I have started learning ffmpeg few weaks ago. At the moment I am able to transcode any video to mp4 using h264/AVC codec. The main scheme is something like that:
-open input
-demux
-decode
-encode
-mux
The actual code is below:
#include <iostream>
#include <math.h>
extern "C" {
#
ifndef __STDC_CONSTANT_MACROS# undef main /* Prevents SDL from overriding main() */ # define __STDC_CONSTANT_MACROS# endif
# pragma comment(lib, "avcodec.lib")# pragma comment(lib, "avformat.lib")# pragma comment(lib, "swscale.lib")# pragma comment(lib, "avutil.lib")
#include <libavcodec\avcodec.h>
#include <libavformat\avformat.h>
#include <libswscale\swscale.h>
#include <libavutil\mem.h>
#include <libavutil/opt.h>
#include <libavutil\channel_layout.h>
#include <libavutil\common.h>
#include <libavutil\imgutils.h>
#include <libavutil\mathematics.h>
#include <libavutil\samplefmt.h>
}
using namespace std;
void open_video(AVFormatContext * oc, AVCodec * codec, AVStream * st) {
int ret;
AVCodecContext * c;
c = st - > codec;
/*open codec */
cout << "probably starts here" << endl;
ret = avcodec_open2(c, codec, NULL);
cout << "and ends here" << endl;
if (ret < 0) {
cout << ("Could not open video codec") << endl;
}
}
/*This function will add a new stream to our file.
#param
oc -> Format context that the new stream will be added.
codec -> codec of the stream, this will be passed.
codec_id ->
chWidth->
chHeight->
*/
AVStream * addStream(AVFormatContext * oc, AVCodec ** codec, enum AVCodecID codec_id, int chWidth, int chHeight, int fps) {
AVCodecContext * c;
AVStream * st;
//find encoder of the stream, it passes this information to #codec, later on
//it will be used in encoding the video # avcodec_encode_video2 in loop.
* codec = avcodec_find_encoder(AV_CODEC_ID_H264);
if (( * codec) == NULL)
cout << "ERROR CAN NOT FIND ENCODER! ERROR! ERROR! AVCODEC_FIND_ENCODER FAILED !!!1 "
"" << endl;
if (!( * codec))
printf("Could not find encoder for ' %s ' ", avcodec_get_name(codec_id));
//create a new stream with the found codec inside oc(AVFormatContext).
st = avformat_new_stream(oc, * codec);
if (!st)
cout << " Cannot allocate stream " << endl;
//Setting the stream id.
//Since, there can be other streams in this AVFormatContext,
//we should find the first non used index. And this is oc->nb_streams(number of streams) - 1
st - > id = oc - > nb_streams - 1;
c = st - > codec;
//setting the stream's codec's properties.
c - > codec_id = codec_id;
c - > bit_rate = 4000000;
c - > width = chWidth;
c - > height = chHeight;
c - > time_base.den = fps;
//fps;
c - > time_base.num = 1;
c - > gop_size = 12;
c - > pix_fmt = AV_PIX_FMT_YUV420P;
if (c - > codec_id == AV_CODEC_ID_MPEG2VIDEO) {
/* just for testing, we also add B frames */
c - > max_b_frames = 2;
}
if (c - > codec_id == AV_CODEC_ID_MPEG1VIDEO) {
/* Needed to avoid using macroblocks in which some coeffs overflow.
* This does not happen with normal video, it just happens here as
* the motion of the chroma plane does not match the luma plane. */
c - > mb_decision = 2;
}
/* Some formats want stream headers to be separate. */
if (oc - > oformat - > flags & AVFMT_GLOBALHEADER)
c - > flags |= CODEC_FLAG_GLOBAL_HEADER;
//returning our lovely new brand stream.
return st;
}
int changeResolution(string source, int format) {
//Data members
struct SwsContext * sws_ctx = NULL;
AVFrame * pFrame = NULL;
AVFrame * outFrame = NULL;
AVPacket packet;
uint8_t * buffer = NULL;
uint8_t endcode[] = {
0,
0,
1,
0xb7
};
AVDictionary * optionsDict = NULL;
AVFormatContext * pFormatCtx = NULL;
AVFormatContext * outputContext = NULL;
AVCodecContext * pCodecCtx;
AVCodec * pCodec;
AVCodec * codec;
AVCodec * videoCodec;
AVOutputFormat * fmt;
AVStream * video_stream;
int changeWidth;
int changeHeight;
int frameFinished;
int numBytes;
int fps;
int lock = 0;
//Register all codecs & other important stuff. Vital!..
av_register_all();
//Selects the desired resolution.
if (format == 0) {
changeWidth = 320;
changeHeight = 180;
} else if (format == 1) {
changeWidth = 640;
changeHeight = 480;
} else if (format == 2) {
changeWidth = 960;
changeHeight = 540;
} else if (format == 3) {
changeWidth = 1024;
changeHeight = 768;
} else {
changeWidth = 1280;
changeHeight = 720;
}
// Open video file
int aaa;
aaa = avformat_open_input( & pFormatCtx, source.c_str(), NULL, NULL);
if (aaa != 0) {
cout << " cannot open input file \n" << endl;
cout << "aaa = " << aaa << endl;
return -1; // Couldn't open file
}
// Retrieve stream information
if (av_find_stream_info(pFormatCtx) < 0)
return -1; // Couldn't find stream information
//just checking duration casually for no reason
/*int64_t duration = pFormatCtx->duration;
cout << "the duration is " << duration << " " << endl;*/
//this writes the info about the file
av_dump_format(pFormatCtx, 0, 0, 0);
cin >> lock;
// Find the first video stream
int videoStream = -1;
int i;
for (i = 0; i < 3; i++)
if (pFormatCtx - > streams[i] - > codec - > codec_type == AVMEDIA_TYPE_VIDEO) {
videoStream = i;
cout << " lel \n ";
break;
}
if (videoStream == -1)
return -1; // Didn't find a video stream
// Get a pointer to the codec context for the video stream
pCodecCtx = pFormatCtx - > streams[videoStream] - > codec;
fps = pCodecCtx - > time_base.den;
//Find the decoder of the input file, for the video stream
pCodec = avcodec_find_decoder(pCodecCtx - > codec_id);
if (pCodec == NULL) {
fprintf(stderr, "Unsupported codec!\n");
return -1; // Codec not found
}
// Open codec, you must open it first, in order to use it.
if (avcodec_open2(pCodecCtx, pCodec, & optionsDict) < 0)
return -1; // Could not open codec
// Allocate video frame ( pFrame for taking the packets into, outFrame for processed frames to packet.)
pFrame = avcodec_alloc_frame();
outFrame = avcodec_alloc_frame();
i = 0;
int ret;
int video_frame_count = 0;
//Initiate the outFrame set the buffer & fill the properties
numBytes = avpicture_get_size(PIX_FMT_YUV420P, changeWidth, changeHeight);
buffer = (uint8_t * ) av_malloc(numBytes * sizeof(uint8_t));
avpicture_fill((AVPicture * ) outFrame, buffer, PIX_FMT_YUV420P, changeWidth, changeHeight);
int pp;
int frameNo = 0;
//allocate the outputContext, it will be the AVFormatContext of our output file.
//It will try to find the format by giving the file name.
avformat_alloc_output_context2( & outputContext, NULL, NULL, "myoutput.mp4");
//Cant find the file extension, using MPEG as default.
if (!outputContext) {
printf("Could not deduce output format from file extension: using MPEG.\n");
avformat_alloc_output_context2( & outputContext, NULL, "mpeg", "myoutput.mp4");
}
//Still cant set file extension, exit.
if (!outputContext) {
return 1;
}
//set AVOutputFormat fmt to our outputContext's format.
fmt = outputContext - > oformat;
video_stream = NULL;
//If fmt has a valid codec_id, create a new video stream.
//This function will set the streams codec & codecs desired properties.
