[Live-devel] Basic question

[Bill Russell]

At 03:53 PM 11/10/2003 -0800, you wrote:

>> From my encoder, I can choose to bring in full MPEG frames or, blocks of 
>> say, 1400 bytes. Which would be more efficient. In the latter case I 
>> have to use MPEG4VideoStreamFramer, so is it true that that would be 
>> less efficient?
>
>Actually, you'll need to use a "MPEG4VideoStreamFramer" in either case, 
>because the 'sink' object ("MPEG4ESVideoRTPSink") requires that its input 
>source be a "MPEG4VideoStreamFramer" (because it looks for special MPEG-4 
>specific parameters that only "MPEG4VideoStreamFramer" supplies).
>
>However, you still have a choice between having your 
>"MPEG4VideoStreamFramer" read full MPEG frames, or fixed size (e.g., 1400 
>byte) blocks.  Both will work, but I suggest reading full MPEG frames, as 
>this will be slightly more efficient (by making the framer's job easier).

I also would like to send JPEG video through the Live server (currently our 
camera sends JPEG frames to a custom ActiveX). The frames are typically up 
to 50KBytes in size. If I derive a class from JPEGVideoSource and use the 
deliverFrame() function, it appears that I must parse through the JPEG 
frame based on the fMaxSize parameter (over multiple calls) which is 
specified by the caller. I can't specify the amount of data I'd like to 
give the caller (in order to give an entire JPEG frame at once) - right? 
This will also apply to my decision to deal with MPEG data as full frames.

Bill