The 4:2:0 is a standard format, derived from CCIR-601 (1982). To quote Benoit (ch2 Digitization of Video Signals): For applications that are less demanding in terms of resolution, and in view of the bit-rate reduction, a certain number of by-products of the 4:2:2 standard have been defined as follows: The 4:2:0 format This format is obtained from the 4:2:2 format by using the same chroma samples for two successive lines, in order to reduce the amount of memory required in processing circuitry while at the same time giving a vertical resolution of the same order as the horizontal resolution. Luminance and horizontal chrominance resolutions are the same as for the 4:2:2 format [...snip...] In order to avoid the chrominance line flickering observed in SECAM at sharp horizontal transients ... Cb and Cr samples are obtained by interpolating 4:2:2 samples of the two successive lines they will "colourize" at display time. This 4:2:0 format is of special importance as it is the input format used for D2-MAC and MPEG-2 (MP@ML) coding. from Benoit, H. Digital Television: MPEG-1, MPEG-2 and Principles of the DVB System. Arnold 1997 ISBN 0 340 69190 5 I recommend this book.itu.int/itudoc/itu-r/rec/index.html http://www.amazon.co.uk/exec/obidos/ASIN/0340691905/o/qid=992447125/sr=8-4/202-9394619-8581457 Oh, in searching Amazon to get the above reference for you I found that of the two customer reviews the second (5*) is also from a Philips employee. This is clearly the book we crib all our stuff from ;-) I've tried but failed to find the ITU specification itself online (not least because their website seems to hate Mozilla - but that's a different problem). There's a useful summary (with a link to the 4:2:0 variation) at http://www.electric-words.com/dict/9/601ccir.html Chris R. On Tuesday 12 June 2001 23:12, you wrote: > On 11-Jun-01 Gerd Knorr wrote: > >> If V4L/V4L2 is going to include the Phillips packed 4:2:0 format, then > >> it'd probably be best to call it PHILLIPS_420 or something like that, > >> since there are obviously very many possible 4:2:0 packed format > >> component orderings. > > > > Exactly that's why I'd like to know whenever the format used by the > > phillips webcam is a standard format or some dirty hack by phillips > > to make the hardware cheap: 420 instead of 422 to get more frames/sec > > over usb and packed instead of planar because it's easier to handle. > > Well, since I am the author of this Philips driver[1], I can probably shed > some light on this... The format produced _now_ by the driver is this: > > even lines: YYYY UU YYYY UU YYYY ... > odd lines: YYYY VV YYYY VV YYYY ... > > (NB: this was clearly documented in the convertor routines). If this is > standard or not, I don4t know. > > This is what comes out of the camera over the USB bus when in > non-compressed mode. Since there was no definition given of YUV420, I > choose that as the format for this palette. This is also the output of the > decompressor routines (though it can also deliver YUV420P without penalty). > Since there was also a definition of YUV420P, I added that with a simple > convertor routine (it4s not really a convertor, just shuffling around bytes > to the correct place). > > > I don't like the idea of adding non-standard formats to v4l/v4l2 ... > > That4s why I added a bunch of convertors, and this is exactly why I got so > pissed off when Alan ripped out that code. The driver is now appearantly > even less than brain-dead, since not a single tool, apart from my own > camstream, can now use this driver. It delivers a format nobody > understands, and the format it could use isn4t allowed. > > But, if nobody else uses this format, we might as well promote it to > 4YUV4204 or ditch it all together. But in that case I must include code to > go from internal format to YUV420P. >itu.int/itudoc/itu-r/rec/index.html > Oh yes, something else: it seems YUYV == YUV422 as well. > > - Nemosoft > > [1] One 4l4 only! > > --------------------------------------------------------------------------- >-- Try SorceryNet! One of the best IRC-networks around! > irc.sorcery.net:9000 URL: never IRC: nemosoft IscaBBS > (bbs.isca.uiowa.edu): Nemosoft > > >> Never mind the daylight << > > _______________________________________________ > Video4linux-list mailing list > Video4linux-list@xxxxxxxxxx > https://listman.redhat.com/mailman/listinfo/video4linux-list
