Star Wars laserdisc cap at 720 when horiz res is only 425?

[steve996]

Like a lot of guys here, I am trying to figure out the best approach at converting my Star Wars Definitive collection laserdisc set to a XVCD or CVD format, preferably rescaling the horizontal scanlines to create an an anamorphic widescreen disc. I know the rescaled scan lines will be made up, but I think it will look a lot better on my hdtv than non anamorphic with the gaps between the scanlines. I'm hoping my Panasonic RP82 with the Farouja deintelacer will do a good job at providing a decent progressive scan. I've searched this forum and others for the past three weeks without a definitive answer about the best horizontal resolution to capture U.S ntsc laserdiscs at. Laserdiscs have a max horizontal resolution of 425 lines, so would it make any sense, or even be of any benefit to capture at what I read some guys are capturing at 720 or 704, then encoding the mpeg at that resolution? Wouldn't it make more sense to capture at a value equal or slightly greater than 425, and save the bitrate for a higher encoding quality?

As far as the vertical resolution of 480, I intend to crop out the black bars during capture, and during mpeg conversion in TMPGEnc, vertically stretch the movie with ~360 scanlines, and fill the rest with black bars, creating an anamorphic disk. Any input would be appreciated and TIA.

Steve

[borghe]

So far for me, I am capturing at 720x480, cropping out the top and bottom 104 lines (leaving me with 272) resizing that to 360, then adding 60 line black bars at the top and bottom. avisynth script:

LoadPlugin("Decomb.dll") # used to inverse telecine
AVISource("swotep4s1.avi") # first side of disc 1 for testing
Trim(898,50171) # trim the intros and side change
Telecide(guide=1,blend=false,post=false) # deinterlace
Decimate() # inverse 3:2 pulldown
LanczosResize(720,360,0,104,720,272) # crop the black bars then resize to anamorphic
AddBorders(0,60,0,60) # pad black bars to make it 720x480

This gets me decent results. My two things left are to possibly recapture (not entirely happy with the current one) and then find the best method to encode. So far it likes like VBR 4500Kbps avg is my only option...

[solarfox]

Laserdiscs have a max horizontal resolution of 425 lines, so would it make any sense, or even be of any benefit to capture at what I read some guys are capturing at 720 or 704, then encoding the mpeg at that resolution? Wouldn't it make more sense to capture at a value equal or slightly greater than 425, and save the bitrate for a higher encoding quality?

Short answer: No, because "lines of resolution" doesn't mean the same thing as "pixels".

Longer answer: You must remember that Laserdiscs are an analog medium; thus, their resolution is not rated in the same way that a digital system is. 425 lines of resolution doesn't mean that only 425 discrete pixels exist on each scan line; what it means is that a maximum of 425 alternating black-and-white lines can be cleanly displayed, due to frequency-bandwidth limitations, and that anything finer than that will tend to blur.

However -- that only applies to those sharp black-white-black-white alternating lines, because they represent the maximum signal swing from peak to peak and, thus, the maximum frequency response of the system. It is still theoretically possible for finer details to be registered if they do not involve signal swings large enough to run up against the bandwidth limit (such as subtle color shading, for example).

The other reason you wouldn't want to do what you're suggesting is that 425 (or "slightly greater than") isn't a valid DVD resolution, so when you encode to DVD-compliant MPEG your software is going to interpolate it out to 720 anyway and you won't gain any significant savings in bit rate.

[steve996]

Thanks guys. Borghe, your script seems to make sense, I'll take a closer look at it and perhaps try it out also. Solarfox, your explanation was very concise and now I full understand. As you said, I was unsure as whether to treat the horizontal resolution as a one would a digital pixel, or as an analog spectrum as you describe. Now I just have to figure out how to create an acceptable anamorphic conversion.