From rjn@csn.net Wed Mar 1 12:44:03 EST 1995 Article: 26259 of alt.video.laserdisc Path: casaba.srv.cs.cmu.edu!rochester!udel!news.sprintlink.net!howland.reston.ans.net!vixen.cso.uiuc.edu!sdd.hp.com!hplabs!hplextra!news.dtc.hp.com!col.hp.com!fc.hp.com!rjn From: rjn@fc.hp.com (Bob Niland) Newsgroups: alt.video.laserdisc Subject: Re: Capacity of Next LD format: New Details Date: 1 Mar 1995 15:38:38 GMT Organization: Colorado SuperNet Lines: 71 Message-ID: <3j24du$ih1@tadpole.fc.hp.com> References: <9502261934.AA07480@omnifest.uwm.edu> <3itu49$icl@tadpole.fc.hp.com> <3iu24i$hib@vixen.cso.uiuc.edu> Reply-To: rjn@csn.net NNTP-Posting-Host: hpfcesd.fc.hp.com X-Newsreader: TIN [version 1.2 PL1.4] Ricardo David (davidric@uxa.cso.uiuc.edu) wrote: >> It's about all we have, since no one has come up with a measure for the >> area in which DVD behaves the worst, temporal resolution, i.e. judder, >> nor for transient spatial resolution losses. > Can you explain the above a bit more Bob, what's judder? The movie industry discovered early on that the original 16-18 frames/sec silent camera speeds resulted in "jerky" images. You can easily sense each frame as a still picture. By the time of sound, the rate had increased to 24 (each frame flashed on-screen twice [48 fps] to eliminate flicker, which is a separate problem). Studies show that moving images need to update at between 24 and 60 frames/sec, in order for the average viewer to not consciously experience the program as a sequence of jerky still frames. When digital motion imagery is compressed, a certain amount of redundant and/or predicatable information can be "losslessly" flagged, discarded, and completely reconstructed during decompression. For example, since film is 24 fps, and video 30, 20% of the video content of a film transfer is duplicate fields which can be discarded. Compression also works great on video that is two hours of unchanging still life. However, once the compression algorithms run out of "easy" stuff to discard, particularly during long scenes with high rates of change and motion, the algorithms have to start discarding or simplifying stuff that cannot be completely reconstructed on playback. Although they will attempt to pick material you are less likely to notice, they will often fail, with visible consequences. The results are: - effective frame rate is reduced, causing visible jerky motion (judder) - scene detail is "simplified" resulting in loss of fine (or possibly even gross) detail. - aspects of the coding method become visible, e.g. scene objects from the original image (say, an actor jumping, rain falling, two shots dissolving, or even a simple fade) degenerate into a number of blocks (superpixels) of digital image only vaguely resembling the original. - smooth variations in tone become abrupt "terraced" bands with fewer tonal values. Those unfortunate enough to own CDi/FMV or VideoCD are already very familiar with these problems. Those buying 18-inch sat dishes are beginning to notice. The rest of us see them from time-to-time on crappy standards converions (Olympics from PAL countries) and on the occasional compressed domestic sat feed. (Of course, on MTV, they do this stuff deliberately, because the talentless boobs producing those attention deficit disorder videos think its "artsy".) Learn to identify this garbage. Complain about it, or be prepared for "HDTV" to stand for "Herkyjerky Detail-Trashed Video". > Temporal resolution. I don't think I seen these terms in your > excellent LD FAQs. The LD FAQs don't need to address the issue, because LD has the full temporal resolution of the parent video standard (NTSC or PAL). Some people are sensitive to the mild judder in 3:2 pulldown (see LD#12), but that is a video transfer issue, and not specific to LD. (It's also avoidable, with 2.5 field electronic pull-down.) Regards, 1001-A East Harmony Road Bob Niland Suite 503 Internet: rjn@csn.net Fort Collins Colorado 80525 USA