>No, this is wrong!  You should not be creating/using a "H264VideoStreamFramer" 
>at all.  That class should be used *only* when the input is a >byte stream 
>(e.g., from a file).  If - as in your case - the input is a discrete sequence 
>of NAL units (i.e., one NAL unit at a time), then you should >use a 
>"H264VideoStreamDiscreteFramer" only.  So, you should replace the line

>             return H264VideoStreamFramer::createNew(envir(), discFramer); with

>             return discFramer;


Ok, doing that the problem with fMaxSize is fixed and its value is the one I 
have specified in the OutPacketBuffer::maxSize

However, in the player I don't see anything just the 'loading screen'.

Because of the fact that I should not include start codes in the NAL Units I 
deactivate them in the encoder

According with my encoder specification 
http://docs.nvidia.com/cuda/samples/3_Imaging/cudaEncode/doc/nvcuvenc.pdf p.28

I have few options for this:

            0 implies that the encoder will add the start codes

           1, 2, 4: length prefixed NAL units of size 1, 2, or 4 bytes


If I set up the parameter to 0 the Discreteframer complains with the following 
message 'H264VideoStreamDiscreteFramer error: MPEG 'start code' seen in the 
input\n";'     I guess that's expected because I should not include start codes 
at this point all clear. However, with the parameter in the encoder set to 1, 2 
or  4 it didn't complain at all but I still do not visualize anything in the 
player.

If I keep using the H264VideoStreamFramer as I was using before -I know it is 
wrong- with encoder parameter set to 0 -start codes-- I visualize the player 
with artifacts as I already explained in previous posts. Meanwhile with the 
parameter sets to 1,2 or 4 I do not visualize anything at all means similar 
results that what I get when I'm using just the discrete framer.

I wonder why are the implications with start codes or prefixed NAL units size 
and the discreteframer..


Pablo
_______________________________________________
live-devel mailing list
[email protected]
http://lists.live555.com/mailman/listinfo/live-devel

Reply via email to