If you send 50i (or 60i) footage to a flatscreen it has to do in-frame interpolation (between the 2 fields) but also in-time interpolation (it creates a new frame in between two normal ones). It works this way. Good (all new) LCD TV sets use a special algorithm to view in progressive (they are all in progressive.. there is no interlace on flat screens). Let's say your footage is made by interlaced frames in this way: 1U1L, 2U2L, 3U3L, etc... What they do they transform it to i(1U1L), i(1L(mU)2U(mD)), i(2U2L), i(2L(mU)3U(mD)), etc... To explain better.. They create 50 frames out of 25 (or 60 out of 30) by creating in-between frames using the lower field of the first frame but mooving it up and the upper field from the next frame, but moving it down.. and then they interpolate the two fields. This means that a 50i DVD stream on an LCD has single frames that have a quality of a normal interpolation from 50i to 25p but they are 50 and not 25.
If your DVD is progressive (25p or 30p) the flatscreen just shows what you are sending.. with no in frame interpolation and no in-time interpolation.
So... if you have interlaced footage to start with... whatever you do... the final quality of the single frame (ie in stop motion) comes from the interpolation of the two fields.
If you do the interpolation before (when you export) you will find yourself with a 25p (or 30p) stream. This will give a "film look" (thus less frames) but will have the advantage of controlling this interpolation. You might find that some software is better than others and, being something YOU control, you can take all the time you need to do your tests, etc...
If you output interlaced (thus not interpolating) you'll leave this job to the flatscreen. The advantage is that you'll have more frames at the end (double) and so you'll have a "TV look" but you'll not be able to control the quality of the interpolation because it will be left to the flatscreen to do. On some older models you might even risk a double frame creation by field duplicating and not by the process I described above.
Obviously the best thing for progressive output is to shoot in progressive. In this way the frame has a real full resolution and not an interpolated one. But I guess this is not the case.
siRoma di Marcello Mazzilli
Corporate video productions in Italy
http://www.siroma.com
