
            layer_size[layer] += fbi.sLayerInfo[layer].pNalLengthInByte[i];
        size += layer_size[layer];
    }
289
    av_log(avctx, AV_LOG_DEBUG, "%d slices\n", fbi.sLayerInfo[fbi.iLayerNum - 1].iNalCount);
290

291
    if ((ret = ff_alloc_packet2(avctx, avpkt, size, size))) {
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
        av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
        return ret;
    }
    size = 0;
    for (layer = first_layer; layer < fbi.iLayerNum; layer++) {
        memcpy(avpkt->data + size, fbi.sLayerInfo[layer].pBsBuf, layer_size[layer]);
        size += layer_size[layer];
    }
    avpkt->pts = frame->pts;
    if (fbi.eFrameType == videoFrameTypeIDR)
        avpkt->flags |= AV_PKT_FLAG_KEY;
    *got_packet = 1;
    return 0;
}

AVCodec ff_libopenh264_encoder = {
    .name           = "libopenh264",
309
    .long_name      = NULL_IF_CONFIG_SMALL("OpenH264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
310 311 312 313 314 315
    .type           = AVMEDIA_TYPE_VIDEO,
    .id             = AV_CODEC_ID_H264,
    .priv_data_size = sizeof(SVCContext),
    .init           = svc_encode_init,
    .encode2        = svc_encode_frame,
    .close          = svc_encode_close,
316
    .capabilities   = AV_CODEC_CAP_AUTO_THREADS,
317 318
    .pix_fmts       = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P,
                                                    AV_PIX_FMT_NONE },
319 320
    .priv_class     = &class,
};