FFMPEG C api h.264 encoding / MPEG2 ts streaming problems - c++11

Question

Class prototype is as follows:
#ifndef _FULL_MOTION_VIDEO_STREAM_H_
#define _FULL_MOTION_VIDEO_STREAM_H_
#include <memory>
#include <string>
#ifndef INT64_C
# define INT64_C(c) (c ## LL)
# define UINT64_C(c) (c ## ULL)
#endif
extern "C"
{
#include "libavutil/opt.h"
#include "libavcodec/avcodec.h"
#include "libavutil/channel_layout.h"
#include "libavutil/common.h"
#include "libavutil/imgutils.h"
#include "libavutil/mathematics.h"
#include "libavutil/samplefmt.h"
#include "libavformat/avformat.h"
#include <libavutil/timestamp.h>
#include <libswscale/swscale.h>
#include <libswresample/swresample.h>
}
class FMVStream
{
public:
struct OutputStream
{
OutputStream() :
st(0),
next_pts(0),
samples_count(0),
frame(0),
tmpFrame(0),
sws_ctx(0)
{
}
AVStream *st;
/* pts of the next frame that will be generated */
int64_t next_pts;
int samples_count;
AVFrame *frame;
AVFrame *tmpFrame;
struct SwsContext *sws_ctx;
};
///
/// Constructor
///
FMVStream();
///
/// Destructor
///
~FMVStream();
///
/// Frame encoder helper function
///
/// Encodes a raw RGB frame into the transport stream
///
int EncodeFrame(uint8_t* frame);
///
/// Frame width setter
///
void setFrameWidth(int width);
///
/// Frame width getter
///
int getFrameWidth() const;
///
/// Frame height setter
///
void setFrameHeight(int height);
///
/// Frame height getter
///
int getFrameHeight() const;
///
/// Stream address setter
///
void setStreamAddress(const std::string& address);
///
/// Stream address getter
///
std::string getStreamAddress() const;
private:
///
/// Video Stream creation
///
AVStream* initVideoStream(AVFormatContext* oc);
///
/// Raw frame transcoder
///
/// This will convert the raw RGB frame to a raw YUV frame necessary for h.264 encoding
///
void CopyFrameData(uint8_t* src_frame);
///
/// Video frame allocator
///
AVFrame* AllocPicture(PixelFormat pix_fmt, int width, int height);
///
/// Debug print helper function
///
void print_sdp(AVFormatContext **avc, int n);
///
/// Write the frame to the stream
///
int write_frame(AVFormatContext *fmt_ctx, const AVRational *time_base, AVStream *st, AVPacket *pkt);
///
/// initialize the frame data
///
void initFrame();
// formatting data needed for output streaming and the output container (MPEG 2 TS)
AVOutputFormat* format;
AVFormatContext* format_ctx;
// structure container for our video stream
OutputStream stream;
AVIOContext* io_ctx;
std::string streamFilename;
int frameWidth;
int frameHeight;
};
#endif
This block starts the class declaration.
#include "FullMotionVideoStream.h"
#include <stdexcept>
#include <iostream>
FMVStream::FMVStream()
: format(0),
format_ctx(0),
stream(),
io_ctx(0),
streamFilename("test.mpeg"),
frameWidth(640),
frameHeight(480)
{
// Register all formats and codecs
av_register_all();
avcodec_register_all();
// Init networking
avformat_network_init();
// Find format
this->format = av_guess_format("mpegts", NULL, NULL);
// allocate the AVFormatContext
this->format_ctx = avformat_alloc_context();
if (!this->format_ctx)
{
throw std::runtime_error("avformat_alloc_context failed");
}
this->format_ctx->oformat = this->format;
//sprintf_s(this->format_ctx->filename, sizeof(this->format_ctx->filename), "%s", this->streamFilename.c_str());
this->stream.st = initVideoStream(this->format_ctx);
this->initFrame();
// Allocate AVIOContext
int ret = avio_open(&this->io_ctx, this->streamFilename.c_str(), AVIO_FLAG_WRITE);
if (ret != 0)
{
throw std::runtime_error("avio_open failed");
}
this->format_ctx->pb = this->io_ctx;
// Print some debug info about the format
av_dump_format(this->format_ctx, 0, NULL, 1);
// Begin the output by writing the container header
avformat_write_header(this->format_ctx, NULL);
AVFormatContext* ac[] = { this->format_ctx };
print_sdp(ac, 1);
}
FMVStream::~FMVStream()
{
av_write_trailer(this->format_ctx);
avcodec_close(this->stream.st->codec);
avio_close(io_ctx);
avformat_free_context(this->format_ctx);
av_frame_free(&this->stream.frame);
