encode using vaapi

ffmpeg-encoder3.cpp

// from https://stackoverflow.com/questions/59666753/encode-video-from-c-using-libavcodec-and-vaapi

#ifndef ENCODER_H
#define ENCODER_H
#include <cassert>

extern "C" {
#include <libavcodec/avcodec.h>
#include <libavformat/avformat.h>
#include <libavutil/opt.h>
#include <libavutil/hwcontext.h>

#include <libavfilter/avfilter.h>
#include <libavfilter/buffersink.h>
#include <libavfilter/buffersrc.h>
}

class Encoder
{
public:
    Encoder(const bool hwAccel);
    void addFrame(AVFrame* frame);
    void flush();

    static constexpr int s_width = 640;
    static constexpr int s_height = 480;
    static constexpr int s_fps = 25;
private:
    void setup();
    void setupEncoder();
    void initFilters();
    void initInputFilters(AVFilterInOut* inputs);
    void initOutputFilters(AVFilterInOut* outputs);
    void filterFrame(AVFrame* inFrame, AVFrame* outFrame);
    void encodeFrame(AVFrame* frame);

    // members
    int m_frameId = 1;
    const bool m_hardwareAcceleration = false;

    AVCodecContext* m_encoder = nullptr;
    AVFormatContext* m_muxer = nullptr;
    AVStream* m_avStream = nullptr;
    AVBufferRef* m_device = nullptr;

    AVFrame* m_hwFrame = nullptr;

    AVFilterGraph* m_filterGraph = nullptr;
    AVFilterContext* m_bufferSrc = nullptr;
    AVFilterContext* m_bufferSink = nullptr;
    AVFilterContext* m_formatFilter = nullptr;
};

#endif // ENCODER_H


#include "encoder.h"

extern "C" {

static enum AVPixelFormat get_vaapi_format(AVCodecContext*, const enum AVPixelFormat *pix_fmts)
{
    const enum AVPixelFormat *p;
    for (p = pix_fmts; *p != AV_PIX_FMT_NONE; p++) {
        if (*p == AV_PIX_FMT_VAAPI)
            return *p;
    }
    fprintf(stderr, "Failed to get HW surface format.\n");
    return AV_PIX_FMT_NONE;
}

}

Encoder::Encoder(const bool hwAccel)
  : m_hardwareAcceleration(hwAccel)
{
    setup();
}
void Encoder::addFrame(AVFrame* frame)
{
    AVFrame* frameToEncode = frame;
    if(m_hardwareAcceleration) {
        filterFrame(frame, m_hwFrame);
        assert(m_hwFrame->format == AV_PIX_FMT_VAAPI);
        frameToEncode = m_hwFrame;
    }

    frameToEncode->pts = m_frameId++;
    encodeFrame(frameToEncode);
}
void Encoder::flush()
{
    encodeFrame(nullptr);
    av_write_trailer(m_muxer);
}

void Encoder::setup()
{
    assert(avformat_alloc_output_context2(&m_muxer, nullptr, "matroska", nullptr) == 0);
    assert(m_muxer != nullptr);

    setupEncoder();

    m_avStream = avformat_new_stream(m_muxer, nullptr);
    assert(m_avStream != nullptr);
    m_avStream->id = m_muxer->nb_streams-1;
    m_avStream->time_base = m_encoder->time_base;

    // Some formats want stream headers to be separate.
    if(m_muxer->oformat->flags & AVFMT_GLOBALHEADER)
        m_encoder->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;

    assert(avcodec_parameters_from_context(m_avStream->codecpar, m_encoder) == 0);
    assert(avio_open(&m_muxer->pb, m_hardwareAcceleration? "hardware.mkv": "software.mkv", AVIO_FLAG_WRITE) == 0);
    assert(avformat_write_header(m_muxer, nullptr) == 0);
}
void Encoder::setupEncoder()
{
    const char* encoderName = m_hardwareAcceleration? "h264_vaapi": "libx264";
    AVCodec* videoCodec = avcodec_find_encoder_by_name(encoderName);
    m_encoder = avcodec_alloc_context3(videoCodec);
    m_encoder->bit_rate = s_width * s_height * s_fps * 2;
    m_encoder->width = s_width;
    m_encoder->height = s_height;
    m_encoder->time_base = (AVRational){1, s_fps};
    m_encoder->framerate = (AVRational){s_fps, 1};

    m_encoder->gop_size = s_fps;  // have at least 1 I-frame per second
    m_encoder->max_b_frames = 1;
    m_encoder->pix_fmt = AV_PIX_FMT_YUV420P;