//Stream's codec will be passed to videoCodec for later usage.
if (fmt - > video_codec != AV_CODEC_ID_NONE)
video_stream = addStream(outputContext, & videoCodec, fmt - > video_codec, changeWidth, changeHeight, fps);
//open the video using videoCodec. by avcodec_open2() i.e open the codec.
if (video_stream)
open_video(outputContext, videoCodec, video_stream);
//Creating our new output file.
if (!(fmt - > flags & AVFMT_NOFILE)) {
ret = avio_open( & outputContext - > pb, "toBeStreamed.264", AVIO_FLAG_WRITE);
if (ret < 0) {
cout << " cant open file " << endl;
return 1;
}
}
//Writing the header of format context.
//ret = avformat_write_header(outputContext, NULL);
if (ret >= 0) {
cout << "writing header success !!!" << endl;
}
//Start reading packages from input file.
while (av_read_frame(pFormatCtx, & packet) >= 0) {
// Is this a packet from the video stream?
if (packet.stream_index == videoStream) {
// Decode video package into frames
ret = avcodec_decode_video2(pCodecCtx, pFrame, & frameFinished, & packet);
if (ret < 0) {
printf(" Error decoding frame !!..");
return ret;
}
if (frameFinished) {
printf("video_frame n:%d coded_n:%d\n", video_frame_count++, pFrame - > coded_picture_number);
}
av_free_packet( & packet);
//do stuff with frame, in this case we are changing the resolution.
static struct SwsContext * img_convert_ctx_in = NULL;
if (img_convert_ctx_in == NULL) {
img_convert_ctx_in = sws_getContext(pCodecCtx - > width,
pCodecCtx - > height,
pCodecCtx - > pix_fmt,
changeWidth,
changeHeight,
PIX_FMT_YUV420P,
SWS_BICUBIC,
NULL,
NULL,
NULL);
}
//scale the frames
sws_scale(img_convert_ctx_in,
pFrame - > data,
pFrame - > linesize,
0,
pCodecCtx - > height,
outFrame - > data,
outFrame - > linesize);
//initiate the pts value
if (frameNo == 0)
outFrame - > pts = 0;
//calculate the pts value & set it.
outFrame - > pts += av_rescale_q(1, video_stream - > codec - > time_base, video_stream - > time_base);
//encode frames into packages. Package passed in #packet.
if (avcodec_encode_video2(outputContext - > streams[0] - > codec, & packet, outFrame, & pp) < 0)
cout << "Encoding frames into packages, failed. " << endl;
frameNo++;
//write the packages into file, resulting in creating a video file.
av_interleaved_write_frame(outputContext, & packet);
}
}
av_free_packet( & packet);
//av_write_trailer(outputContext);
avio_close(outputContext - > pb);
// Free the RGB image
av_free(buffer);
av_free(outFrame);
// Free the YUV frame
av_free(pFrame);
// Close the codec
avcodec_close(video_stream - > codec);
avcodec_close(pCodecCtx);
// Close the video file
avformat_close_input( & pFormatCtx);
return 0;
}
at the end of the process I get my desired file with desired codec & container & resolution.
My problem is, in a part of our project I need to get elementary video streams IN file. Such as example.264. However I can not add a stream without creating an AVFormatContext. I can not create an AVFormatContext because 264 files does not have a container,they are just raw video?, as far as I know.
I have tried the way in decoding_encoding.c which uses fwrite. However that example was for mpeg-2 codec and when I try to adapt that code to H264/AVC codec, I got "floating point division by zero" error from mediainfo and moreover, some of the properties of the video was not showing (such as FPS & playtime & quality factor). I think it has to do with the "endcode" the example adds at the end of the code. It is for mpeg-2. ( uint8_t endcode[] = { 0, 0, 1, 0xb7 }; )
Anyway, I would love to get a startpoint for this task. I have managed to come this far by using internet resources ( quite few & outdated for ffmpeg) but now I'm stuck a little.
Related
I writting a software for my classes that receive a h264 stream from a drone and i need to convert the video stream to opencv Mat.
I have no trouble to receive the frame and if i save it to a .264 file i can read the output with VLC.
The drone is sending IDR-Frame and P-Frame , since i don't need to see the video stream juste some frame of it, i was thinking of only using the IDR-Frame to get the image but i have trouble to understand how to use libavcodec from FFMPEG, how can i create a AVFrame from my IDR-Frame and how to convert it to cv::Mat after.
i have try the following code, i have store the full frame in .raw file and i try to decode them , but i get read error when i try to parse the packet to a frame, i don't thing a initialize the buffer of ARPacket the right way :
AVFormatContext* fc = 0;
int vi = -1; // vi veut dire video index
inline cv::Mat avframe_to_mat(const AVFrame* avframe)
{
AVFrame dst;
cv::Mat m;
memset(&dst, 0, sizeof(dst));
int w = avframe->width, h = avframe->height;
m = cv::Mat(h, w, CV_8UC3);
dst.data[0] = (uint8_t*)m.data;
avpicture_fill((AVPicture*)&dst, dst.data[0], AV_PIX_FMT_BGR24, w, h);
struct SwsContext *convert_ctx = NULL;
enum AVPixelFormat src_pixfmt = AV_PIX_FMT_BGR24;
enum AVPixelFormat dst_pixfmt = AV_PIX_FMT_BGR24;
convert_ctx = sws_getContext(w, h, src_pixfmt, w, h, dst_pixfmt, SWS_FAST_BILINEAR, NULL, NULL, NULL);
sws_scale(convert_ctx, avframe->data, avframe->linesize, 0, h, dst.data, dst.linesize);
sws_freeContext(convert_ctx);
return m;
}
inline bool init_stream(unsigned char* data, int len)
{
const char* file = "test.avi";
const AVCodecID codec_id = AV_CODEC_ID_H264;
AVCodec* codec = avcodec_find_encoder(codec_id);
// Crée le container pour le stream
fc = avformat_alloc_context();
/*
AVOutputFormat *of = av_guess_format(0, file, 0);
fc = avformat_alloc_context();
fc->oformat = of;
strcpy(fc->filename, file);
*/
int br = 1000000;
int w = 640;
int h = 360;
int fps = 24;
// ajoute un stream video
AVStream* pst = avformat_new_stream(fc, codec); // Pourquoi je passe pas le codec ici ?
vi = pst->index;
codec_context = avcodec_alloc_context3(codec);
codec_context->bit_rate = br;
codec_context->width = w;
codec_context->height = h;
codec_context->time_base = {1,fps};
codec_context->gop_size = 10; // Emit one intra frame every ten frames
codec_context->max_b_frames = 1;
codec_context->pix_fmt = AV_PIX_FMT_YUV420P;
// Vu quon n'est en h264
av_opt_set(codec_context->priv_data, "preset", "slow", 0);
// Ouvre notre codec
if(avcodec_open2(codec_context, codec,nullptr) < 0)
{
cerr << "Impossible d'ouvrir le codec" << endl;
return false;
}
if (!(fc->oformat->flags & AVFMT_NOFILE))
avio_open(&fc->pb, fc->filename,0);
// avformat_write_header(fc,nullptr);
return true;
}
inline void append_stream(uint8_t* data, int len)
{
if( 0 > vi)
{
cerr << "video index is less than 0" << endl;
return;
}
AVStream* pst = fc->streams[vi];
AVPacket pkt;
// Init un nouveau packet
av_init_packet(&pkt);
pkt.flags |= AV_PKT_FLAG_KEY;
pkt.data = data;
pkt.stream_index = pst->index;
pkt.size = len;
pkt.dts = AV_NOPTS_VALUE;
pkt.pts = AV_NOPTS_VALUE;
// ERROR accessing location
av_interleaved_write_frame(fc, &pkt);
}
inline void execute_staging_test(const fs::path& folder, int nbr_trame)
{
fs::path file_name = folder / "stream.bin";
if(!fs::exists(file_name))
{
cerr << "The file " << file_name.string() << " does not exists" << endl;
return;
}
avcodec_register_all();
av_log_set_level(AV_LOG_DEBUG);
int length = 0;
char* buffer;
for(int i = 0; i < nbr_trame;i++)
{
fs::path file = std::to_string(i) + ".raw";
file = folder / file;
cout << "Got frame on " << file.string() << endl;
ifstream ifs(file, ofstream::binary);
// Get la longeur du fichier pour savoir le buffer a prendre
ifs.seekg(0, ios::end);
length = ifs.tellg();
ifs.seekg(0, ios::beg);
if (length == 0) {
std::cerr << "No data in file " << file << std::endl;
return;
}
buffer = new char[length];
std::cout << "File " << file << " length is " << length << std::endl;
ifs.read(buffer, length);
cv::VideoWriter vw;
int codec = cv::VideoWriter::fourcc('X', '2', '6', '4');
if(!fc)
{
if(!init_stream((unsigned char*)buffer, length))
{
return;
}
}
if(fc)
{
append_stream((unsigned char*)buffer, length);
}
}
}
Thanks you very mutch if you can help me , i'm a novice in c++ and i have never deal with video stream. If you want to see the full code its host on github repo to this project
One of the things I'm trying to achieve is parallel encoding via FFmpeg's c API. This looks to work out of the box quite nicely; however, I've changed the goal posts slightly:
In an existing application, I already have a thread pool at hand. Instead of using another thread pool via FFmpeg, I would like reuse the existing thread pool in my application. Having studied the latest FFmpeg trunk docs, it very much looks possible.