av_free(this->format);
}
AVFrame* FMVStream::AllocPicture(PixelFormat pix_fmt, int width, int height)
{
// Allocate a frame
AVFrame* frame = av_frame_alloc();
if (frame == nullptr)
{
throw std::runtime_error("avcodec_alloc_frame failed");
}
if (av_image_alloc(frame->data, frame->linesize, width, height, pix_fmt, 1) < 0)
{
throw std::runtime_error("av_image_alloc failed");
}
frame->width = width;
frame->height = height;
frame->format = pix_fmt;
return frame;
}
void FMVStream::print_sdp(AVFormatContext **avc, int n)
{
char sdp[2048];
av_sdp_create(avc, n, sdp, sizeof(sdp));
printf("SDP:\n%s\n", sdp);
fflush(stdout);
}
AVStream* FMVStream::initVideoStream(AVFormatContext *oc)
{
AVStream* st = avformat_new_stream(oc, NULL);
if (st == nullptr)
{
std::runtime_error("Could not alloc stream");
}
AVCodec* codec = avcodec_find_encoder(AV_CODEC_ID_H264);
if (codec == nullptr)
{
throw std::runtime_error("couldn't find mpeg2 encoder");
}
st->codec = avcodec_alloc_context3(codec);
st->codec->codec_id = AV_CODEC_ID_H264;
st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
st->codec->bit_rate = 400000;
st->codec->width = this->frameWidth;
st->codec->height = this->frameHeight;
st->time_base.num = 1;
st->time_base.den = 30;
st->codec->framerate.num = 1;
st->codec->framerate.den = 30;
st->codec->max_b_frames = 2;
st->codec->gop_size = 12;
st->codec->pix_fmt = PIX_FMT_YUV420P;
st->id = oc->nb_streams - 1;
if (oc->oformat->flags & AVFMT_GLOBALHEADER)
{
st->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
}
// option setup for the codec
av_opt_set(st->codec->priv_data, "profile", "baseline", AV_OPT_SEARCH_CHILDREN);
if (avcodec_open2(st->codec, codec, NULL) < 0)
{
throw std::runtime_error("avcodec_open failed");
}
return st;
}
void FMVStream::initFrame()
{
// Allocate a tmp frame for converting our raw RGB data to YUV for encoding
this->stream.tmpFrame = this->AllocPicture(PIX_FMT_RGB24, this->frameWidth, this->frameHeight);
// Allocate a main frame
this->stream.frame = this->AllocPicture(PIX_FMT_YUV420P, this->frameWidth, this->frameHeight);
}
This block is attempting to convert from the raw RGB to our needed YUV format for h.264 encoding.
void FMVStream::CopyFrameData(uint8_t* data)
{
// fill image with our raw RGB data
//avpicture_alloc((AVPicture*)this->stream.tmpFrame, PIX_FMT_RGB24, this->stream.st->codec->width, this->stream.st->codec->height);
int numBytes = avpicture_get_size(PIX_FMT_RGB24, this->stream.st->codec->width, this->stream.st->codec->height);
uint8_t* buffer = (uint8_t*) av_malloc(numBytes * sizeof(uint8_t));
avpicture_fill((AVPicture*)this->stream.tmpFrame, buffer, PIX_FMT_RGB24, this->stream.st->codec->width, this->stream.st->codec->height);
for (int y = 0; y < this->stream.st->codec->height; y++)
{
for (int x = 0; x < this->stream.st->codec->width; x++)
{
int offset = 3 * (x + y * this->stream.st->codec->width);
this->stream.tmpFrame->data[0][offset + 0] = data[x + y * this->stream.st->codec->width]; // R
this->stream.tmpFrame->data[0][offset + 1] = data[x + y * this->stream.st->codec->width + 1]; // G
this->stream.tmpFrame->data[0][offset + 2] = data[x + y * this->stream.st->codec->width + 2]; // B
}
}
// convert the RGB frame to a YUV frame using the sws Context
this->stream.sws_ctx = sws_getContext(this->stream.st->codec->width, this->stream.st->codec->height, PIX_FMT_RGB32, this->stream.st->codec->width, this->stream.st->codec->height, PIX_FMT_YUV420P, SWS_FAST_BILINEAR, NULL, NULL, NULL);
// use the scale function to transcode this raw frame to the correct type
sws_scale(this->stream.sws_ctx, this->stream.tmpFrame->data, this->stream.tmpFrame->linesize, 0, this->stream.st->codec->height, this->stream.frame->data, this->stream.frame->linesize);
}
This is the block that encodes the raw data to h.264, and then send it out the Mpeg2 ts. I believe the problem lies within this block. I can put a break point in my write frame block and see that frames are being written, however, opening the resulting file in VLC results in a blank video. The file is approx 2Mb.
