[Qt-interest] Intercepting Phonon VideoWidget frames as they are pushed

[Trenton Tuggle]

I have an interesting challenge I've been working on:

I'm working on video processing, and am investigating using Phonon for  
platform-independent back-ends.

My interest is not the usual, though -- I have little interest in  
showing the video on-screen (although that would be nice, also) but  
primarily, I'm interested in obtaining the video frames.

Now I know it's easy to render() a widget (such as VideoWidget) onto a  
pixmap or other, but I'm really interested in obtaining the frames in  
a more automated way -- such as getting a signal when the next frame  
is ready, etc.  My thought is that since VideoWidget can be embedded  
in a graphics scene, it must be possible to intercept its painting  
events and redirect it to a qimage or something: but so far I've failed.

So that's my question: how can I embed a VideoWidget in something that  
can intercept paint events, redirect them to a QImage, then signal a  
processor that a frame is ready for processing?

I've tried stuff along these lines:

http://www.kdedevelopers.org/node/3644

The thought was that I'll create a Phonon path to play on a  
VideoWidget embedded in something like that which would capture the  
frame, then signal my processor....

But so far, I've been unable to get the widget I create to receive any  
paint events, or intercept anything at all -- it doesn't even get  
resized properly when a video is loaded!

If anyone has any advice or input on this, I'd really appreciate it!

Thanks in advance!

Trent Tuggle <Trent at Tuggle.Org>