[mdryja]

After all this discussion regarding 1080i-->1080p on this board, I have a couple of questions:


1) Why did the Powers That Be even *decide* to do 1080i as one of the two high-def resolutions, and why aren't more broadcasters (i.e., any except ABC) choosing 720p instead? It seems that with all the problems with 480i-->480p conversion, that a "p" resolution would have been selected for HD (or only progressive resolutions, and no i resolutions). I understand that CRT technology is still somewhat prevelant (although how much for HD sets is debateable), but inasmuch as a lot of these CRT's can do 480p, surely they could also just easily as do 1080p?


2) With respect to 1080i->1080p, it seems to me that all the algorithms that have been "tried and true" with 480i->480p (Faroudja's DCDi, Silicon Image's, etc.), it wouldn't be very difficult to implement the same algs with 1080i->1080p, and just boost the horsepower to compensate for the increased resolution? Is this the case? If not, why not? I would presume that with the advent of faster processing capabilities, it wouldn't be hard to do 1080i->1080p these days. But, it seems that only the ancient F5000 and the new KD HD Leeza do the trick . . . . .

 

[oferlaor]

effective bandwidth of 1080i is lower because it sends 540 lines per sync signal, whereas 720p has to send 720 lines.


At first this took me by surprise as well, but we should really thank our lucky stars that this is what many broadcasters picked because it leaves the door wide open for 1080p.


As this standard is obviously going to be with us for a long long time, it's best that it contains as much possible data to accomodate the next decade (at the very least). 1080i does just that by allowing us to use just 540p now, and "upgrading" (via processing) to 1080p relatively effortlessly.


It is just a matter of time before more scaling engines are created to allow for 1080i deinterlacing. The technology is well known, it's just a matter of market availability and processing power. Both DCDi and DVDO's algorithm (Pure Progressive) can be adapted to 1080i, and I believe they probably will at one point or another.

 

[mdryja]

Ofer, I was hoping you'd respond -- thanks for the info! Insofar as there are now two 1920x1080p fixed-pixel displays on the horizon (a Sharp LCD and a Samsung or LG plasma), I'm looking forward to 1080i-->1080p becoming more common, insofar as it can only help 1080i-->1366 x 768p (assuming that what would be done would be 1080i-->1080p-->1366 x 768p, instead of the mroe common 1080i-->540p-->1366 x 768p).

 

[Dale Adams]

Quote:

Originally posted by mdryja 2) With respect to 1080i->1080p, it seems to me that all the algorithms that have been "tried and true" with 480i->480p (Faroudja's DCDi, Silicon Image's, etc.), it wouldn't be very difficult to implement the same algs with 1080i->1080p, and just boost the horsepower to compensate for the increased resolution? Is this the case?

That's certainly the case with the Silicon Image deinterlacer. There's no reason why exactly the same algorithms and processing stages couldn't be applied to a 1080i signal as a 480i or 576i signal. It's largely a matter of increasing field/frame buffer storage and bandwidth into, inside of, and out of the deinterlacer.


- Dale Adams

 

[oferlaor]

Dale,


Any ideas if SI are going to push for something like that?


From what I'm seeing on the forum (note that this might not apply to the real world, but I suspect it would) there's a large market waiting for such a chip to come out with full 3:2 cadence detection (and actually 2:2 pulldown as well because [email protected] is starting to pick up steam in Europe and in Australia).

 

[Dale Adams]

Ofer,


I don't really know what Silicon Image has in the planning or development stages. Even if I did, I would probably be under NDA and unable to tell anyone about it. It does seem like a logical thing to do, however.


- Dale Adams

 

[Allan Jayne]

After figuring in the horizontal retrace intervals and scan lines used for the vertical retrace intervals, 720p and 1080i have the same bitrate after decoding if you equate a pixel to a bit, and therefore both have the same theoretical baseband video bandwidth requirement of about 37 MHz.


With 1080i manufacturers can skimp more on bandwidth before the viewer notices it, since the design horizontal pixel count is so much larger. If the system bandwidth (for a mostly analog video path such as inside most HDTV sets) was halved, there would still be 960 max pixels across for 1080i horizontal resolution but just 640 across for 720p.


With 1080i the picture tube scan rate is quite similar to that for 480p, for each 1/30'th second 1125 total scan lines versus twice 525 or 1050 for 480p.


Fixed pixel displays such as LCD can do a de-facto weave de-interlacing right in the display element.


With an external de-interlacer, 1080p needs a 74 MHz path to the TV.


Video hints:
http://members.aol.com/ajaynejr/video.htm