Question about digital video and analog playback.
 

When playing back a DVD or other digital video file (say with a storage resolution of 720 pixels by 480 pixels) back on a device that outputs a standard analog NTSC-M signal, are we not loosing some of the horizontal resolution?

If my arithmetic is correct, the highest possible horizontal resolution for analog NTSC video would be ~333 lines, and that is for luma. On one of our vintage color sets with a color trap, or a cheap B/W set with 2 or 3 IF stages, it should be ~250 to ~280 lines, given that we're no longer working with the full 4.2 MHz of luma bandwidth. The ~330 line limit should hold true for any set fed via an RF modulator, and also for composite video. A set with component would obviously not be limited in such a way.

Why store the video on a DVD with 720 horizontal pixels then? Why not 640? 560? Cutting the horizontal storage resolution would save, if not tremendously, on storage space. Given the storage limitations of a DVD, and the fact that it came out in the late 1990s, one has to wonder why they specified 720 horizontal pixels in the first place? I don't recall finding a tremendous number of sets from the 1990s with component video inputs.

High end DVD setups in the 1990s did indeed use component video, and it made quite a visible difference in terms of picture quality. I think DVD video was designed not just for home use, but also for higher end setups using video projectors. My own home theatre is from 2000, and it has a Runco projector, which accepts component video. Switching between composite and component video makes a big difference in terms of picture quality. Also, don't forget the S-video connectors which were very very common on 1990s consumer gear. About any TV sold with a DVD player would have accepted S-video if not component, and this was sort of half way there.

DVD was the first consumer video format which had superior image quality to 16mm film, so it was also used with projectors in educational and institutional settings, where the highest possible resolution was very beneficial.

S-Video separates the luma and the chroma, but you're still limited to ~330 lines luma resolution, and the chroma resolution is still constrained with S-Video. Even S-Video doesn't live up to the full potential of the DVD standard.

Perhaps ignorance on the part of consumers played a role? I know my parents had a 40 inch Trinitron with component in, but never the component inputs. My father used the composite video cable that came with the first DVD player he bought around 2001 or 2002.

According to the interweb, you could buy players with component video out as early as 1997, and DVD always supported 16x9 resolution, so I would say it's reasonable to conclude they designed it with the future in mind. VHS had a good 20 year run, I bet they thought DVD would too.

My parents got a Sony 43" rear projection TV and a DVD player in 1999, it was a huge improvement over the 20" trinitron it replaced.. I don't think anyone in our house cared about composite versus component, even through comoosite the equipment was so clearly superior for watching movies to what it replaced.

There was plenty of talk of HD and ED TV going back into the 80's so it is plausible that they didn't want the resolution to look bad in 20 years. However there are two possibilities that seem more likely to me. One is over sampling to avoid pixelation on any output. The other is dvds played via PC even the lower end ones of the period DVD was introduced could do that resolution or better, and PCs with CD ROM helped increase market share of CD before then. I'd imagine that the engineers and product planners behind DVD took that into account when developing it.

I remember back in the days of 486s and Pentium (nothin') processors that Compaq (remember them?) predicted that one day, all video would be digital, and reproduced using computer technology disguised as video gear. I thought they were nuts, based on the computer-based video of the day. Turns out...

Seems DVDs were developed without much focus on then-current display technology.

With baseband (S video) input instead of RF, you are not limited to 330 lines, but to whatever the video output amplifier bandwidth is in the TV, the same as with component video, assuming the luminance signal is input at a point beyond any chroma trap.

As for 720 pixels, it comes from the earliest studio digital video recorder standard, D1, which used a fortuitous common clock frequency for both 525 line/60 Hz video and 625 line/50 Hz video. Since conversions always produce some degradation, it was much better to retain the same pixel numbers for DVDs.

https://en.wikipedia.org/wiki/D-1_(Sony)

Lets be thankful they did make DVD's better than they had to, you can watch one today on a 50 or 60 inch flat screen and still have an acceptable picture, though they can vary wildly in PQ depending on how well they were mastered.

The 720 pixel standard predates the D1 recorders. The relevant standards documents are CCIR601, CCIR656 and SMPTE125M. These standards emerged from a lot of R&D work in several countries, mainly the US and UK but also France, Germany and Japan. The serial digital interface that became universal was subsequently invented by Sony. The original SDI in the standards documents was never used.

The choice of a common 13.5MHz pixel clock for both 525 and 625 was good at the time but has left us with some legacy problems. Non square pixels with different aspect ratios on 625 and 525 are a nuisance for a lot of processing.

All later HD standards have common image formats and square pixels. This gives different pixel clocks for 50Hz and 59.97Hz related standards. Also huge H blanking intervals on some standards such as 720/24p which are a sort of wasted space. Fortuantely they don't have to be transmitted.