I've always been a little puzzled by how analog horizontal resolution worked, and reading the wikipedia entry on the resolution of different things, it appears that old 480i SD broadcast television and Beta SP had 330 "lines" of horizontal resolution, which on the page is estimated to be the equivalent to around 440 pixels. I guess since it's analog, it doesn't mathematically need to be a 4:3 ratio in order to work on a 4:3 display?
So I'm wondering - since 4:3 DVDs are mathematically 4:3, does this mean the video has to be scaled/interpolated horizontally when it's transferred from videotape, thereby softening the image? I have noticed that when I play tape-based DVDs on my imac, they actually look sharper when I downscale the player than they do at the default size, although at the expense of a little bit of fine detail.
+ Reply to Thread
Results 1 to 8 of 8
Tape is analog so it doesn't come in discrete units along the horizontal axis, it's a continuous waveform. When digitizing it you can use as few or as many samples as you want. The bandwidth of that analog waveform on VHS tape is such that it can only resolve about 350 alternating black and white vertical lines (color resolution is far less, about 40 lines). It's customary to sample standard definition analog video with ~704 samples, usually with 8 pixels of padding at the left and right edges -- 720 pixels total. That's enough to capture the resolution of studio video tape.
NTSC DVDs are almost always 720x480. That's a 3:2 aspect ratio. But the encoded picture is either 4:3 or 16:9 when displayed. The player or TV adjusts for the display aspect ratio. Any time you enlarge a small image you get a fuzzy big image. Any time you reduce a large image to a small image you get an apparent sharpening.
Last edited by jagabo; 19th Jun 2014 at 10:25.
A film sourced standard definition DVD would be considered a higher resolution than a standard definition broadcast tape, though, correct? I always figured that to mean 720p is a higher horizontal resolution than whatever broadcast tapes typically were, regardless of how that would be measured, since they're both 480 vertically.
Continuing from what jagabo said,
Transfers to digital from film sources are either digitized directly at a higher rez (HD/2k, UHD/4k) and then downsampled to SD for DVD, or they are digitized directly at SD spec (not common anymore), or they are converted to analog video (usually SD, but possibly HD) and then digitized. #2 is clearly using 720x480/576 or 704x480/576.
#3 works just the same way it would if you were transferring from any other studio SD analog tape. Analog has a continuous horizontal signal that is bandwidth-constrained, which sampling at 704/720 pixels per line should fully cover. The vertical is 525/625 lines, but of course that includes vertical blanking. Vertical blanking isn't necessary in a stored digital file, because the file format & metadata make intrinsic assumptions about the resolution and so the VB can be, and is, automatically re-created on playback out to digital or analog transmission paths (HDMI, DVI, Composite, etc). Thus, out of 525 discreet rows, only 480 (or 486 for studio formats) include active picture.
As far as 4:3, 16:9, etc., the 704/720 sampling of the continuous analog line works out to (when compared with the 480 line active picture height) a non-square Pixel/Sample Aspect Ratio. All devices that work with SD digital video SHOULD take this non-square feature into account.
4:3/16:9 DVDs are mathematically 720/704 in one dimension and 480 in another dimension. IMO, the conflation of these 2 dimensions into a Storage Aspect Ratio (of ~3:2 in this case) is a faulty assumption, because the ratio, per se, is never worked on independently, just the 2 raw dimensions individually. So a 720/704x480 DVD can be 4:3 (due to non-square PAR) or it can be 16:9 (due to a differently sampled non-square PAR). It is not scaled during sampling, usually only when necessary (such as during full-screen playback to Square-pixel displays). Then, it could be either scaled up/down horizontally-only, or vertically-only (though horizontal-only scaling is preferred for analog-originated & interlaced material).
Film resolution is also "analog", and the quality depends greatly on the film stock used (grain size), age of the film, the optical/chemical processing done, which pos/neg generation is being used. For standard 35mm, this can be anywhere from ~ED+ quality (~1000x540) to ~6k/UHD quality. This also varies depending on which criteria is being used to evaluate it. For the most part 2k or 4k are commonly considered to be equivalent digital conversions. There are other factors that one can compare with Film vs. Video, including dynamic range/latitude, colorspace/gamut, etc., so there is almost always a drop going from analog film to SD or HD analog or digital video. How much of the detail is retained in the A->D sampling and scaling conversions depends on the method (#1, #2, or #3), but it seems clear to most that a pristine, original Film Negative laser-scan sampled to 4k and then downsampled to SD for use on DVD retains much more detail "quality" than those other methods, which is why preservationist companies such as Criterion attempt to do much of their transfers that way (and also advertise as such).
Last edited by newpball; 8th Jul 2015 at 15:40.