Preparing To Start
The first requirement to consider is encoding. We have a combination of MPEG-2 and MPEG-4 encoded video. All video will need to be MPEG-2, so we already know some of the video will have to be re-encoded. Almost all of the MPEG-4 video is encoded at 23.976fps which is the correct way to encode film material for NTSC playback. After encoding to MPEG-2 at this framerate we can apply pulldown flags to duplicate fields. This will change the framerate to NTSC's 29.97fps. All of the MPEG-2 files are encoded at 29.97fps interlaced. They have hard pulldown applied, which effectively means lower quality progressive scan playback and a higher bitrate than should be required for the quality. While this will work for our DVD, most of the video should be re-encoded because the source files use a very high bitrate. We should be able to drastically lower it with minimal effect on quality.
PAL to NTSCOne problem will be the main feature, The 39 Steps. It's encoded at PAL's 25fps, a 4% speed increase over the original film. Since the speed change isn't enough to drastically effect the audio, we're going to leave it at 25fps and use a non-standard pulldown variant to duplicate fields. This will give us correct NTSC playback with no audio change required.
Field DominanceBy creating an index of our MPEG-2 files in DGIndex, we'll be able to access information our software will need for reassemble the original film sources from interlaced video. We also have an MPEG-4 file that's had hard pulldown used to create it, but since it was encoded progressive, there's no longer a field order to help us deconstruct the process. We'll have to use a little trial and error instead.
ResizingAll of the MPEG-2 files have Full D1 NTSC resolution (720x480) already but the MPEG-4 video will need to be resized.This will be handled by AviSynth, and you'll have several different resizers to choose from.
Optional InformationDepending on what MPEG encoder we're going to be using for the next step, encoding video assets, we may also need some colorimetry information. Colorimetry is a set of formulas used to convert RGB color into MPEG's YUV format. Most sources use only a couple of standards, including Rec.709 from the International Telecommunications Union (ITU), which is equivalent SMPTE 240M from the Society of Motion Picture and Television Engineers (SMPTE) and Rec.601 or the equivalent SMPTE 170M .
DVD compliance requires the use of Rec.709, which is what a standard MPEG encoder will use. However, the encoder also needs to know what colorimetry the source was encoded with in order to read the video to be encoded. Most encoders can detect this, but Cinemacraft Encoder Basic, one of the encoders that can be used for the next step, assumes Rec.601 for the input video. This results in colors that are "slightly off" when the input source uses some other colorimetry. DGIndex will be able to tell us each MPEG-2 file's colorimetry. Since all MPEG-4 ASP encoders encode in Rec.601 we'll assume that for the rest of our files.