Using some FFmpeg sample code, I've created a sample application to demonstrate what I'm trying to achieve (see below). The sample app generates a video-only mpeg2 ts using the mp2v codec.
The problem I'm experiencing is that the custom 'thread_execute' or 'thread_execute2' are never invoked. This is despite the fact that the codec appears to indicate that threading is supported. Please be aware that I have not yet plumbed in the thread pool just yet. My first goal is for it to call the custom function pointer.
I've tried to get assistance on the FFmpeg mailing lists but to no avail.
#include <iostream>
#include <thread>
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include <cstring>
#include <future>
extern "C"
{
#include <libavutil/avassert.h>
#include <libavutil/channel_layout.h>
#include <libavutil/opt.h>
#include <libavutil/timestamp.h>
#include <libavformat/avformat.h>
//#include <libswscale/swscale.h>
#include <libswresample/swresample.h>
}
#define STREAM_DURATION 1000.0
#define STREAM_FRAME_RATE 25 /* 25 images/s */
#define STREAM_PIX_FMT AV_PIX_FMT_YUV420P /* default pix_fmt */
#define SCALE_FLAGS SWS_BICUBIC
// a wrapper around a single output AVStream
typedef struct OutputStream {
AVStream *st;
AVCodecContext *enc;
/* pts of the next frame that will be generated */
int64_t next_pts;
int samples_count;
AVFrame *frame;
AVFrame *tmp_frame;
float t, tincr, tincr2;
struct SwsContext *sws_ctx;
struct SwrContext *swr_ctx;
} OutputStream;
/////////////////////////////////////////////////////////////////////////////
// The ffmpeg variation raises compiler warnings.
char *cb_av_ts2str(char *buf, int64_t ts)
{
std::memset(buf,0,AV_TS_MAX_STRING_SIZE);
return av_ts_make_string(buf,ts);
}
/////////////////////////////////////////////////////////////////////////////
// The ffmpeg variation raises compiler warnings.
char *cb_av_ts2timestr(char *buf, int64_t ts, AVRational *tb)
{
std::memset(buf,0,sizeof(AV_TS_MAX_STRING_SIZE));
return av_ts_make_time_string(buf,ts,tb);
}
/////////////////////////////////////////////////////////////////////////////
// The ffmpeg variation raises compiler warnings.
char *cb_av_err2str(char *errbuf, size_t errbuf_size, int errnum)
{
std::memset(errbuf,0,errbuf_size);
return av_make_error_string(errbuf,errbuf_size,errnum);
}
int thread_execute(AVCodecContext* s, int (*func)(AVCodecContext *c2, void *arg2), void* arg, int* ret, int count, int size)
{
// Do it all serially for now
std::cout << "thread_execute" << std::endl;
for (int k = 0; k < count; ++k)
{
ret[k] = func(s, arg);
}
return 0;
}
int thread_execute2(AVCodecContext* s, int (*func)(AVCodecContext* c2, void* arg2, int, int), void* arg, int* ret, int count)
{
// Do it all serially for now
std::cout << "thread_execute2" << std::endl;
for (int k = 0; k < count; ++k)
{
ret[k] = func(s, arg, k, count);
}
return 0;
}
static void log_packet(const AVFormatContext *fmt_ctx, const AVPacket *pkt)
{
char s[AV_TS_MAX_STRING_SIZE];
AVRational *time_base = &fmt_ctx->streams[pkt->stream_index]->time_base;
printf("pts:%s pts_time:%s dts:%s dts_time:%s duration:%s duration_time:%s stream_index:%d\n",
cb_av_ts2str(s,pkt->pts), cb_av_ts2timestr(s,pkt->pts, time_base),
cb_av_ts2str(s,pkt->dts), cb_av_ts2timestr(s,pkt->dts, time_base),
cb_av_ts2str(s,pkt->duration), cb_av_ts2timestr(s,pkt->duration, time_base),
pkt->stream_index);
}
static int write_frame(AVFormatContext *fmt_ctx, const AVRational *time_base, AVStream *st, AVPacket *pkt)
{
/* rescale output packet timestamp values from codec to stream timebase */
av_packet_rescale_ts(pkt, *time_base, st->time_base);
pkt->stream_index = st->index;
/* Write the compressed frame to the media file. */
log_packet(fmt_ctx, pkt);
return av_interleaved_write_frame(fmt_ctx, pkt);
}
/* Add an output stream. */
static void add_stream(OutputStream *ost, AVFormatContext *oc,
AVCodec **codec,
enum AVCodecID codec_id)
{
AVCodecContext *c;
int i;
/* find the encoder */
*codec = avcodec_find_encoder(codec_id);
if (!(*codec)) {
fprintf(stderr, "Could not find encoder for '%s'\n",
avcodec_get_name(codec_id));
exit(1);
}
ost->st = avformat_new_stream(oc, NULL);
if (!ost->st) {
fprintf(stderr, "Could not allocate stream\n");
exit(1);
}
ost->st->id = oc->nb_streams-1;
c = avcodec_alloc_context3(*codec);
if (!c) {
fprintf(stderr, "Could not alloc an encoding context\n");
exit(1);
}
ost->enc = c;
switch ((*codec)->type)
{
case AVMEDIA_TYPE_AUDIO:
c->sample_fmt = (*codec)->sample_fmts ?
(*codec)->sample_fmts[0] : AV_SAMPLE_FMT_FLTP;
c->bit_rate = 64000;
c->sample_rate = 44100;
if ((*codec)->supported_samplerates) {
c->sample_rate = (*codec)->supported_samplerates[0];
for (i = 0; (*codec)->supported_samplerates[i]; i++) {
if ((*codec)->supported_samplerates[i] == 44100)
c->sample_rate = 44100;
}
}
c->channels = av_get_channel_layout_nb_channels(c->channel_layout);
c->channel_layout = AV_CH_LAYOUT_STEREO;
if ((*codec)->channel_layouts) {
c->channel_layout = (*codec)->channel_layouts[0];
for (i = 0; (*codec)->channel_layouts[i]; i++) {
if ((*codec)->channel_layouts[i] == AV_CH_LAYOUT_STEREO)
c->channel_layout = AV_CH_LAYOUT_STEREO;
}
}
c->channels = av_get_channel_layout_nb_channels(c->channel_layout);
ost->st->time_base = (AVRational){ 1, c->sample_rate };
break;
case AVMEDIA_TYPE_VIDEO:
c->codec_id = codec_id;
c->bit_rate = 400000;
/* Resolution must be a multiple of two. */
c->width = 352;
c->height = 288;
/* timebase: This is the fundamental unit of time (in seconds) in terms
* of which frame timestamps are represented. For fixed-fps content,
* timebase should be 1/framerate and timestamp increments should be
* identical to 1. */
ost->st->time_base = (AVRational){ 1, STREAM_FRAME_RATE };
c->time_base = ost->st->time_base;
c->gop_size = 12; /* emit one intra frame every twelve frames at most */
c->pix_fmt = STREAM_PIX_FMT;
if (c->codec_id == AV_CODEC_ID_MPEG2VIDEO) {
/* just for testing, we also add B-frames */
c->max_b_frames = 2;
}
if (c->codec_id == AV_CODEC_ID_MPEG1VIDEO) {
/* Needed to avoid using macroblocks in which some coeffs overflow.
