I recently convinced the video guy for a play I'm sound designing to use PD instead of Max/MSP/Jitter so that we can work together and run the audio and video off of the same laptop. The problem is that I don't have a lot of experience doing video in PD or using PD on OSX, but I've been doing PD on linux for live audio processing for about 4 years.
He's done about a month's worth of research into using Jitter, so I want to make the transition as easy as possible, and perhaps win another PD convert. Mostly what we need to do is record video with background subtraction, and then play multiple layers of it back with some effects so that the actor can interact with himself. Is it more computationally efficient to use straight PDP, straight GEM, or some combination? I'd like to use GEM's [pix_video] object, because it can talk to the OSX video system, but I'd imagine the PDP video processing stuff is more efficient, as it's not concerned with texturing a rectangle and displaying it in 3D. We're running the show on a Macbook pro, so it's hardware accelerated. I've got the GEM<->PDP conversion figured out in both directions using [pix_2pdp] and [pdp2gem].
Any OSX pitfalls I should be aware of? I don't have access to a macbook pro myself, so I can't test things before I try to get it working on his machine. Should I expect any problems from GEM display or from the [pdp_glx] output?
thanks!
-spencer