You might have to resample the audio so that you can get nice even sector sizes for the audio per frame perhaps? 2,940 samples/frame@30fps could be handled but messy whereas 3072 would be ideal. So resample 44.1k to 46.08k might make it easier. Having a separate file or a "file" embedded in the video would be easier still. You only lose a millisecond or so switching to another block read using sector pointers for video and audio.
I've got this working with PNut too now.
Also, the cogless version of sdspi can now run the videos (although one needs 300 MHz clock to do it).
Cogless speed much improved after removing a REP loop from the inline assembly.
Comments
For my audio buffers I use a size of 3072 x2 to maintain smooth audio between frames.
edit: 44.1kHz 16-bit mono samples btw
Need to get A & V in sync...
Not sure that 71 is optimum, but sounds good.
I'm trying to post some example videos to my website. We'll see if that works... They are too big to attach here....
Update. Some video settings were hard coded to this specific video. An update to the code is in the top post.
http://www.rayslogic.com/Propeller2/P2Video/P2Video.html
But, the sample video links don't work. Not sure why not...
Download works now... Seems I had to .zip it up...
Maybe I’ll try to figure out how pixmix works...
This one has some dialog:
Also, the cogless version of sdspi can now run the videos (although one needs 300 MHz clock to do it).
Cogless speed much improved after removing a REP loop from the inline assembly.