Posted by Jukka Aho on 12/06/06 13:45

Lincoln Spector wrote:

> Part of the DVD signal identifies the program as 4x3 or 16x9. Ever
> watch DVDs on a computer in a window (as opposed to full-screen). The
> window changes shape when it goes to 16x9.

Of course, but that's beside the point. The aspect ratio switching works
on a computer screen because the (software-based implementation of a)
DVD player on a computer can directly read the aspect ratio flags off
the original MPEG-2 stream. The situation is different when you're
watching DVDs using a set-top DVD player and a regular tv set. The tv
set does not have any direct access to the original MPEG-2 stream, much
less to the aspect ratio flags within that stream. Only the DVD player
has access to that information.

It is possible to relay the aspect information from a set-top DVD player
to the tv set - and that is indeed what is being done in 625/50
countries. For that to work, however, you need an aspect ratio signaling
specification: some sort of standard that is universally agreed upon
among DVD player manufacturers and tv manufacturers.

Two such widescreen signaling standards have been devised for the 625/50
system. Both are widely used in the PAL countries. But neither of these
methods is directly applicable to the 525/60 ("NTSC") signals.

What I'm interested in, and what my original question was about, is
whether there are similar (widely-used and widely-supported) widescreen
signaling standards for the 525/60 world - and if so, how do they work,
and where are they specified.

> Whether that information gets to the TV, I'd imagine, depends on what
> type of connection you're using. I'm using component video, and it
> works fine. I suspect it also works with HDMI and doesn't work with
> composite video. I don't know about S-Video.

Not necessarily. For example, the line 23 ("PALplus") widescreen
signaling - which is one of the methods used in 625/50 countries - is
designed in such way that it survives all commonly-used analog signal
types - CVBS, Y/C, YPbPr, RGB. It even survives recording onto a VHS
tape. (The VHS VCR does not need to support this in any special way.)

Now that I scoured the web, it seems that there actually _is_ a
specifically tailored version of the PALplus line 23 WSS signaling for
the 525/60 systems. It carries the same information, albeit on different
lines (20 and 283.) See this Intersil application note (page 3):

<http://www.intersil.com/data/an/an9716.pdf>

Without further investigation, it is hard to say whether this is the
exact WSS method that is being employed when you're watching DVDs using
your set-top player, but it would seem plausible. If it is, I would
expect the aspect ratio signaling to work on the same equipment even
over composite video (yellow RCA) connections.

> You also have to tell your DVD player that you have a widescreen TV
> (there's a setting in there somewhere). Without telling it that, the
> player will convert all 16x9 signals to 4x3, adding letterboxing and
> reducing the vertical resolution by 25%.

It's the same over here in 625/50 land, but that "16:9" setting is not
strictly for 16:9 sets only. It's more like a "is your tv set capable of
handling full-frame 16:9 signals all by itself?" setting.

For example, I have a 4:3 SDTV set (a direct-view CRT) that can handle
full-frame ("anamorphic") 16:9 signals just fine and display them in
their correct aspect ratio - letterboxed within the 4:3 screen. Hence,
I'm using a "16:9" setting on my DVD player, even though it is not
connected to a 16:9 set. (When that TV set encounters video that is
signaled as being "16:9", it changes its vertical deflection
sufficiently so that you'll get a letterbox effect. It still draws all
the lines on the screen - just constrained in a tighter space. This way,
you'll get at least marginally better resolution than when you let the
DVD player do the letterboxing up front. The actual increase in
resolution depends on a number of different factors, of course - such as
the resolution of the shadow mask and the focusing of the electron
beam.)

--
znark

[Back to original message]