* This does not happen with normal video, it just happens here as
* the motion of the chroma plane does not match the luma plane. */
c->mb_decision = 2;
}
break;
default:
break;
}
if (c->codec->capabilities & AV_CODEC_CAP_FRAME_THREADS ||
c->codec->capabilities & AV_CODEC_CAP_SLICE_THREADS)
{
if (c->codec->capabilities & AV_CODEC_CAP_FRAME_THREADS)
{
c->thread_type = FF_THREAD_FRAME;
}
if (c->codec->capabilities & AV_CODEC_CAP_SLICE_THREADS)
{
c->thread_type = FF_THREAD_SLICE;
}
c->execute = &thread_execute;
c->execute2 = &thread_execute2;
c->thread_count = 4;
// NOTE: Testing opaque.
c->opaque = (void*)0xff;
}
/* Some formats want stream headers to be separate. */
if (oc->oformat->flags & AVFMT_GLOBALHEADER)
c->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
}
/**************************************************************/
/* video output */
static AVFrame *alloc_picture(enum AVPixelFormat pix_fmt, int width, int height)
{
AVFrame *picture;
int ret;
picture = av_frame_alloc();
if (!picture)
return NULL;
picture->format = pix_fmt;
picture->width = width;
picture->height = height;
/* allocate the buffers for the frame data */
ret = av_frame_get_buffer(picture, 32);
if (ret < 0) {
fprintf(stderr, "Could not allocate frame data.\n");
exit(1);
}
return picture;
}
static void open_video(AVFormatContext *oc, AVCodec *codec, OutputStream *ost, AVDictionary *opt_arg)
{
int ret;
AVCodecContext *c = ost->enc;
//AVDictionary *opt = NULL;
//av_dict_copy(&opt, opt_arg, 0);
/* open the codec */
ret = avcodec_open2(c, codec, NULL);
//av_dict_free(&opt);
if (ret < 0) {
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Could not open video codec: %s\n", cb_av_err2str(s,AV_ERROR_MAX_STRING_SIZE,ret));
exit(1);
}
/* allocate and init a re-usable frame */
ost->frame = alloc_picture(c->pix_fmt, c->width, c->height);
if (!ost->frame) {
fprintf(stderr, "Could not allocate video frame\n");
exit(1);
}
/* If the output format is not YUV420P, then a temporary YUV420P
* picture is needed too. It is then converted to the required
* output format. */
ost->tmp_frame = NULL;
if (c->pix_fmt != AV_PIX_FMT_YUV420P) {
ost->tmp_frame = alloc_picture(AV_PIX_FMT_YUV420P, c->width, c->height);
if (!ost->tmp_frame) {
fprintf(stderr, "Could not allocate temporary picture\n");
exit(1);
}
}
/* copy the stream parameters to the muxer */
ret = avcodec_parameters_from_context(ost->st->codecpar, c);
if (ret < 0) {
fprintf(stderr, "Could not copy the stream parameters\n");
exit(1);
}
}
/* Prepare a dummy image. */
static void fill_yuv_image(AVFrame *pict, int frame_index,
int width, int height)
{
int x, y, i;
i = frame_index;
/* Y */
for (y = 0; y < height; y++)
for (x = 0; x < width; x++)
pict->data[0][y * pict->linesize[0] + x] = x + y + i * 3;
/* Cb and Cr */
for (y = 0; y < height / 2; y++) {
for (x = 0; x < width / 2; x++) {
pict->data[1][y * pict->linesize[1] + x] = 128 + y + i * 2;
pict->data[2][y * pict->linesize[2] + x] = 64 + x + i * 5;
}
}
}
static AVFrame *get_video_frame(OutputStream *ost)
{
AVCodecContext *c = ost->enc;
/* check if we want to generate more frames */
if (av_compare_ts(ost->next_pts, c->time_base,
STREAM_DURATION, (AVRational){ 1, 1 }) >= 0)
return NULL;
/* when we pass a frame to the encoder, it may keep a reference to it
* internally; make sure we do not overwrite it here */
if (av_frame_make_writable(ost->frame) < 0)
exit(1);
if (c->pix_fmt != AV_PIX_FMT_YUV420P) {
/* as we only generate a YUV420P picture, we must convert it
* to the codec pixel format if needed */
/*if (!ost->sws_ctx) {
ost->sws_ctx = sws_getContext(c->width, c->height,
AV_PIX_FMT_YUV420P,
c->width, c->height,
c->pix_fmt,
SCALE_FLAGS, NULL, NULL, NULL);
if (!ost->sws_ctx) {
fprintf(stderr,
"Could not initialize the conversion context\n");
exit(1);
}
}
fill_yuv_image(ost->tmp_frame, ost->next_pts, c->width, c->height);
sws_scale(ost->sws_ctx,
(const uint8_t * const *)ost->tmp_frame->data, ost->tmp_frame->linesize,
0, c->height, ost->frame->data, ost->frame->linesize);*/
} else {
fill_yuv_image(ost->frame, ost->next_pts, c->width, c->height);
}
ost->frame->pts = ost->next_pts++;
return ost->frame;
}
/*
* encode one video frame and send it to the muxer
* return 1 when encoding is finished, 0 otherwise
*/
static int write_video_frame(AVFormatContext *oc, OutputStream *ost)
{
int ret;
AVCodecContext *c;
AVFrame *frame;
int got_packet = 0;
AVPacket pkt = { 0 };
c = ost->enc;
frame = get_video_frame(ost);
if (frame)
{
ret = avcodec_send_frame(ost->enc, frame);
if (ret < 0)
{
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Error encoding video frame: %s\n", cb_av_err2str(s, AV_ERROR_MAX_STRING_SIZE, ret));
exit(1);
}
}
av_init_packet(&pkt);
ret = avcodec_receive_packet(ost->enc,&pkt);
if (ret < 0)
{
if (ret == AVERROR(EAGAIN)) { ret = 0; }
else
{
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Error receiving packet: %s\n", cb_av_err2str(s,AV_ERROR_MAX_STRING_SIZE,ret));
exit(1);
}
}
else
{
got_packet = 1;
ret = write_frame(oc, &c->time_base, ost->st, &pkt);
}
if (ret < 0) {
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Error while writing video frame: %s\n", cb_av_err2str(s,AV_ERROR_MAX_STRING_SIZE,ret));
exit(1);
}
return (frame || got_packet) ? 0 : 1;
}
static void close_stream(AVFormatContext *oc, OutputStream *ost)
{
avcodec_free_context(&ost->enc);
av_frame_free(&ost->frame);
av_frame_free(&ost->tmp_frame);
//sws_freeContext(ost->sws_ctx);
//swr_free(&ost->swr_ctx);
}
/**************************************************************/
/* media file output */
int main(int argc, char **argv)
{
OutputStream video_st = { 0 }, audio_st = { 0 };
const char *filename;
AVOutputFormat *fmt;
AVFormatContext *oc;
AVCodec /**audio_codec,*/ *video_codec;
int ret;
int have_video = 0, have_audio = 0;
int encode_video = 0, encode_audio = 0;
AVDictionary *opt = NULL;
int i;
/* Initialize libavcodec, and register all codecs and formats. */
av_register_all();
avformat_network_init();
if (argc < 2) {
printf("usage: %s output_file\n"
"API example program to output a media file with libavformat.\n"
"This program generates a synthetic audio and video stream, encodes and\n"
"muxes them into a file named output_file.\n"
"The output format is automatically guessed according to the file extension.\n"
"Raw images can also be output by using '%%d' in the filename.\n"
"\n", argv[0]);
return 1;
}
filename = argv[1];
for (i = 2; i+1 < argc; i+=2) {
if (!strcmp(argv[i], "-flags") || !strcmp(argv[i], "-fflags"))
av_dict_set(&opt, argv[i]+1, argv[i+1], 0);
}
const char *pfilename = filename;
/* allocate the output media context */
avformat_alloc_output_context2(&oc, NULL, "mpegts", pfilename);
if (!oc) {
printf("Could not deduce output format from file extension: using MPEG.\n");
avformat_alloc_output_context2(&oc, NULL, "mpeg", pfilename);
}
if (!oc)
return 1;
fmt = oc->oformat;
