Harsh, "aliased" sound with digital TV converter box.

G-squared wrote:
On Mar 15, 7:26 am, jwvm wrote:
snip
You are still wrong here. Color difference signals do carry
luminance
information. Color differences for a given feature will be larger
in
bright regions of an image compared to dimmer regions.

I'm done, and I'm outta here.

All right. Let this thread rest in peace!!

There is no luminance in the color _difference_ signals. You
apparently have never seen a vectorscope either analog or digital.

From the looks of this (lengthy) thread, I suspect there are a few who
might find this interesting. It's the beta version of a "scope" pair
that works on still images.

I actually do compute color difference signals (UV) to drive the
vectorscope, but I don't make them available at the line scope. Still,
the results can be instructive. A nice SMPTE color bar BMP is
included.

Take a look:

http://www.microdexterity.com/demos/PicScope.zip

--
Jim

I said I wouldn't post again (and don't need to, because other people are
giving good info about color-difference signals -- and I've bookmarked their
references for further study), but I want to repeat my request for a copy of
the Sams book.

I'm pretty certain it's Sams. It dates from the mid '60s, as far as I know.
It is anxcruciatingly complete analysis of how color TV works. I don't know
the title, and my attempts to find it have so far failed. If anyone can
help, it would be most appreciated.

Thanks.

On Mar 15, 10:24*am, "William Sommerwerck"
wrote:

Now let's return to my original claim that the creation of color-difference
signals removed all the high-frequency detail from them.

I was wrong.

That's enough for me.

(At the moment I write this,
the female personality on CNN is wearing a vivid orange-red dress. She's
next to a black background, yet the edge of her dress doesn't appear blurred
or smeared.) The eye, apparently is not disturbed by this, because it "adds
in" the sharp luminance transition for an overall impression of sharpness.

This makes it possible to reduce the bandwidth of the color-difference
signals without losing (much) information that the eye cannot "regenerate"..

Yup. That's why color difference signals are also used in DTV.

I should also point out that, as the color-difference signals do not carry
luminance information, they convey less information than a color primary
would, and should require less storage space.

Since the color difference signals are defined as R-Y and B-Y, I
suppose that "by definition" these color difference signals remove Y
from the color primary. However, since that Y information is carried
in the Y component, it is not bandwidth that is saved. It is merely
bandwidth that has been reassigned elsewhere in the signal. The
contribution to Y from each of the color primaries (R, G, and B) gets
collected in the Y component, that's all.

All's well that ends well.

Bert

"Albert Manfredi" wrote in message
...
On Mar 15, 10:24 am, "William Sommerwerck"
wrote:

I should also point out that, as the color-difference signals do not
carry
luminance information, they convey less information than a color primary
would, and should require less storage space.

Since the color difference signals are defined as R-Y and B-Y, I
suppose that "by definition" these color difference signals remove Y
from the color primary. However, since that Y information is carried
in the Y component, it is not bandwidth that is saved. It is merely
bandwidth that has been reassigned elsewhere in the signal. The
contribution to Y from each of the color primaries (R, G, and B) gets
collected in the Y component, that's all.

All's well that ends well.

It doesn't.

I'm not going to respond to these final remarks, except to say that, just
because I'm not making a counter-argument -- which I'm not -- doesn't mean I
agree with them. Which I don't.

"William Sommerwerck" wrote in message
. ..

The best book on color television -- ever -- was published by Sams, ca.
1960. It was beyond belief, going into the details in excruciating depth.
One unique feature was page of two columns of colored rectangles -- blue
and
green -- that were large at the top of the page and got progressively
smaller. At the bottom of the page the colors looked a lot more "alike"
than
they had at the top. The writer did not expect the reader to blindly
believe
what he said. (If anyone knows the title of this book, please let me know.
I've contacted Biblio and Sams for more information, but they had none. I
called a store in Canada, but the book they had was actually about selling
color TVs.)

The full, correct title is (was) Color TV Training Manual. You may Google
for
"color TV Training Manual" sams for 16 links. The original was 1956. I
had
the 2nd Edition, 1965. Some copies may be available. It was, indeed, a
very
fine book. I'm sorry I lost mine.

I knew about the two columns of color squares. I thought specifically about
them when I wrote about the and cones -- relevantly, by the way, despite
your demurrer -- and I should have included a mention.

