Please help to understand pre_push functionality in GStVideoEncoderClass ?

[Tim Müller]

On Mon, 2016-02-08 at 07:40 -0800, ssshukla26 wrote:

Hi,

> I thought that handle_frame and pre_push functions works as top half
> and bottom half of an encoding process.

Not sure why you thought that, I don't see anything in the
documentation suggesting that's how it's supposed to be used, and none
of the existing H.264 encoders I checked implement a pre_push function
either.

The pre_push function is called from gst_video_encoder_finish_frame(),
so only once *you* call gst_video_encoder_finish_frame(). It's main
purpose it to fix up the output in some way, if needed. Usually it's
not needed. (For example, one could convert the data from avc to byte-
stream format or vice versa, or add a prefix or something, or
additional metadata.)

> *2) under pre_push function I copied the encoded data from the queue
> to the
> output buffer*
> 
> static GstFlowReturn *gst_myh264encoder_pre_push* (GstVideoEncoder
> *encoder,GstVideoCodecFrame *frame)
> {
>     GstFlowReturn ret = GST_FLOW_OK;
>   
>     if(*fill_output_buffer*(encoder,frame))
>         ret = GST_FLOW_ERROR;
> 
>     if(GST_FLOW_OK == ret)
>     {   
>         ret = gst_video_encoder_finish_frame (encoder,frame);
>     }   
> 
>     return ret;
> }

Why don't you do this in the handle_frame() function?

Cheers
 -Tim

-- 
Tim Müller, Centricular Ltd - http://www.centricular.com