int FMVStream::EncodeFrame(uint8_t* data)
{
AVCodecContext* c = this->stream.st->codec;
AVRational one;
one.den = one.num = 1;
// check to see if we want to keep writing frames we can probably change this to a toggle switch
if (av_compare_ts(this->stream.next_pts, this->stream.st->codec->time_base, 10, one) >= 0)
{
this->stream.frame = nullptr;
}
else
{
// Convert and load the frame data into the AVFrame struct
CopyFrameData(data);
}
// setup the timestamp stepping
AVPacket pkt = { 0 };
av_init_packet(&pkt);
this->stream.frame->pts = (int64_t)((1.0 / this->stream.st->codec->framerate.den) * 90000.0 * this->stream.next_pts++);
int gotPacket, out_size, ret;
out_size = avcodec_encode_video2(c, &pkt, this->stream.frame, &gotPacket);
if (gotPacket == 1)
{
ret = write_frame(this->format_ctx, &c->time_base, this->stream.st, &pkt);
}
else
{
ret = 0;
}
if (ret < 0)
{
std::cerr << "Error writing video frame" << std::endl;
}
av_free_packet(&pkt);
return ((this->stream.frame != nullptr) || gotPacket) ? 0 : 1;
}
int FMVStream::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;
return av_interleaved_write_frame(fmt_ctx, pkt);
}
void FMVStream::setFrameWidth(const int width)
{
this->frameWidth = width;
}
int FMVStream::getFrameWidth() const
{
return this->frameWidth;
}
void FMVStream::setFrameHeight(const int height)
{
this->frameHeight = height;
}
int FMVStream::getFrameHeight() const
{
return this->frameHeight;
}
void FMVStream::setStreamAddress(const std::string& address)
{
this->streamFilename = address;
}
std::string FMVStream::getStreamAddress() const
{
return this->streamFilename;
}
Here is the Main function.
#include "FullMotionVideoStream.h"
#include <iostream>
#include <thread>
#include <chrono>
int main(int argc, char** argv)
{
FMVStream* fmv = new FMVStream;
fmv->setFrameWidth(640);
fmv->setFrameHeight(480);
std::cout << "Streaming Address: " << fmv->getStreamAddress() << std::endl;
// create our alternating frame of black and white to test the streaming functionality
uint8_t white[640 * 480 * sizeof(uint8_t) * 3];
uint8_t black[640 * 480 * sizeof(uint8_t) * 3];
std::memset(white, 255, 640 * 480 * sizeof(uint8_t) * 3);
std::memset(black, 0, 640 * 480 * sizeof(uint8_t)* 3);
for (auto i = 0; i < 100; i++)
{
auto ret = fmv->EncodeFrame(white);
if (ret != 0)
{
std::cerr << "There was a problem encoding the frame: " << i << std::endl;
}
std::this_thread::sleep_for(std::chrono::milliseconds(10));
}
for (auto i = 0; i < 100; i++)
{
auto ret = fmv->EncodeFrame(black);
if (ret != 0)
{
std::cerr << "There was a problem encoding the frame: " << i << std::endl;
}
std::this_thread::sleep_for(std::chrono::milliseconds(10));
}
delete fmv;
}
Here is the resultant output via the console / my print SDP function.