/* Add the audio and video streams using the default format codecs
* and initialize the codecs. */
if (fmt->video_codec != AV_CODEC_ID_NONE) {
add_stream(&video_st, oc, &video_codec, fmt->video_codec);
have_video = 1;
encode_video = 1;
}
/*if (fmt->audio_codec != AV_CODEC_ID_NONE) {
add_stream(&audio_st, oc, &audio_codec, fmt->audio_codec);
have_audio = 1;
encode_audio = 1;
}*/
/* Now that all the parameters are set, we can open the audio and
* video codecs and allocate the necessary encode buffers. */
if (have_video)
open_video(oc, video_codec, &video_st, opt);
//if (have_audio)
// open_audio(oc, audio_codec, &audio_st, opt);
av_dump_format(oc, 0, pfilename, 1);
/* open the output file, if needed */
if (!(fmt->flags & AVFMT_NOFILE)) {
ret = avio_open(&oc->pb, pfilename, AVIO_FLAG_WRITE);
if (ret < 0) {
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Could not open '%s': %s\n", pfilename,
cb_av_err2str(s,AV_ERROR_MAX_STRING_SIZE,ret));
return 1;
}
}
/* Write the stream header, if any. */
ret = avformat_write_header(oc, &opt);
if (ret < 0) {
char s[AV_ERROR_MAX_STRING_SIZE];
fprintf(stderr, "Error occurred when opening output file: %s\n",
cb_av_err2str(s,AV_ERROR_MAX_STRING_SIZE,ret));
return 1;
}
while (encode_video || encode_audio) {
/* select the stream to encode */
if (encode_video &&
(!encode_audio || av_compare_ts(video_st.next_pts, video_st.enc->time_base,
audio_st.next_pts, audio_st.enc->time_base) <= 0)) {
encode_video = !write_video_frame(oc, &video_st);
} else {
//encode_audio = !write_audio_frame(oc, &audio_st);
}
//std::this_thread::sleep_for(std::chrono::milliseconds(35));
}
/* Write the trailer, if any. The trailer must be written before you
* close the CodecContexts open when you wrote the header; otherwise
* av_write_trailer() may try to use memory that was freed on
* av_codec_close(). */
av_write_trailer(oc);
/* Close each codec. */
if (have_video)
close_stream(oc, &video_st);
if (have_audio)
close_stream(oc, &audio_st);
if (!(fmt->flags & AVFMT_NOFILE))
/* Close the output file. */
avio_closep(&oc->pb);
/* free the stream */
avformat_free_context(oc);
return 0;
}
//
Environment:
Ubuntu Zesty (17.04)
FFmpeg version 3.2.4 (via package manager)
gcc 6.3 (C++)
You have to do following:
call avcodec_alloc_context3(...). This call will set default execute and execute2 functions in new context
set c->thread_count = number_of_threads_in_your_thread_pool()
call avcodec_open2(...).
set c->execute and c->execute2 to point to your functions
call ff_thread_free(c). This function isnt exposed in libavcodec headers but you can add following line:
extern "C" void ff_thread_free(AVCodecContext *s);
Drawback is that libavcodec will create internal thread pool after avcodec_open2(...) call, and that pool will be deleted in ff_thread_free() call.
Internal thread pool is very efficient, but its not good if you plan to do parallel encoding of multiple video feeds. In that case libavcodec will create separate thread pool for each encoding video feed.
I want to decode H.264 video from a collection of MPEG-2 Transport Stream packets but I am not clear what to pass to avcodec_decode_video2
The documentation says to pass "the input AVPacket containing the input buffer."
But what should be in the input buffer?
A PES packet will be spread across the payload portion of several TS packets, with NALU(s) inside the PES. So pass a TS fragment? The entire PES? PES payload only?
This Sample Code mentions:
BUT some other codecs (msmpeg4, mpeg4) are inherently frame based, so
you must call them with all the data for one frame exactly. You must
also initialize 'width' and 'height' before initializing them.
But I can find no info on what "all the data" means...
Passing a fragment of a TS packet payload is not working:
AVPacket avDecPkt;
av_init_packet(&avDecPkt);
avDecPkt.data = inbuf_ptr;
avDecPkt.size = esBufSize;
len = avcodec_decode_video2(mpDecoderContext, mpFrameDec, &got_picture, &avDecPkt);
if (len < 0)
{
printf(" TS PKT #%.0f. Error decoding frame #%04d [rc=%d '%s']\n",
tsPacket.pktNum, mDecodedFrameNum, len, av_make_error_string(errMsg, 128, len));
return;
}
output
[h264 # 0x81cd2a0] no frame!
TS PKT #2973. Error decoding frame #0001 [rc=-1094995529 'Invalid data found when processing input']
EDIT
Using the excellent hits from WLGfx, I made this simple program to try decoding TS packets. As input, I prepared a file containing only TS packets from the Video PID.
It feels close but I don't know how to set up the FormatContext. The code below segfaults at av_read_frame() (and internally at ret = s->iformat->read_packet(s, pkt)). s->iformat is zero.
Suggestions?
EDIT II - Sorry, for got post source code **
**EDIT III - Sample code updated to simulate reading TS PKT Queue
/*
* Test program for video decoder
*/
#include <stdio.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
extern "C" {
#ifdef __cplusplus
#define __STDC_CONSTANT_MACROS
#ifdef _STDINT_H
#undef _STDINT_H
#endif
#include <stdint.h>
#endif
}
extern "C" {
#include "libavcodec/avcodec.h"
#include "libavformat/avformat.h"
#include "libswscale/swscale.h"
#include "libavutil/imgutils.h"
#include "libavutil/opt.h"
}
class VideoDecoder
{
public:
VideoDecoder();
bool rcvTsPacket(AVPacket &inTsPacket);
private:
AVCodec *mpDecoder;
AVCodecContext *mpDecoderContext;
AVFrame *mpDecodedFrame;
AVFormatContext *mpFmtContext;
};
VideoDecoder::VideoDecoder()
{
av_register_all();
// FORMAT CONTEXT SETUP
mpFmtContext = avformat_alloc_context();
mpFmtContext->flags = AVFMT_NOFILE;
// ????? WHAT ELSE ???? //
// DECODER SETUP
mpDecoder = avcodec_find_decoder(AV_CODEC_ID_H264);
if (!mpDecoder)
{
printf("Could not load decoder\n");
exit(11);
}
mpDecoderContext = avcodec_alloc_context3(NULL);
if (avcodec_open2(mpDecoderContext, mpDecoder, NULL) < 0)
{
printf("Cannot open decoder context\n");
exit(1);
}
mpDecodedFrame = av_frame_alloc();
}
bool
VideoDecoder::rcvTsPacket(AVPacket &inTsPkt)
{
bool ret = true;
if ((av_read_frame(mpFmtContext, &inTsPkt)) < 0)
{
printf("Error in av_read_frame()\n");
ret = false;
}
else
{
// success. Decode the TS packet
int got;
int len = avcodec_decode_video2(mpDecoderContext, mpDecodedFrame, &got, &inTsPkt);
if (len < 0)
ret = false;
if (got)
printf("GOT A DECODED FRAME\n");
}
return ret;
}
int
main(int argc, char **argv)
{
if (argc != 2)
{
printf("Usage: %s tsInFile\n", argv[0]);
exit(1);
}
FILE *tsInFile = fopen(argv[1], "r");
if (!tsInFile)
{
perror("Could not open TS input file");
exit(2);
}
unsigned int tsPktNum = 0;
uint8_t tsBuffer[256];
AVPacket tsPkt;
av_init_packet(&tsPkt);
VideoDecoder vDecoder;
while (!feof(tsInFile))
{
tsPktNum++;
tsPkt.size = 188;
tsPkt.data = tsBuffer;
fread(tsPkt.data, 188, 1, tsInFile);
vDecoder.rcvTsPacket(tsPkt);
}
}
I've got some code snippets that might help you out as I've been working with MPEG-TS also.
