home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.compression:2816 rec.video:9597 rec.video.satellite:3781 rec.video.production:1140
- Newsgroups: comp.compression,rec.video,rec.video.satellite,rec.video.production
- Path: sparky!uunet!munnari.oz.au!labtam!graeme
- From: graeme@labtam.labtam.oz.au (Graeme Gill)
- Subject: Re: Compression artifacts in NBC Olympic coverage
- Organization: Labtam Australia Pty. Ltd., Melbourne, Australia
- Date: Mon, 27 Jul 1992 04:21:01 GMT
- Message-ID: <1992Jul27.042101.9732@labtam.labtam.oz.au>
- Followup-To: rec.video
- Summary: Probably PAL -> NTSC
- References: <MONTA.92Jul26181344@image.mit.edu>
- Lines: 22
-
- In article <MONTA.92Jul26181344@image.mit.edu>, monta@image.mit.edu (Peter Monta) writes:
- > While I was watching the US-Angola basketball game and the coverage
- > of the early gymnastics compulsories, I thought I saw several image
- > artifacts. Finally I taped a segment of the gymnastics and waited for
- > a particularly bad example, then went through it frame by frame.
- >
- > There's no question about it: there are block artifacts in the
- > regions where a fast-moving object occludes or reveals the background.
- > For example, NBC showed the compulsory floor exercise of Maria
- > Neculita (Romania) at around 5:30pm EDT. Look at the first tumbling
- > run, in which Neculita's legs flash past the audience.
-
- Being from Europe the TV signal probably originated as a PAL signal
- and has to be converted to NTSC, wich means both the colour encoding,
- frame rate and number of lines has to be converted to the US standard.
- Simple interpolation and smoothing of the frame rate (50Hz -> 60Hz) usually
- leads to smearing of fast motion, so the "better" standards converters have
- various "smarts" to recognize motion and transcode it specially. I guess the
- motion estimation circuits work on a block basis, and you may be seeing these
- block based motion artifacts.
-
- Graeme Gill
-