Xref: utzoo comp.graphics:6823 rec.video:7494 sci.electronics:7388 Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!brutus.cs.uiuc.edu!apple!sun-barr!newstop!sun!vector!poynton From: poynton@vector.Sun.COM (Charles A. Poynton) Newsgroups: comp.graphics,rec.video,sci.electronics Subject: Re: HDTV and ATV Glossary (TN32) Summary: A raster is fully specified by five numbers, the rest are derived. I prefer these five, illustrated by the trademark Poynton raster diagram. Keywords: 525/59.94, 625/50, NTSC, PAL, SECAM, Component, Composite, Message-ID: <121089@sun.Eng.Sun.COM> Date: 11 Aug 89 23:24:39 GMT References: <120919@sun.Eng.Sun.COM> Sender: news@sun.Eng.Sun.COM Lines: 83 A correspondent writes: >> 1250/24, 2048-by-1152, 74.25 MHz. Dare to be square! > I don't understand how 1250 becomes 2048-by-1152, and what the active > elements stuff is about and how 1250/24 becomes 74.25 MHz. Well, briefly, you get to pick five numbers and the rest fall out of the wash. The best numbers to work with are sample rate, total and picture samples per line, and total and picture lines per frame. The best numbers to publish comprise that set, except publish frame rate instead of total samples per line. Total samples per line needs to be roughly 15% more than picture samples to accommodate horizontal scanning, Total lines per frame needs to be roughly 4% more than picture lines per frame to accommodate vertical scanning. Frame rate just falls out as sample rate, divided by total samples per line, divided by total lines per frame. So in my proposal, total samples per line would be 2475, about 20% greater than 2048, but you can derive the 2475 from the five numbers I gave. You really need to see the trademark Poynton raster diagram to make this all clear. How much can I do with ASCII text ... 74.25 MHz 2475 +---------------------------------------+ | V blank | | +-------------------------------+ | | 2048 | | | | | | | 1250 | H | 1152 picture | | blank | | | | (16:9 aspect ratio) | | | | | | | +-------+-------------------------------+ Video monitors are usually specified by horizontal line rate (easily derived as sampling frequency divided by total samples per line), and frame rate. Unfortunately many frame buffers (whoops, graphics cards) specify some rounded versions of H and V rates, and it's a real nuisance to work backwards to try to figure what the real parameters are. My 1250/24, 2048-by-1152, 74.25 MHz HDTV proposal is, briefly, - use 1250 total lines and 1152 picture lines, as in current Eu95 proposals, to appeal to the Europeans; - use the proposed Japanese [and semi-recommended SMPTE 240M] sampling frequency of 74.25 MHz; - use 2048 samples per picture width and square pixels to appeal to the computer industry; - use 24 Hz to be fully conformable to film, for the Hollywood production community and to utilize current film libraries which will be the source of 90% of the initial ATV programming, with ABSOLUTELY NO temporal artifacts; - accomplish down-conversion to current broadcast standards in exactly the way it is done today from film, that is, 3-2 pulldown 0.1% slow for 525/59.94 and 2-2 pulldown 4% fast for 625/50, with artifacts identical to those seen in today's film transfers; and - display at 72 Hz to satisfy even the most flicker-sensitive, Euro-gonomic, high-ambient viewers, including computer workstation users. Of course a 2048-by-1152 monitor at 72 Hz is quite a way off, so this should mollify the U.S. broadcasters, who are concerned that an early entry into ATV could cost them a lot of money. Don't laugh about this being a political compromise. This is serious business! C. ----- Charles A. Poynton Sun Microsystems Inc. 2550 Garcia Avenue, MS 8-04 415-336-7846 Mountain View, CA 94043 "As at the ski resorts where girls go looking for husbands, and husbands go looking for girls, the situation is not as symmetrical as it might seem at first." -- attributed to Alan Kay -----