Starting with my packet thread which checks each packet against the stream ID's which I've already found and got the codec contexts:
void *FFMPEG::thread_packet_function(void *arg) {
FFMPEG *ffmpeg = (FFMPEG*)arg;
for (int c = 0; c < MAX_PACKETS; c++)
ffmpeg->free_packets[c] = &ffmpeg->packet_list[c];
ffmpeg->packet_pos = MAX_PACKETS;
Audio.start_decoding();
Video.start_decoding();
Subtitle.start_decoding();
while (!ffmpeg->thread_quit) {
if (ffmpeg->packet_pos != 0 &&
Audio.okay_add_packet() &&
Video.okay_add_packet() &&
Subtitle.okay_add_packet()) {
pthread_mutex_lock(&ffmpeg->packet_mutex); // get free packet
AVPacket *pkt = ffmpeg->free_packets[--ffmpeg->packet_pos]; // pre decrement
pthread_mutex_unlock(&ffmpeg->packet_mutex);
if ((av_read_frame(ffmpeg->fContext, pkt)) >= 0) { // success
int id = pkt->stream_index;
if (id == ffmpeg->aud_stream.stream_id) Audio.add_packet(pkt);
else if (id == ffmpeg->vid_stream.stream_id) Video.add_packet(pkt);
else if (id == ffmpeg->sub_stream.stream_id) Subtitle.add_packet(pkt);
else { // unknown packet
av_packet_unref(pkt);
pthread_mutex_lock(&ffmpeg->packet_mutex); // put packet back
ffmpeg->free_packets[ffmpeg->packet_pos++] = pkt;
pthread_mutex_unlock(&ffmpeg->packet_mutex);
//LOGI("Dumping unknown packet, id %d", id);
}
} else {
av_packet_unref(pkt);
pthread_mutex_lock(&ffmpeg->packet_mutex); // put packet back
ffmpeg->free_packets[ffmpeg->packet_pos++] = pkt;
pthread_mutex_unlock(&ffmpeg->packet_mutex);
//LOGI("No packet read");
}
} else { // buffers full so yield
//LOGI("Packet reader on hold: Audio-%d, Video-%d, Subtitle-%d",
// Audio.packet_pos, Video.packet_pos, Subtitle.packet_pos);
usleep(1000);
//sched_yield();
}
}
return 0;
}
Each decoder for audio, video and subtitles have their own threads which receive the packets from the above thread in ring buffers. I've had to separate the decoders into their own threads because CPU usage was increasing when I started using the deinterlace filter.
My video decoder reads the packets from the buffers and when it has finished with the packet sends it back to be unref'd and can be used again. Balancing the packet buffers doesn't take that much time once everything is running.
Here's the snipped from my video decoder:
void *VideoManager::decoder(void *arg) {
LOGI("Video decoder started");
VideoManager *mgr = (VideoManager *)arg;
while (!ffmpeg.thread_quit) {
pthread_mutex_lock(&mgr->packet_mutex);
if (mgr->packet_pos != 0) {
// fetch first packet to decode
AVPacket *pkt = mgr->packets[0];
// shift list down one
for (int c = 1; c < mgr->packet_pos; c++) {
mgr->packets[c-1] = mgr->packets[c];
}
mgr->packet_pos--;
pthread_mutex_unlock(&mgr->packet_mutex); // finished with packets array
int got;
AVFrame *frame = ffmpeg.vid_stream.frame;
avcodec_decode_video2(ffmpeg.vid_stream.context, frame, &got, pkt);
ffmpeg.finished_with_packet(pkt);
if (got) {
#ifdef INTERLACE_ALL
if (!frame->interlaced_frame) mgr->add_av_frame(frame, 0);
else {
if (!mgr->filter_initialised) mgr->init_filter_graph(frame);
av_buffersrc_add_frame_flags(mgr->filter_src_ctx, frame, AV_BUFFERSRC_FLAG_KEEP_REF);
int c = 0;
while (true) {
AVFrame *filter_frame = ffmpeg.vid_stream.filter_frame;
int result = av_buffersink_get_frame(mgr->filter_sink_ctx, filter_frame);
if (result == AVERROR(EAGAIN) ||
result == AVERROR(AVERROR_EOF) ||
result < 0)
break;
mgr->add_av_frame(filter_frame, c++);
av_frame_unref(filter_frame);
}
//LOGI("Interlaced %d frames, decode %d, playback %d", c, mgr->decode_pos, mgr->playback_pos);
}
#elif defined(INTERLACE_HALF)
if (!frame->interlaced_frame) mgr->add_av_frame(frame, 0);
else {
if (!mgr->filter_initialised) mgr->init_filter_graph(frame);
av_buffersrc_add_frame_flags(mgr->filter_src_ctx, frame, AV_BUFFERSRC_FLAG_KEEP_REF);
int c = 0;
while (true) {
AVFrame *filter_frame = ffmpeg.vid_stream.filter_frame;
int result = av_buffersink_get_frame(mgr->filter_sink_ctx, filter_frame);
if (result == AVERROR(EAGAIN) ||
result == AVERROR(AVERROR_EOF) ||
result < 0)
break;
mgr->add_av_frame(filter_frame, c++);
av_frame_unref(filter_frame);
}
//LOGI("Interlaced %d frames, decode %d, playback %d", c, mgr->decode_pos, mgr->playback_pos);
}
#else
mgr->add_av_frame(frame, 0);
#endif
}
//LOGI("decoded video packet");
} else {
pthread_mutex_unlock(&mgr->packet_mutex);
}
}
LOGI("Video decoder ended");
}
As you can see, I'm using a mutex when passing packets back and forth.
Once a frame has been got I just copy the YUV buffers from the frame for later use into another buffer list. I don't convert the YUV, I use a shader which converts the YUV to RGB on the GPU.
The next snippet adds my decoded frame to my buffer list. This may help understand how to deal with the data.