    if(m_hardwareAcceleration) {
        m_encoder->pix_fmt = AV_PIX_FMT_VAAPI;
        m_encoder->get_format = get_vaapi_format;

        assert(av_hwdevice_ctx_create(&m_device, AV_HWDEVICE_TYPE_VAAPI, "/dev/dri/renderD128", nullptr, 0) == 0);
        const AVHWDeviceContext* deviceCtx = (AVHWDeviceContext*) m_device->data;
        assert(deviceCtx->type == AV_HWDEVICE_TYPE_VAAPI);

        initFilters();

        m_encoder->hw_device_ctx = nullptr;
        m_encoder->hw_frames_ctx = av_buffer_ref(av_buffersink_get_hw_frames_ctx(m_bufferSink));
    }

    assert(avcodec_open2(m_encoder, videoCodec, nullptr) == 0);

    if(m_hardwareAcceleration) {
        m_encoder->hw_device_ctx = av_hwframe_ctx_alloc(m_device);
        m_hwFrame = av_frame_alloc();
        av_hwframe_get_buffer(m_encoder->hw_device_ctx, m_hwFrame, 0);
    }

    m_muxer->video_codec_id = videoCodec->id;
    m_muxer->video_codec = videoCodec;
}
void Encoder::initFilters()
{
    AVFilterInOut* inputs = nullptr;
    AVFilterInOut* outputs = nullptr;
    m_filterGraph = avfilter_graph_alloc();
    assert(avfilter_graph_parse2(m_filterGraph, "format=nv12,hwupload", &inputs, &outputs) == 0);

    for(unsigned i=0; i<m_filterGraph->nb_filters; i++) {
        m_filterGraph->filters[i]->hw_device_ctx = av_buffer_ref(m_device);
        assert(m_filterGraph->filters[i]->hw_device_ctx != nullptr);
    }

    initInputFilters(inputs);
    initOutputFilters(outputs);

    assert(avfilter_graph_config(m_filterGraph, nullptr) == 0);
}
void Encoder::initInputFilters(AVFilterInOut* inputs)
{
    assert(inputs != nullptr);
    assert(inputs->next == nullptr);

    char args[512];
    snprintf(args, sizeof(args),
            "video_size=%dx%d:pix_fmt=%d:time_base=%d/%d:pixel_aspect=%d/%d",
            s_width, s_height, AV_PIX_FMT_YUV420P,
            1, s_fps,
            1, 1);

    assert(avfilter_graph_create_filter(&m_bufferSrc, avfilter_get_by_name("buffer"), "in",
                                        args, nullptr, m_filterGraph) == 0);
    assert(avfilter_link(m_bufferSrc, 0, inputs->filter_ctx, inputs->pad_idx) == 0);
}
void Encoder::initOutputFilters(AVFilterInOut* outputs)
{
    assert(outputs != nullptr);
    assert(outputs->next == nullptr);

    assert(avfilter_graph_create_filter(&m_bufferSink, avfilter_get_by_name("buffersink"), "out",
                                       nullptr, nullptr, m_filterGraph) == 0);
    assert(avfilter_graph_create_filter(&m_formatFilter, avfilter_get_by_name("format"), "format",
                                       "vaapi_vld", nullptr, m_filterGraph) == 0);
    assert(avfilter_link(outputs->filter_ctx, outputs->pad_idx, m_formatFilter, 0) == 0);
    assert(avfilter_link(m_formatFilter, 0, m_bufferSink, 0) == 0);
}
void Encoder::filterFrame(AVFrame* inFrame, AVFrame* outFrame)
{
    assert(av_buffersrc_add_frame_flags(m_bufferSrc, inFrame, AV_BUFFERSRC_FLAG_KEEP_REF) == 0);
    assert(av_buffersink_get_frame(m_bufferSink, outFrame) == 0);
}
void Encoder::encodeFrame(AVFrame* frame)
{
    assert(avcodec_send_frame(m_encoder, frame) == 0);

    AVPacket packet;
    av_init_packet(&packet);
    int ret = 0;
    while(ret >= 0) {
        ret = avcodec_receive_packet(m_encoder, &packet);
        if(ret == AVERROR(EAGAIN) || ret == AVERROR_EOF) {
            return;  // nothing to write
        }
        assert(ret >= 0);

        av_packet_rescale_ts(&packet, m_encoder->time_base, m_avStream->time_base);
        packet.stream_index = m_avStream->index;
        av_interleaved_write_frame(m_muxer, &packet);
        av_packet_unref(&packet);
    }
}