|DV Pix - Sampling Methods||
||4:2:2 sampling is used in ITU-R BT601, D-1, D-5, Ampex DCT, Digital Betacam, Digital-S, and DVCPRO50 formats.|
||4:1:1 sampling is used in 525/59.94 ("NTSC") DV and DVCAM, and in both 525/59.94 and 625/50 ("PAL") DVCPRO.|
||Co-sited 4:2:0 sampling is used in 625/50 ("PAL") DV and DVCAM formats. Cr and Cb samples are on alternate lines in each field (a frame view would show two lines of YCr,Y pairs, then two lines of YCb,Y pairs). Note that this is different from the 4:2:0 chroma sample positioning in JPEG, MPEG, and H.261 formats!|
|schematics courtesy of Panasonic Broadcast and Digital Systems.|
4:1:1 Sampling: detail
4:2:0 Sampling detail
|This image, shot with a 525/59.94 DV format VX1000, is the best demo
I have showing the nature of 4:1:1 sampling. The 72-pixel by 48-pixel detail
clearly shows the coarser horizontal sampling of the color: once every
four luminance samples. The color sample is co-sited (located at the same
place) as the first brightness sample in each 4-pixel group, and the color
at that location is simply replicated over the following three pixels (thus
there is an average 2-pixel "lag" or Y/C delay of the color data compared
to the brightness data).
The yellow and green LEDs in the image show the same artifacts; it's not specifically a "red smear" problem. However, the red LEDs show the most color contrast with the surrounding background, so I show red ones here for clarity.
While this looks pretty horrible, remember that the color "sharpness" of DV is comparable to that of BetacamSP. It's quite sufficiently sharp and detailed for many purposes (including much chroma-keying, HDTV upconversion, and film transfer work) although for extremely critical work a 4:2:2 format such as Digital-S or DVCPRO50 will give noticeably superior results in a side-by-side comparison.
The 45-pixel by 30-pixel 4:2:0 detail, kindly supplied by Karel Holubièka in Prague, shows the more evenly-balanced subsampling of the red LEDs on his audio mixer panel as seen by his 625/50 DV DCR-VX1000.
Also see Pix: Codecs for more 4:1:1 sampling details
Technical Details: Images were shot on Sony DCR-VX1000
3-chip DV camcorders. The 525/59.94 tape was played in a DHR-1000 DV VTR
and captured digitally using the DPS Spark (Adaptec AHA-8940) IEEE-1394
I/O card on a P133 Wintel machine. The resulting AVI video clip was opened
in Adobe Premiere 4.2 and the desired frame was exported as 24-bit Windows
BMP files using the Adaptec DVSoft codec; the BMPs was then read into Adobe
Photoshop LE. The full-screen image was downsized to 320x240 using bicubic
interpolation and saved as low-quality JPEG for efficient transmission;
full screen image is shown as reference only, not as a sample of the quality
inherent in the original BMP. The detail image was upsampled from the original
BMP using "nearest neighbor" interpolation and saved as high-quality JPEG
to preserve visual quality; this JPEG is very close in appearance to the
original image. The 625/50 image was similarly captured using a DPS Spark
in Prague; the AVI was emailed to me and I extracted a frame from it for
processing. All images are shown with square pixels, although 525 pixels
are 12.5% narrower than square, while 625 pixels are 6.25% fatter than
square: 525 pixels should be slightly taller than wide, while 625 pixels
are very slightly wider than tall.
Copyright (c) 1998 by Adam J. Wilt.
You are granted a nonexclusive right to duplicate, print, link to, frame, or otherwise repurpose this material,
as long as all authorship, ownership and copyright information is preserved, and a link to this site is displayed.
For use of the 4:2:2, 4:1:1, and 4:2:0 schematics, contact Panasonic Broadcast & Digital Systems.
Last updated 16 August 1998.