"Sal M. Onella"
wrote in message ...

"William Sommerwerck" wrote in message
. ..

The best book on color television -- ever -- was published by Sams, ca.
1960. It was beyond belief, going into the details in excruciating depth.
One unique feature was page of two columns of colored rectangles -- blue
and green -- that were large at the top of the page and got progressively
smaller. At the bottom of the page the colors looked a lot more "alike"
than they had at the top. The writer did not expect the reader to blindly
believe what he said. (If anyone knows the title of this book, please let
me know.

The full, correct title is (was) Color TV Training Manual. You may Google
for "color TV Training Manual" sams for 16 links. The original was 1956.
I had the 2nd Edition, 1965. Some copies may be available. It was, indeed,
a a very fine book. I'm sorry I lost mine.

Others think so, apparently. I'm surprised at the prices it's going for.
There's also a third, 1973 edition.

Thank you for helping out. I've started collecting classic works of
technical literature, the first being the Philbrick manual on op amps, an
amazing work. This will be the second.


I knew about the two columns of color squares. I thought specifically
about
them when I wrote about rods and cones -- relevantly, by the way, despite
your demurrer -- and I should have included a mention.

I sincerely apologize for my demurral.

What's interesting about the color squares is that even the smallest ones
are huge compared to the cones' foveal density. Yet, these green and blue
squares look more "alike" than the big ones at the top of the page. Clearly
(???), "something else" is going on in the eye.

By the way, if you've never seen a Land-retinex demo, be prepared to be
startled. Two photos are taken on B&W film, one through a red filter, the
other through green. The images are then projected, the red image through a
red filter, the green unfiltered. The combined image shows a fair
approximation of the original colors, including reds, greens, blues, and
yellows! In "Insisting on the Impossible", the author documents the grief
Dr. Land went through trying to get respect for his research from the
scientific establishment.

The Smithsonian used to have a retinex demo. I don't know if it's still on
display.

PS: Wasn't "Salmon Ella" a relative of Typhoid Mary?

On Mar 15, 8:45*pm, "William Sommerwerck"
wrote:

"Albert Manfredi" wrote in message

Since the color difference signals are defined as R-Y and B-Y, I
suppose that "by definition" these color difference signals remove Y
from the color primary. However, since that Y information is carried
in the Y component, it is not bandwidth that is saved. It is merely
bandwidth that has been reassigned elsewhere in the signal. The
contribution to Y from each of the color primaries (R, G, and B) gets
collected in the Y component, that's all.
All's well that ends well.

It doesn't.

"It doesn't" what?

Are you saying that it doesn't end well, or are you saying that Y is
not composed of the weighted sum of the luminance carried by the three
primary colors?

If the former, it's up to you. If the latter, then you still don't get
it, and all because you refuse to do the math.

Y = 0.30R + 0.59G + 0.11B
(http://www.scantips.com/lumin.html)

Therefore, when you subtract Y from R and B, in the Pr and Pb
components, but at the same time you create this new Y component that
was not present in RGB, it is undeniable that by whatever amount info
is removed in the difference signals, that info is added to the
luminance. This is simple arithmetic, or at best Algebra I.

At the end of the chain, at the display, to reconstruct RGB from Y Pb
Pr, all you do is a little algebra. You reverse the above weighted
sum. Restore Y in R-Y (Pr), restore Y in B-Y (Pb), and then solve for
G using those reconstituted R and B values, which now include
luminance info. I thought you had figured that out when you
acknowledged that the Y Pb Pr is no less demanding of bandwidth than
RGB, if the difference signal aren't subsequently filtered down:

Now let's return to my original claim that the creation of color-difference
signals removed all the high-frequency detail from them.

I was wrong.

You really need to think these things through. Furthermore, as I
indicated previously, the difference signals MAY actually have to
carry exactly as much info as the Y component, when not low-pass
filtered. The example was the case of zero R or zero B in the
reconstructed primaries, which would only be possible if Pr = -Y, or
Pb = -Y.

Long monologues cannot dispute any of this. This is simple math.

Bert

"William Sommerwerck" wrote in message
. ..
PS: Wasn't "Salmon Ella" a relative of Typhoid Mary?

"William Sommerwerck" wrote in message
. ..
PS: Wasn't "Salmon Ella" a relative of Typhoid Mary?



Good one.