I'm sure this question would have been asked before but I've searched and can't find anything specific to help a solution.
I'll start out outlining the initial concerns and if more indepth technical information is needed then I can give it. Hopefully there is enough information for the initial question(s).
I'm writing an app using c++ and directshow in visual studio 2010. The main project specification is for a live preview and, at any time of choosing, record the video to mpeg2 to harddrive then to dvd to be played in a standard dvd player, all the time the live preview is not to be interrupted.
The capturing seems a pretty trivial standard straight forward thing to do with directshow.
There are a couple of custom filters that i wrote. Nothing amazing but we wanted our own custom screen overlay information - time and date etc - this must be in the preview and the recorded file. I use the avi decompressor connected to the capture card video out pin, and connect the avi decompressor to my filter to give me an rgb image that i can manipulate. The output from this filter is then split via an inftee filter, one goes to the screen, the other goes into the ms mpeg2 encoder. The audio goes from the capture card audio out into the same mpeg2 encoder. Output from the mpeg2 encoder then goes to a file. That file then gets authored for dvd and burnt to dvd.
So my questions are...
Where and how would be the best place to allow starting and stopping of only mpeg2 file output, to be done via user action?
I have tried using smart tee filters - 1 for video and 1 for audio as the last filter BEFORE the mpeg2 encoder, then using the iamstreamcontrol interface to turn off the pins at the appropriate time. Should this cause any timing issues with the final mpeg2? as the output file will play via mplayer and vlc etc but doesnt get converted to be mpeg2 dvd compliant ( for testing - via any dvd authoring software - complaints of a broken file and somteimes gives time references ) - is it possible that time stamps in the file are a problem and giving an error? If the file is captured from the first moment that capture commences ( as opposed to say after 5 mins of streaming ) then everything is ok.
I did think of using the streambuffer route - http://msdn.microsoft.com/en-gb/library/windows/desktop/dd693041(v=vs.85).aspx - but I'm not sure on the best direction to takes things. It seems that are possibly a few choices for the best direction.
Any help and tips would be greatly appreciated. Especially if theres websites/books/information of DirectShow filters,pins,graphs and how they all flow together.
EDIT: I was thinking of making my own copy of the 'Smart Tee' filter that in that I would have 2 pins coming in - audio and video - and 4 out pins - 2 video ( 1 preview and 1 capture ) and 2 of the same for audio, but would I end up with the same issue? And what is the correct way to handle 'switching off' the capture pins of that custom filter. Would I be wasting my time to work on something like this? Is it a simple case of overriding the Active/Inactive methods of the output pin(s) and either send or not send the sample? I feel its not that easy?
Many thanks!
Where and how would be the best place to allow starting and stopping of only mpeg2 file output, to be done via user action?
For this kind of action I would recommend GMFBridge. Creating your own filter is not easy. GMFBridge allows you to use two separate graphs with a dynamic connection. Use the first graph for the preview and the second graph for the file output. And only connect the graphs after a user action.