[libx264 # 000000ac95f58440] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2
AVX FMA3 AVX2 LZCNT BMI2
[libx264 # 000000ac95f58440] profile Constrained Baseline, level 3.0
Output #0, mpegts, to '(null)':
Stream #0:0: Video: h264 (libx264), yuv420p, 640x480, q=-1--1, 400 kb/s, 30
tbn
SDP:
v=0
o=- 0 0 IN IP4 127.0.0.1
s=No Name
t=0 0
a=tool:libavformat 56.23.104
m=video 0 RTP/AVP 96
b=AS:400
a=rtpmap:96 H264/90000
a=fmtp:96 packetization-mode=1
a=control:streamid=0
Streaming Address: test.mpeg
[libx264 # 000000ac95f58440] frame I:45 Avg QP: 0.51 size: 1315
[libx264 # 000000ac95f58440] frame P:136 Avg QP: 0.29 size: 182
[libx264 # 000000ac95f58440] mb I I16..4: 99.7% 0.0% 0.3%
[libx264 # 000000ac95f58440] mb P I16..4: 0.1% 0.0% 0.1% P16..4: 0.1% 0.0
% 0.0% 0.0% 0.0% skip:99.7%
[libx264 # 000000ac95f58440] final ratefactor: -68.99
[libx264 # 000000ac95f58440] coded y,uvDC,uvAC intra: 0.5% 0.5% 0.5% inter: 0.0%
0.1% 0.1%
[libx264 # 000000ac95f58440] i16 v,h,dc,p: 96% 0% 3% 0%
[libx264 # 000000ac95f58440] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 1% 10% 85% 0% 3%
0% 1% 0% 0%
[libx264 # 000000ac95f58440] i8c dc,h,v,p: 100% 0% 0% 0%
[libx264 # 000000ac95f58440] ref P L0: 46.8% 25.2% 28.0%
[libx264 # 000000ac95f58440] kb/s:0.03
I know there are probably many issues with this program, I am very new with FFMPEG and multimedia programming in general. Ive used many pieces of code found through searching google/ stack overflow to get to this point as is. The file has a good size but comes up as length 0.04 tells me that my time stamping must be broken between the frames / pkts, but I am unsure on how to fix this issue.
I tried inspecting the file with ffmpeg.exe using ffmpeg -i and outputting to a regular TS. It seems my code works more then I originally intended however, I am simply trying to output a bunch of all white frames.
ffmpeg -i test.mpeg test.ts
ffmpeg version N-70125-g6c9537b Copyright (c) 2000-2015 the FFmpeg developers
built with gcc 4.9.2 (GCC)
configuration: --disable-static --enable-shared --enable-gpl --enable-version3
--disable-w32threads --enable-avisynth --enable-bzlib --enable-fontconfig --ena
ble-frei0r --enable-gnutls --enable-iconv --enable-libass --enable-libbluray --e
nable-libbs2b --enable-libcaca --enable-libfreetype --enable-libgme --enable-lib
gsm --enable-libilbc --enable-libmodplug --enable-libmp3lame --enable-libopencor
e-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enabl
e-librtmp --enable-libschroedinger --enable-libsoxr --enable-libspeex --enable-l
ibtheora --enable-libtwolame --enable-libvidstab --enable-libvo-aacenc --enable-
libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-l
ibwebp --enable-libx264 --enable-libx265 --enable-libxavs --enable-libxvid --ena
ble-lzma --enable-decklink --enable-zlib
libavutil 54. 19.100 / 54. 19.100
libavcodec 56. 26.100 / 56. 26.100
libavformat 56. 23.104 / 56. 23.104
libavdevice 56. 4.100 / 56. 4.100
libavfilter 5. 11.101 / 5. 11.101
libswscale 3. 1.101 / 3. 1.101
libswresample 1. 1.100 / 1. 1.100
libpostproc 53. 3.100 / 53. 3.100
Input #0, mpegts, from 'test.mpeg':
Duration: 00:00:00.04, start: 0.000000, bitrate: 24026 kb/s
Program 1
Metadata:
service_name : Service01
service_provider: FFmpeg
Stream #0:0[0x100]: Video: h264 (Constrained Baseline) ([27][0][0][0] / 0x00
1B), yuv420p, 640x480, 25 fps, 25 tbr, 90k tbn, 50 tbc
File 'test.ts' already exists. Overwrite ? [y/N] y
Output #0, mpegts, to 'test.ts':
Metadata:
encoder : Lavf56.23.104
Stream #0:0: Video: mpeg2video, yuv420p, 640x480, q=2-31, 200 kb/s, 25 fps,
90k tbn, 25 tbc
Metadata:
encoder : Lavc56.26.100 mpeg2video
Stream mapping:
Stream #0:0 -> #0:0 (h264 (native) -> mpeg2video (native))
Press [q] to stop, [?] for help
frame= 3 fps=0.0 q=2.0 Lsize= 9kB time=00:00:00.08 bitrate= 883.6kbits/
s dup=0 drop=178
video:7kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing ove
rhead: 22.450111%

Answer 1

avpicture_fill does not do what you think it does. It does not fill the picture using data from ptr, as the source, It fills the picture using ptr as the destination. So basically, you are clearing the image before you encode it.

Answer 2

on your av_packet_rescale_ts(pkt, *time_base, st->time_base);
you are using AvCodecContext::time_base
and you set AvCodecContext::framerate instead.
st->time_base.num = 1;
st->time_base.den = 30;
st->codec->framerate.num = 1;
st->codec->framerate.den = 30;
change to:
st->time_base.num = 1;
st->time_base.den = 30;
st->codec->time_base = st->time_base;