Comment on color preprocessing
Posted: Thu Apr 26, 2007 4:11 pm
1. My last paragraph is not directed at Mr.AL
Considering security CCTV:
Many older cameras sync to the line voltage to simplify the process of
putting multiple cameras on one screen. They also have an independent
horizontal oscillator with an indefinite amount of interlaced lines, such
that it would be possible to get ultra high resolution of a criminal if he
stood still for a second, that would be 15750 (more or less) lines. If
forensics have "CSI-like" tech than they could really do those magic
"image enhancement" zooms easily on the analog CCTV.
Considering color preprocessing:
B&W TV has none of it's own color circuits and is not as bandwidth
limited as color TV's are. Color processing is so necessarily sharp that
S-video supposedly exists to bypass the pair of color filters in the
two connected components. Composite NTSC video contains a color "burst"
sync pulse train to reset an oscillator which creates a reference to
the zero phase of the carrier. Different colors are different phases,
and as I and Q, (2 phases intermediate in converting to RGB), one
of the signals (I or Q) is more visible than the other. The lesser is
transmitted less for efficiency. A delay line brings these signals to
R-Y and B-Y, from which G-Y is generated and then all converted
to RGB.
I believe from experience that cyan and orange are represented by
one of the two phases referenced by the colorburst; since they are
complementary colors I will assume that they are the weak ones
(Just looked it up: orange and cyan are both close to the weaker phase)
and also may trick the average eye into seeing full color because
they are complementary to normal primary colors. But it is certainly
easy to extract these from a color signal, easier still to generate a
picture with them using other image sources than NTSC video,
for example, the VGA port, which is ANALOG RGB, at twice the
horizontal scan rate of NTSC. We know what can come out of the
VGA port is better than best HDTV.
orange/cyan has a resolution advantage over RGB in this idea, but
we could get weird and say blue is the least used color, so lets give
Red and Green 90% of the lines and 10% to blue, and use fast
progressive scan. Too Silly and Complicated for watching a football game.
No color wheels are suggested, they have a major flicker problem.
(BTW The only retail HDTV that doesn't flicker is LCD. )
My colored-BW multi-frame B&W tube HDTV idea is probably like
SECAM or PAL (phase alternating lines) format, except the colored
lines are grouped together on the screen. Cheap mirrors and magnifying
lenses (optically correct, vs astigmatic) are all that's needed for hardware,
and tweaking vertical hold, rotating the yoke wires, and only vert-syncing
alternating colorframes is most of the job electronically. The rest is altering
the input signal. I do think external components (such as a DVD player
with S-video, or a VCR tuner, maybe even a "replaced for something newer"
HD home theater device) may provide help decoding the video to
my format. DVD players decode MPEG although I've never tapped the
decoder and assume that most of them today hide the connections inside
a big chip.
HDTV over the air seems to involve an 8-phase "randomized" (supposedly
not meant as encryption but to look like very mild static) signal just below
the Y (B&W signal) carrier. Broadcast flag law would force filtering out
that signal or allowing the "digital tuner" to be remote controlled by the station.
B&W tuners are not considered digital, after all, they're so ancient that we
were still using "wireless telegraphs" when The Tube was invented. Rumor has
it that beyond the random key is MPEG data.
I encourage experimenting with small black & white TV's, just don't pull that
suction cup off the boob tube and use it as a pacifier. And don't whiz on the
electric fence. It won't defibrillate your love life as much as a DIY HDTV.
Considering security CCTV:
Many older cameras sync to the line voltage to simplify the process of
putting multiple cameras on one screen. They also have an independent
horizontal oscillator with an indefinite amount of interlaced lines, such
that it would be possible to get ultra high resolution of a criminal if he
stood still for a second, that would be 15750 (more or less) lines. If
forensics have "CSI-like" tech than they could really do those magic
"image enhancement" zooms easily on the analog CCTV.
Considering color preprocessing:
B&W TV has none of it's own color circuits and is not as bandwidth
limited as color TV's are. Color processing is so necessarily sharp that
S-video supposedly exists to bypass the pair of color filters in the
two connected components. Composite NTSC video contains a color "burst"
sync pulse train to reset an oscillator which creates a reference to
the zero phase of the carrier. Different colors are different phases,
and as I and Q, (2 phases intermediate in converting to RGB), one
of the signals (I or Q) is more visible than the other. The lesser is
transmitted less for efficiency. A delay line brings these signals to
R-Y and B-Y, from which G-Y is generated and then all converted
to RGB.
I believe from experience that cyan and orange are represented by
one of the two phases referenced by the colorburst; since they are
complementary colors I will assume that they are the weak ones
(Just looked it up: orange and cyan are both close to the weaker phase)
and also may trick the average eye into seeing full color because
they are complementary to normal primary colors. But it is certainly
easy to extract these from a color signal, easier still to generate a
picture with them using other image sources than NTSC video,
for example, the VGA port, which is ANALOG RGB, at twice the
horizontal scan rate of NTSC. We know what can come out of the
VGA port is better than best HDTV.
orange/cyan has a resolution advantage over RGB in this idea, but
we could get weird and say blue is the least used color, so lets give
Red and Green 90% of the lines and 10% to blue, and use fast
progressive scan. Too Silly and Complicated for watching a football game.
No color wheels are suggested, they have a major flicker problem.
(BTW The only retail HDTV that doesn't flicker is LCD. )
My colored-BW multi-frame B&W tube HDTV idea is probably like
SECAM or PAL (phase alternating lines) format, except the colored
lines are grouped together on the screen. Cheap mirrors and magnifying
lenses (optically correct, vs astigmatic) are all that's needed for hardware,
and tweaking vertical hold, rotating the yoke wires, and only vert-syncing
alternating colorframes is most of the job electronically. The rest is altering
the input signal. I do think external components (such as a DVD player
with S-video, or a VCR tuner, maybe even a "replaced for something newer"
HD home theater device) may provide help decoding the video to
my format. DVD players decode MPEG although I've never tapped the
decoder and assume that most of them today hide the connections inside
a big chip.
HDTV over the air seems to involve an 8-phase "randomized" (supposedly
not meant as encryption but to look like very mild static) signal just below
the Y (B&W signal) carrier. Broadcast flag law would force filtering out
that signal or allowing the "digital tuner" to be remote controlled by the station.
B&W tuners are not considered digital, after all, they're so ancient that we
were still using "wireless telegraphs" when The Tube was invented. Rumor has
it that beyond the random key is MPEG data.
I encourage experimenting with small black & white TV's, just don't pull that
suction cup off the boob tube and use it as a pacifier. And don't whiz on the
electric fence. It won't defibrillate your love life as much as a DIY HDTV.