Hmm. I'm definitely not
in the DVDs-match-HDTV resolution camp, so I'll elaborate a bit on my comment above about very poor HDTV approximating very good DVD picture quality.
Lots of threads in the programming forum discuss poor 1080i quality. Don't think it's unreasonable to wind up with a full-screen 16:9 horizontal resolution of only 800-1000 displayed pixels. That's factoring in potential filtering by the director (film or tape), perhaps limitations in the HDTV tape equipment (<1440-pixel HDCAMs in the link), and ~20% filtering at the receiver.
As for vertical resolution, depending on the motion in the images, it might be 800 lines (static) or only 400 lines (dynamic). This Tuesday
(my 11/6 post) is the last time I dragged out these measurements based on data from the FCC's experts. That thread cites (as sublinks) other resolution-limiting factors--plus, absolutely free, some resolution-related Anna Kournikova links. :)
Anyway, if you take a worse case 1080i dynamic image, say 400 X 800, IMO there may be times when it compares to a DVD with a potential horizontal resolution of 700+ (with minimum prefiltering). Also, with a HTPC or scaler, some boost the apparent resolution of DVDs by replicating or artificially generating pixels (not original samples). That may be where some mistakenly think they're seeing HDTV-equivalent resolution. -- John