Where do you get the idea that the VX2000 has "530 vertical lines" of resolution?

It may have "530 TV lines" of resolution. TV lines are not physical lines, but are only calculated based on a formula.

A typical HDV camera may indeed have something like 720 TV lines of resolution, this is regardless of if it is shooting 720P or 1080i or something else.

The more expensive HD cameras have a much higher resolution thanks to much more expensive circuitry, and in practical use also benefit from lenses that cost from $20K to $200K each (the latter for a sports zoom that can follow a golf ball from tee to hole-in-one).

Actually, my VX2000 manual says it's CCDs have "500 lines of horizontal resolution". That's 500 vertical lines isn't it?

For the VX2000 I found "530 lines of horizontal resolution" which means "TV lines". Nothing to do with CCD pixels, other than that those provide an upper theoretical limit.

The original idea was that "530 lines of horizontal resolution" meant that you could just barely discern 265 black vertical lines if the space between them was white.

If you are not shooting just clean black and clean white, it becomes a lot more important to look at the MTF (Modulation Transfer Function). This function determines, popularly described, how much mushier your video will look than what is in front of the lens. It specifies what will happen to not only black lines on a white background, but how well say a dark gray feature will stand out against a medium gray background.

A lens or a video system can have an incredibly high black-on-white resolution, but still look terrible in real life. For example, Nikon still lenses used to be very sharp but had poor MTF in the 60s and 70s, while Canon focused on the MTF early and as a result made lower contrast motifs look much better.

Today all lenses are designed with good MTF in mind, and video is aspiring (and in the high level formats it is successful with it).

See Adam Wilt and the following from him:

"...the common rating [is] 80 TVL/ph (TV lines per picture height) per 1 MHz of bandwidth. Broadcast resolution is normally said to be 336 TVL/ph: 4.2 MHz luma bandwidth."

Someday we may have truly digital systems where a pixel is a pixel and there are no other limiting factors...

OK, now that approaches making sense (to my limited intellect anyway)! So why then does SD rendered from HD look so much better than SD shot that way to begin with?

One reason is the HD color space of 709 vs. 601, it looks more "colorful."

Another reason is that the limited color sampling at least matches between HDV and MPEG-2 DVD, they are both 4:2:0.

When you shoot NTSC DV, you're shooting with 4:1:1 color sampling which throws away 75% of the CCD color information (to save bandwidth/recording time).

4:2:0 also throws away 75% of the color information, but when you translate from 4:1:1 to 4:2:0 you lose half of the color samples, leaving you with 12.5%.

In practice, it's not quite as bad as that. Many MPEG-2 encoders do a bit of chroma smoothing, ie. they try to synthesize some of the missing color information, which helps to a point.

There has also been some technical progress in DSPs which have benefitted the picture on HDV cameras.

That would explain why rendering straight to mpeg2 for DVD authoring looks so much better than rendering to a DV avi first. Yeah, I can see the improvement in the color. I also see the same color improvement if I capture to Cineform directly using DirectHD rather than converting m2t files with Gearshift, though it's a lot more subtle in that case.

The VX2100 is far from being a top shelf DV camera. The difference between a 2/3" CCD DV camera with good glass and the 1/3" CCD cameras with fixed lenses is quite significant. Not only that but I do a lot of work with DigiBetacam but capture that as DV and that looks even better still. In fact for going out to DVD I'm hard pressed to see any advantage in capturing it as 4.2.2 compared to 4.2.0. I guess working in PAL has some bearing on this as well.
Bob.