Most of my material I record is 1080i ProRes. Deinterlacing is easy with Compressor; the fastest (and best quality) workflow is to stick with ProRes for deinterlacing, retiming and resizing before creating a distribution (h.264) movie.
I am in the process of analyzing different methods of deinterlacing and I must admit that this is really splitting hairs here, but I want to make sure my quality stays as pristine as possible.
I've been able to split and deinterlace fields in compressor resulting in twice the frame rate (59.94 fps) but only by guessing and testing. I can't find any documentation supporting which algorithm Compressor (QuickTime) uses to deinterlace: Blend, Weave, Motion Compensate, or Discard.
My one solid question for this post: When removing a field from an interlaced frame how does the frame get filled in with missing information? It's not line doubled (as far as I can see). Is it interpreted or drawn from the other missing field?
Any links to other posts, information or forums would help me in my endeavor.
Aside from this method and ignoring the drop field methods, progressive frames are completed through either averaging or interpolation. High quality method on deinterlace parts of the picture that have movement (motion adaptive).
Two years from now, spam will be solved. - Bill Gates, World Economic Forum 2004