#WFM V-Scope Techniques
11 messages in this thread
Yet another question <grin>
Re: Color Bars
Here are RGB values for Color Bar Test Patterns provided with the Pinnacle
Alladin and the NewTek Video Toaster (Maybe other DTV product owners can check
theirs…)
Color Video Toaster Pinnacle Alladin
R G B R G B
White 189 189 189 255 255 255
Yellow 190 190 0 189 191 0
Cyan 0 189 188 0 192 193
Green 0 189 0 0 192 0
Magenta 189 0 189 189 2 190
Red 191 0 0 189 0 0
Blue 0 0 188 0 3 191
_blue 0 14 61
_white 253 253 253
_purple 34 0 94
_black 0 0 0
_bar1 0 0 0
_bar2 0 0 0
_bar3 9 9 9
_black 0 0 0
The questions I have are why aren't the values the same? Is NTSC (Never the
same color twice) that forgiving? Why don't the Toaster's Bars reflect _bar1
and _bar2, I know that they are supposed to be there, can RGB not do it? or
something?
Thanks for your input.
-John
–
An afterthought John, about what I just said. My recollection is that you are
a FAST user. The spec for the Pinnacle bars looks like the same goofy bars
that the VM generates.
As to the differences in the non-white chips, I think the toaster values are a
tad low – slightly below 75% and the Pinnacle chips are a tad high – slightly
over 75%.
Bob
John,
I'm not sure why you should be concerned about RGB values. What matters is
what comes out the spigot to the final tape/program feed, etc.
I haven't tested Pinnacle (don't have on as yet). BUt I have looked at the
bars from the toaster alot (I have two of those). With the Waveform Vector
scope, the Toaster bars are dead-nuts on in both intensity and phase. And
that's all that matters. Capturing them to some RGB paint program (or worse
yet, two different RGB paint programs) and then looking at the values really
has no bearing on what you want – boradcast legal.
Now, for an additional thought. There are what are known as 75% bars and 100%
bars. They have different values for the bars. I use exclusively 75% so I
can't eleborate on the 100s. But the top white bar is actually not 100% white
– the bottom one is. The top one is actually considered a 75% gray.
Also, if you bar1, bar2, bar3 is a reference to the pluge pattern, it is there
on the Toaster bars. If you have a monitor with blue check, bright, contrast
controls, you can tune the monitor to properly reflect NTSC values. In such a
situation, you turn the black level (brightness) up until all three bars can
just be distiguished. Then you turn the contrast (picture) fo that each step
appears about twice as bright as the the preceding one. Then you turn the
brighness down until the two left hand stripes look alike. You should still be
able to barely see the third stripe. At this point you are ready to set the
chroma gain and phase controls for your monitor.
So basically, unless you have a Waveform/Vscope and/or a properly calibrated
monitor, there is no way to tell which unit is performing to standard. But I
can tell you that most Toasters are dead on with a few that were a little "hot"
(10 IRE black instead of 7.5). I assume the Pinnacle's are too, as they would
not have gotten such favorable press in the pro video rags if they couldn't
even output color bars to spec.
wmc – via Autopilot!
Wayne, I am agreement that what comes out is very important, however, just as a
seperate Y/C signal will almost always look better than a composite signal, I
am always looking for the best picture.
I only looked at the bars shipped standard with every toaster, in ToasterPaint.
Then I compared them to the bars shipped standard with Alladin Paint.
Therefore the difference. The Alladin bars are 100/75 and the Toaster bars are
SMPTE.
However, RGB is RGB and therefore the question of why a "red" of rxgxbx can be
different from a "red" of ryxygy. See what I mean?
Now, the "Pluge" pattern… I figured there was a name for them, thanks alot.
What I can't understand is that, in ToasterPaint the values of _bar1 and _bar2
are both 0 and the value of _bar3 is r9g9b9. Seems to me that (this is crazy)
the encoder is turning a r0g0b0 into something else, which is crazy in the
computer world, because as you probably know, it would take code that says 'hey
these are our bars, let's invent a Pluge pattern'. See what I mean? It has to
be there in the RGB.
Love the technique in your article, thanks Wayne.
