I'm currently realizing a MJPEG decoder and it looks it doesn't act at all as the other decoders.
I used MPEG4 and H264 and after calling Init, the information about the width and height were filled into VideoDecoderParams.info.clip_info.width|height but the MJPEG doesn't fill the information.
I need them to initialize the output buffer... so for now I use this workaround :
int nchannels;
int precision;
JSS sampling;
JCOLOR color;
CJPEGDecoder* m_dec;
m_dec = new CJPEGDecoder;
m_dec->SetSource((Ipp8u*)p_dataIn.GetDataPointer(),p_dataIn.GetDataSize());
m_dec->ReadHeader(&vdp.info.clip_info.width, &vdp.info.clip_info.height, &nchannels,&color,&sampling,&precision);
Furthermore, GetFrame is supposed to return the next decoded frame. So each time we use GetFrame it supposed to use the next frame.
With MJPEG, the datapointer doesn't seem to be moved to the next frame...
In file codec/jpeg/src/umc_mjpeg_video_decoder.cpp:397 you change the data size here :
in->SetDataSize(nBytesRemains);
But I think you would have to do a MoveDataPointer somewhere instead of modifying the DataSize like that.
So after a GetFrame I need to do something like :
p_dataIn.SetBufferPointer( (Ipp8u*)p_dataIn.GetBufferPointer() + p_dataIn.GetBufferSize() - p_dataIn.GetDataSize(), p_dataIn.GetDataSize());
Do I forget something to prevent the use of these tweaks?
Thank you
