You'd have to lookup the specific use cases (as in, what format/service, what software, and what hardware as some might be able to use the video processing blocks). I'm not sure if the platforms themselves might offer more info or not. For instance there's a difference between just having the video mapped to the full display, versus how you actually watch Netflix through the Netflix VR app (which you're apparently in like a CGI cabin or something watching a CGI TV), and I think PSVR has a theater mode where it puts you in a CGI theater type setting for watching video (not sure if you have to do that mode, I think there is a generic "fill the screen" fall back, where it doesn't use any head tracking and just processes the image)
A lot of the 3rd party stuff is likely using the GPUs to interpolate images or doing the 3D stitching, and that will take GPU (not sure how much; I mean, you can peg high end GPUs with certain video processing so it really is dependent on the specifics of what the program is doing).