void VideoManager::add_av_frame(AVFrame *frame, int field_num) {
int y_linesize = frame->linesize[0];
int u_linesize = frame->linesize[1];
int hgt = frame->height;
int y_buffsize = y_linesize * hgt;
int u_buffsize = u_linesize * hgt / 2;
int buffsize = y_buffsize + u_buffsize + u_buffsize;
VideoBuffer *buffer = &buffers[decode_pos];
if (ffmpeg.is_network && playback_pos == decode_pos) { // patched 25/10/16 wlgfx
buffer->used = false;
if (!buffer->data) buffer->data = (char*)mem.alloc(buffsize);
if (!buffer->data) {
LOGI("Dropped frame, allocation error");
return;
}
} else if (playback_pos == decode_pos) {
LOGI("Dropped frame, ran out of decoder frame buffers");
return;
} else if (!buffer->data) {
buffer->data = (char*)mem.alloc(buffsize);
if (!buffer->data) {
LOGI("Dropped frame, allocation error.");
return;
}
}
buffer->y_frame = buffer->data;
buffer->u_frame = buffer->y_frame + y_buffsize;
buffer->v_frame = buffer->y_frame + y_buffsize + u_buffsize;
buffer->wid = frame->width;
buffer->hgt = hgt;
buffer->y_linesize = y_linesize;
buffer->u_linesize = u_linesize;
int64_t pts = av_frame_get_best_effort_timestamp(frame);
buffer->pts = pts;
buffer->buffer_size = buffsize;
double field_add = av_q2d(ffmpeg.vid_stream.context->time_base) * field_num;
buffer->frame_time = av_q2d(ts_stream) * pts + field_add;
memcpy(buffer->y_frame, frame->data[0], (size_t) (buffer->y_linesize * buffer->hgt));
memcpy(buffer->u_frame, frame->data[1], (size_t) (buffer->u_linesize * buffer->hgt / 2));
memcpy(buffer->v_frame, frame->data[2], (size_t) (buffer->u_linesize * buffer->hgt / 2));
buffer->used = true;
decode_pos = (++decode_pos) % MAX_VID_BUFFERS;
//if (field_num == 0) LOGI("Video %.2f, %d - %d",
// buffer->frame_time - Audio.pts_start_time, decode_pos, playback_pos);
}
If there's anything else that I may be able to help with just give me a shout. :-)
EDIT:
The snippet how I open my video stream context which automatically determines the codec, whether it is h264, mpeg2, or another:
void FFMPEG::open_video_stream() {
vid_stream.stream_id = av_find_best_stream(fContext, AVMEDIA_TYPE_VIDEO,
-1, -1, &vid_stream.codec, 0);
if (vid_stream.stream_id == -1) return;
vid_stream.context = fContext->streams[vid_stream.stream_id]->codec;
if (!vid_stream.codec || avcodec_open2(vid_stream.context,
vid_stream.codec, NULL) < 0) {
vid_stream.stream_id = -1;
return;
}
vid_stream.frame = av_frame_alloc();
vid_stream.filter_frame = av_frame_alloc();
}
EDIT2:
This is how I've opened the input stream, whether it be file or URL. The AVFormatContext is the main context for the stream.
bool FFMPEG::start_stream(char *url_, float xtrim, float ytrim, int gain) {
aud_stream.stream_id = -1;
vid_stream.stream_id = -1;
sub_stream.stream_id = -1;
this->url = url_;
this->xtrim = xtrim;
this->ytrim = ytrim;
Audio.volume = gain;
Audio.init();
Video.init();
fContext = avformat_alloc_context();
if ((avformat_open_input(&fContext, url_, NULL, NULL)) != 0) {
stop_stream();
return false;
}
if ((avformat_find_stream_info(fContext, NULL)) < 0) {
stop_stream();
return false;
}
// network stream will overwrite packets if buffer is full
is_network = url.substr(0, 4) == "udp:" ||
url.substr(0, 4) == "rtp:" ||
url.substr(0, 5) == "rtsp:" ||
url.substr(0, 5) == "http:"; // added for wifi broadcasting ability
// determine if stream is audio only
is_mp3 = url.substr(url.size() - 4) == ".mp3";
LOGI("Stream: %s", url_);
if (!open_audio_stream()) {
stop_stream();
return false;
}
if (is_mp3) {
vid_stream.stream_id = -1;
sub_stream.stream_id = -1;
} else {
open_video_stream();
open_subtitle_stream();
if (vid_stream.stream_id == -1) { // switch to audio only
close_subtitle_stream();
is_mp3 = true;
}
}
LOGI("Audio: %d, Video: %d, Subtitle: %d",
aud_stream.stream_id,
vid_stream.stream_id,
sub_stream.stream_id);
if (aud_stream.stream_id != -1) {
LOGD("Audio stream time_base {%d, %d}",
aud_stream.context->time_base.num,
aud_stream.context->time_base.den);
}
if (vid_stream.stream_id != -1) {
LOGD("Video stream time_base {%d, %d}",
vid_stream.context->time_base.num,
vid_stream.context->time_base.den);
}
LOGI("Starting packet and decode threads");
thread_quit = false;
pthread_create(&thread_packet, NULL, &FFMPEG::thread_packet_function, this);
Display.set_overlay_timout(3.0);
return true;
}
EDIT: (constructing an AVPacket)
Construct an AVPacket to send to the decoder...
AVPacket packet;
av_init_packet(&packet);
packet.data = myTSpacketdata; // pointer to the TS packet
packet.size = 188;
You should be able to reuse the packet. And it might need unref'ing.
You must first use the avcodec library to get the compressed frames out of the file. Then you can decode them using avcodec_decode_video2. look at this tutorial http://dranger.com/ffmpeg/
im reading a udp-mjpeg-stream with the ffmpeg-API. When i read and display the Stream with an ARM-Processor i have 2 Problems:
1- The Applikation is too slow and there is a big delay between network cam and displayed video.
2- the memory usage increases every time when i call the function av_read_frame().
The Source code
const char *cam1_url = "udp://192.168.1.1:1234";
AVCodec *pCodec;
AVFrame *pFrame, *pFrameRGB;
AVCodecContext *pCodecCon;
AVDictionary *pUdpStreamOptions = NULL;
AVInputFormat *pMjpegFormat = av_find_input_format("mjpeg");
av_dict_set(&pUdpStreamOptions, "fifo_size", "5000000", 0);
av_register_all();
avdevice_register_all();
avcodec_register_all();
avformat_network_init();
AVFormatContext *pFormatCont = avformat_alloc_context();
if(avformat_open_input(&pFormatCont,cam1_url,pMjpegFormat,&pUdpStreamOptions) < 0)
{
cout << "!! Error !! - avformat_open_input(): failed to open input URL" << endl;
}
if(avformat_find_stream_info(pFormatCont,NULL) < 0)
{
cout << "!! Error !! - avformat_find_stream_info(), Failed to retrieve stream info" << endl;
}
av_dump_format(pFormatCont, 0, cam1_url, 0);
int videoStream;
for(int i=0; i< pFormatCont->nb_streams; i++)
{
if(pFormatCont->streams[i]->codec->codec_type==AVMEDIA_TYPE_VIDEO)
{
videoStream=i;
cout << " videoStream = " << videoStream << endl;
}
}
pCodecCon = pFormatCont->streams[videoStream]->codec;
pCodec = avcodec_find_decoder(pCodecCon->codec_id);
if(NULL == pCodec)
{
cout << "couldnt find codec" << endl;
return EXIT_FAILURE;
}
if(avcodec_open2(pCodecCon,pCodec,NULL) < 0)
{
cout << "!! Error !! - in avcodec_open2()" << endl;
return EXIT_FAILURE;
}
uint8_t *frameBuffer;
int numRxBytes = 0;
AVPixelFormat pFormat =AV_PIX_FMT_BGR24;
int width_rgb = (int)((float)pCodecCon->width);
int height_rgb = (int)((float)pCodecCon->height);
numRxBytes = avpicture_get_size(pFormat,width_rgb,height_rgb);
frameBuffer = (uint8_t *) av_malloc(numRxBytes*sizeof(uint8_t));
avpicture_fill((AVPicture *) pFrameRGB, frameBuffer, pFormat,width_rgb,height_rgb);
AVPacket rx_pkt; // received packet
int frameFinished = 0;
struct SwsContext *imgConvertCtx;
av_init_packet(&rx_pkt);
while(av_read_frame(pFormatCont, &rx_pkt) >= 0)
{
if(rx_pkt.stream_index == videoStream)
{
av_frame_free(&pFrame);
pFrame = av_frame_alloc();
av_frame_free(&pFrameRGB);
pFrameRGB = av_frame_alloc();
avcodec_decode_video2(pCodecCon, pFrame, &frameFinished,&rx_pkt);
if(frameFinished)
{
imgConvertCtx = sws_getCachedContext(NULL, pFrame->width,pFrame->height, AV_PIX_FMT_YUVJ420P,width_rgb,height_rgb,AV_PIX_FMT_BGR24, SWS_BICUBIC, NULL, NULL,NULL);
sws_scale(imgConvertCtx, ((AVPicture*)pFrame)->data, ((AVPicture*)pFrame)->linesize, 0, pCodecCon->height, ((AVPicture *)pFrameRGB)->data, ((AVPicture *)pFrameRGB)->linesize);
av_frame_unref(pFrame);
av_frame_unref(pFrameRGB);
}
}
av_free_packet(&rx_pkt);
av_packet_unref(&rx_pkt);
}
//cvDestroyWindow("Cam1Video");
av_free_packet(&rx_pkt);
avcodec_close(pCodecCon);
av_free(pFrame);
av_free(pFrameRGB);
avformat_close_input(&pFormatCont);
I have read, the reason could be that the ffmpeg-Libs saves the incomming frames in the cache but the arm-processor isnt fast enough to process them. After like 4 minutes the system craches.
