IMileti

Hi,

I need to encode 2 live video and one live audio streams in real time into one asf file. So i basically have an asf writer filter setup with an appropriate profile. Btw, I know this is a very demanding operation. On weaker computers, the resulting asf file has strange audio artifacts and disortions. Im guessing this is because by the time the asf writer encodes two video frames the audio sample may be late and is dropped. So I started wondering if the asf writer encodes all streams in one thread, would i maybe get better preformance if i move the encoding out of the asf writer and into seperate DMO filters upstream from the asf writer. That way each stream should encode in its own thread and my audio problems should stop (i really dont care if a video frame gets dropped occasionally, its very important that the audio stream is good).

I would appreciate any comments, suggestions, ...
Thanks
Ivan



Re: DirectShow Development directshow dmo asf writer

Chris P.

I believe it uses one thread per pin for encoding.

The video encoding is likely spiking the CPU in this case, so I would suggest lowering the encoder complexity in your profile until you see the CPU usage drop to a reasonable level.