Reputation: 856
I would really appreciate some help in this issue. I'm new to FFmpeg, therefore forgive me if this is a dumb question
I have an application streamming H264 raw data from a camera via UDP. That communication works fine. The problem is while trying to decoding
I simply want to decode that data I receive, in order to apply a few Computer Vision techniques to another project.
I got error -22 right after calling
avcodec_send_packet
Does any of you know what could it be? Here is part of my code
Mat DecodificadorH264::h264decode(char* rawH264, int size){
...
...
AVPacket av_packet;
av_init_packet(&av_packet);
uint8_t data;
data = *rawH264;
av_packet.data = &data;
av_packet.size = size;
cout << "size = "<<av_packet.size;
int frame_finished = 0;
l_pCodec = avcodec_find_decoder(AV_CODEC_ID_H264);
l_pAVCodecContext = avcodec_alloc_context3(l_pCodec);
l_pAVCodecContext->width = FRAME_WIDTH;
l_pAVCodecContext->height = FRAME_HEIGHT;
l_pAVCodecContext->extradata = NULL;
l_pAVCodecContext->pix_fmt = AV_PIX_FMT_YUV420P;
int av_return = avcodec_send_packet(l_pAVCodecContext,
(AVPacket*)&av_packet);
if (av_return == AVERROR(EINVAL))cout << "codec not opened, or it is an encoder" << endl;
....
av_return returns "AVERROR(EINVAL)", which is defined as -22
What could be the problem?
Thanks in advance!
Upvotes: 0
Views: 2538
Reputation: 3809
You need one more step to making a usable decode context: avcodec_open2()
codec = avcodec_find_decoder (AV_CODEC_ID_H264);
context = avcodec_alloc_context3 (codec);
AVDictionary *dict = nullptr;
avcodec_open2(context, codec, &dict);
Upvotes: 0
Reputation: 651
Well I see one problem so far:
uint8_t data;
data = *rawH264;
av_packet.data = &data; // all you did here is copy the first char from rawH264 into data
What you need is
av_packet.data = rawH264; // you might have to cast rawH264 to uint8_t* first
According to https://ffmpeg.org/doxygen/0.6/error_8h.html: #define AVERROR_INVALIDDATA AVERROR(EINVAL) Invalid data found when processing input.
So it makes sense the result is EINVAL, because you passed in &data, which is on the stack, and with size parameter, it was reading the stack as if it's actual encoded video data.
Upvotes: 1