How could i solve the Problem.
one option could be to tell ffmpeg to act as frame grabber, also to read frames in real time, with the flag "-re". How can i set this Flag in the c++ source code. Or can anybody help me to solve that Problem.
Thank you very much
When I decode frames from avi file and then decode them in x264 and save to mp4 file, the fps of the output file is always 12,800. Therefore the file is played very fast. But, when I save the encoded in h264 frames in avi format and not mp4, so the fps is as I wanted - 25.
What could be the problem?
Here the code I wrote in VS2010:
#include "stdafx.h"
#include "inttypes.h"
extern "C" {
#include "libavcodec/avcodec.h"
#include "libavformat/avformat.h"
#include "libavutil/avutil.h"
#include <libswscale/swscale.h>
#include <libavutil/opt.h>
#include <libswscale/swscale.h>
#include <libavutil/imgutils.h>
}
#include <iostream>
using namespace std;
int main(int argc, char* argv[])
{
const char* inFileName = "C:\\000227_C1_GAME.avi";
const char* outFileName = "c:\\test.avi";
const char* outFileType = "avi";
av_register_all();
AVFormatContext* inContainer = NULL;
if(avformat_open_input(&inContainer, inFileName, NULL, NULL) < 0)
exit(1);
if(avformat_find_stream_info(inContainer, NULL) < 0)
exit(1);
// Find video stream
int videoStreamIndex = -1;
for (unsigned int i = 0; i < inContainer->nb_streams; ++i)
{
if (inContainer->streams[i] && inContainer->streams[i]->codec &&
inContainer->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO)
{
videoStreamIndex = i;
break;
}
}
if (videoStreamIndex == -1) exit(1);
AVFormatContext* outContainer = NULL;
if(avformat_alloc_output_context2(&outContainer, NULL, outFileType, outFileName) < 0)
exit(1);
// ----------------------------
// Decoder
// ----------------------------
AVStream const *const inStream = inContainer->streams[videoStreamIndex];
AVCodec *const decoder = avcodec_find_decoder(inStream->codec->codec_id);
if(!decoder)
exit(1);
if(avcodec_open2(inStream->codec, decoder, NULL) < 0)
exit(1);
// ----------------------------
// Encoder
// ----------------------------
AVCodec *encoder = avcodec_find_encoder(AV_CODEC_ID_H264);
if(!encoder)
exit(1);
AVStream *outStream = avformat_new_stream(outContainer, encoder);
if(!outStream)
exit(1);
avcodec_get_context_defaults3(outStream->codec, encoder);
// Construct encoder
if(outContainer->oformat->flags & AVFMT_GLOBALHEADER)
outStream->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
outStream->codec->coder_type = AVMEDIA_TYPE_VIDEO;
outStream->codec->pix_fmt = AV_PIX_FMT_YUV420P;
outStream->codec->width = inStream->codec->width;
outStream->codec->height = inStream->codec->height;
outStream->codec->codec_id = encoder->id;
outStream->codec->bit_rate = 500000;
//outStream->codec->rc_min_rate = 600000;
//outStream->codec->rc_max_rate = 800000;
outStream->codec->time_base.den = 25;
outStream->codec->time_base.num = 1;
outStream->codec->gop_size = 250; // Keyframe interval(=GOP length). Determines maximum distance distance between I-frames
outStream->codec->keyint_min = 25; // minimum GOP size
outStream->codec->max_b_frames = 3;//16; // maximum number of B-frames between non-B-frames
outStream->codec->b_frame_strategy = 1; // decides the best number of B-frames to use. Default mode in x264.
outStream->codec->scenechange_threshold = 40;
outStream->codec->refs = 6; // abillity to reference frames other than the one immediately prior to the current frame. specify how many references can be used.
outStream->codec->qmin = 0;//10;
outStream->codec->qmax = 69;//51;
outStream->codec->qcompress = 0.6;
outStream->codec->max_qdiff = 4;
outStream->codec->i_quant_factor = 1.4;//0.71;
outStream->codec->refs=1;//3;
outStream->codec->chromaoffset = -2;
outStream->codec->thread_count = 1;
outStream->codec->trellis = 1;
outStream->codec->me_range = 16;
outStream->codec->me_method = ME_HEX; //hex
outStream->codec->flags2 |= CODEC_FLAG2_FAST;
outStream->codec->coder_type = 1;
if(outStream->codec->codec_id == AV_CODEC_ID_H264)
{
av_opt_set(outStream->codec->priv_data, "preset", "slow", 0);
}
// Open encoder
if(avcodec_open2(outStream->codec, encoder, NULL) < 0)
exit(1);
// Open output container
if(avio_open(&outContainer->pb, outFileName, AVIO_FLAG_WRITE) < 0)
exit(1);
//close_o
AVFrame *decodedFrame = avcodec_alloc_frame();
if(!decodedFrame)
exit(1);
AVFrame *encodeFrame = avcodec_alloc_frame();
if(!encodeFrame)
exit(1);
encodeFrame->format = outStream->codec->pix_fmt;
encodeFrame->width = outStream->codec->width;
encodeFrame->height = outStream->codec->height;
if(av_image_alloc(encodeFrame->data, encodeFrame->linesize,
outStream->codec->width, outStream->codec->height,
outStream->codec->pix_fmt, 1) < 0)
exit(1);
av_dump_format(inContainer, 0, inFileName,0);
//Write header to ouput container
avformat_write_header(outContainer, NULL);
AVPacket decodePacket, encodedPacket;
int got_frame, len;
while(av_read_frame(inContainer, &decodePacket)>=0)
{
if (decodePacket.stream_index == videoStreamIndex)
{
len = avcodec_decode_video2(inStream->codec, decodedFrame, &got_frame, &decodePacket);
if(len < 0)
exit(1);
if(got_frame)
{
av_init_packet(&encodedPacket);
encodedPacket.data = NULL;
encodedPacket.size = 0;
if(avcodec_encode_video2(outStream->codec, &encodedPacket, decodedFrame, &got_frame) < 0)
exit(1);
if(got_frame)
{
if (outStream->codec->coded_frame->key_frame)
encodedPacket.flags |= AV_PKT_FLAG_KEY;
encodedPacket.stream_index = outStream->index;
if(av_interleaved_write_frame(outContainer, &encodedPacket) < 0)
exit(1);
av_free_packet(&encodedPacket);
}
}
}
av_free_packet(&decodePacket);
}
av_write_trailer(outContainer);
avio_close(outContainer->pb);
avcodec_free_frame(&encodeFrame);
avcodec_free_frame(&decodedFrame);
avformat_free_context(outContainer);
av_close_input_file(inContainer);
return 0;
}
The problem was with PTS and DTS of the packet. Before writing the packet to output( before av_interleaved_write_frame command) set PTS and DTS like this
if (encodedPacket.pts != AV_NOPTS_VALUE)
encodedPacket.pts = av_rescale_q(encodedPacket.pts, outStream->codec->time_base, outStream->time_base);
if (encodedPacket.dts != AV_NOPTS_VALUE)
encodedPacket.dts = av_rescale_q(encodedPacket.dts, outStream->codec->time_base, outStream->time_base);