-John
John,
Your confusion on the Pluge is the root of my argument: Not all RGB programs
have the same 'precision' and the mechanisms and algorithms of transfer from
video to RGB encoding are largely manufacturer dependent. So if you REALLY
wanted the correct comparison, run the Pinnacle into a Toaster, run another
Toaster into a Toaster and capture both source's bars in TPaint then look at
the RGB values. Or, if you prefer the more expensive route run one Toaster and
one Pinnacle into another Pinnacle and capture the bars from both sources in
the Switching pinnacle….
You could drive yourself nuts trying to justify the RGB's of two manufactures
products compared in different paint programs. Especially in the case of
ToasterPaint. (I forget which, but it is actually only 12 or 18 bits of color
information which is why you probably didn't see the full Pluge). One problem
with Toatster paint is in trying to do the RGB to NTSC two-step, they use a
dithering pattern that, if sampled even two pixels appart often gives different
color values eventhough it looks the same on the monitor.
Anyway, early color TV engineers probaly went thru the same thing you did –
when is red red enough and that is how the whole WavformVScope thing came
about. They basically said, "to heck with what you feed it, or how you mix it
up, just make it come out the program spigot at x IRE and y phase if its color
a." Then they divised proc amps to "teach the test".
wmc – via Autopilot!
Wayne,
Thanks for the help. It does make since to me now that the RGB values of color
bars for different encoders will vary because of the encoder and that the right
place to measure the values is through a Waveform/Vectorscope connected to
program out.
-John
In order to get the I and Q vectors on SMPTE bars you would need to generate a
negative RGB value. Since it is difficult (if not impossible) for most PC-type
systems to do this, you won't find these generated on most computer systems.
My understanding is that the pluge also requires a negative RGB value for the
blacker-than-black portion. (If you want the true functions as you find them
in test signal generators).
The probable reason for the differing values is that there is a variance in the
encoder/decoder circuits of the devices. Since they use different resostors,
caps, etc., they need to massage the test images they ship with the unit to
give ideal results. This is not uncommon, I know of many companies that do
this in the PC world.
What you need to look at is the end result. If it is RS-170A/SMPTE 170M and
fits your WFM/VS as it should, you are OK.
James Bell Jr. — Multimedia Contributor – CompuServe
PMFJI James –
>>In order to get the I and Q vectors on SMPTE bars you would need to generate
a negative RGB value
Could please amplify on that? I'm not sure what negative rgb values mean or
how they could be generated. 'Course there's lots of stuff I don't know<G>.
Bob.
Since we're talking about gray values, the I and Q (Inphase and Quadrature)
signal are zero – no color at all. The only reason that black is at 7.5 IRE
rather than 0 IRE (0 volts) is to hide the retrace lines on TV sets of decades
gone by.
The Video Toaster does create the three pluge bars correctly (well, more or
less – the Toaster's levels aren't as exact as they could be, especially on the
original Toaster). Luminance values below 0 IRE are definately not
NTSC/RS-170A legal; they're starting to encroach upon sync territory at that
point. Any chrominance riding on a low luminance value (ie- blues & magentas)
are allowed to drop below the 0 IRE value since they will be filtered out
anyway before reaching the sync circuits.
Color bar signals and color bar generators are not all the same. As with any
test equipment, I would want to verify what is claimed versus what is
generated. Some equipment can generate full field color bars (ie- MX50) that
might serve as a reference, but I would much prefer using a proven source of
SMPTE color bars.
In measuring the color bars, I'm not confident that they can be inspected via
framegrab and software techniques with any precision. This would explain the
pluge measurement anomally in the Toaster. If you think about it, NewTek made
it very difficult (maybe impossible) to generate blacks less than 7.5 IRE (much
to everyone's dissatisfaction). Yet we see a 6 IRE pluge bar; obviously they
have some trick to load the framestore with such values.
Bruce
Bruce,
Thanks for the explanation! I went back to the Toaster and when I load a
framestore from the Switcher, I do indeed get the correct pluge pattern.
However, If I load the same framestore into Toaster Paint and then render to
Program Out, I only get one of the three bars. Also, If I grab the correct
pluge pattern into Toaster Paint and then rerender to program out, I lose the
pluge pattern. It is an interesting feature. 🙂
-John
James,
Thanks for the information re: negative RGB values. I think that I had a
feeling that there was something like that involved. Is the blacker than black
signal called the "floor" in the video world or do you know what it is called.
Also, do you know if it is used for anything special, like at the beginning of
a commercial or something? Thanks in advance…
-John