Last Updated 19 sept 2008.

Errors found by Cosmin Truta


page 179, the text "4:1:1 sampling" should be "4:2:0 sampling". 4:2:2 sampling is 2h1v (which
you correctly mention), 4:1:1 sampling is 4h1v (not 2h2v), and 4:2:0 sampling is 2h2v. Very
confusing, indeed. 

Section 5.9. There are two issues, one about the red, green, and blue proportions as defined
by CCIR 601, and the other is about the suitability of CCIR 601 itself for JPEG. Here is a
short discussion:

CCIR 601 defines the YUV color space, which is a particular kind of YCbCr; specifically, Y is
restricted to the range 16-235, and Cb,Cr are restricted to 16-240. You correctly point this
out. However, JPEG (more specifically, JPEG-JFIF, which is the hugely popular JPEG flavor which
99.9% of the people identify it with the JPEG itself) does NOT use CCIR 601. JPEG-JFIF and
JPEG-2000 only use the R,G,B proportions from CCIR 601, but not the YUV color space. Both of
them use the full-blown YCbCr color space, in which both luminance and chrominance span over
the entire range 0-255.

One of the side effects is that, while JPEG's do not use YUV, but MPEG's (-1, -2, -4...) do,
it's necessary to collapse the 0-255 range linearly to 16-235 when inserting a JPEG frame into
an MPEG movie. It's also necessary to expand the 16-235 range into 0-255 when extracting an
MPEG video frame and converting it to JPEG. 

---------

Errors found by Junaid Qadir

Page 13, line 6: "Our codes are much longer than $k$ bits" should be "Our codes are much longer than $m$ bits"

---------

Unto those Three Things which the Ancients held impossible, there should be added this Fourth,
to find a Book Printed without erratas.

-Alfonso de Cartagena (1385-1456)