Re: Even more about monitors and gun control, that last word <G>
Re: Even more about monitors and gun control, that last word <G>
- Subject: Re: Even more about monitors and gun control, that last word <G>
- From: email@hidden
- Date: Fri, 15 Jun 2001 15:25:23 EDT
In a message dated 6/15/01 1:22:48 PM, email@hidden writes:
>
1. On the Chroma4, there is one little square in the corner - this is
>
the
>
>
refresh detection sensor.
>
>
2. Colorvision's data sheet on the Spyder says "eight silicon
>
>
photodetectors, seven filtered sensors"
Thats exactly how I recalled it; wouldn't you want one unfiltered (full
range) sensor, plus seven varying filters, to get the 8 most useful data sets?
>
>
>
>
> Ah, this would be the basis of my differing assumptions. I look at it
>
in
>
>
> terms of color detection, and assume that three filters (like the eye's
>
>
three
>
>
> cone types) are blind to a lot, and that more sensors is moving more
>
in
>
>
the
>
>
> direction of a defraction spectro, with many bands. Now if the intent
>
is
>
>
> simply to build gamma curves for the three guns, then yes, a sensor tuned
>
>
for
>
>
> each monitor color should do the trick. But if the intention is to
>
>
actually
>
>
> distinguish whether the green phosphor is a bit more yellow, or a bit
>
more
>
>
> cyan, wouldn't extra sensors tuned to distinguish in between Blue and
>
>
Green
>
>
> or Red and Green add more detailed data to this process? So for simply
>
>
> downloading curves to the video card, yes, three sensors, or even a single
>
>
> luminance meter, should do the trick, but for defining the phosphor colors
>
>
> for a matrix profile (or, if we dare consider it, taking readings
>
>
>throughout the color space for a true 3d lookup table) would be improved
>
by
>
>
increased
>
>
>data.
>
>
>
>
>
> So the theoretical differences here would be your model where piecing
>
>
> together more parts is likely to introduce more error, versus my model
>
>
>where a ruler divided to eights is more detailed and accurate than one
>
>
divided
>
>
>only to thirds.
>
>
>
>
>
1. If the eye can't see it, it's not color.
Hmmm... the eye sees the light from an OTT-Lite as looking exactly like
daylight, even though the Ott-litespectrum is full of peaks and valleys at
different wavelengths while true daylight has a very smooth spectrum... and
what is illuminated by the Ott-Lite might not respond exactly the same as it
does under daylight... so I'd call the eye the final target value, but not
the best tool to get you there.
>
>
2. The intent is to measure monitor color to accurately set the white
>
>
point and balance the grays.
Thats is part of it; but only part of it; if you are not accurately measuring
the phosphor or dye colors as well, there is no opportunity to adjust the
result to account for monitor primaries versus the image colors! In other
words: video card level color correction is not as good a profile based color
correction (such as Photoshop offers) due to the inability to correct for the
phosphors (or dyes) at the video card level (I suspect this may be the most
profound statement in this entire thread, to those doing development work).
>
>
3. A seven channel spectro is not very good. Even the best spectro's
>
are
>
>
not
>
>
very good at measuring a varying light source like a monitor. That's why
>
LMT
>
>
makes a $35,000 colorimeter to measure monitors.
From the same guy who yesterday said: "making the Chroma4 the most accurate
device available today for measuring monitors."
So I think we both agree the goal here is accurate monitor calibration with
an affordable device, not which unaffordable spectrophotometric device can do
the very most accurate job of measuring monitor output.
>
>
4. If two of the light sensors measure overlapping colors then somehow
>
>
the value of the twice-measured color must be accounted for. This is
>
>
difficult
>
>
to do so I believe the Spyder has very sharp edge filters to try to
>
>
eliminate this problem. If seven sensors is a superior technology,
>
>
why are the specs so bad?
Measuring each color only once has the even greater weakness of not really
measuring the color at all; only measuring the luminance of that color, with
no meaningful spectral component: "its exactly this bright, but who knows
whether its red or a bit orange..."
Epson's listed scanner dynamic range is not as good as Umax's; but I find
Epson scanners to offer cleaner, more detailed darks, so I take the
published specs to be somewhat relative, after all they were measured by the
manufacturer. You don't really believe Sear's vacuum cleaners are 36 horse
power do you? <G>
>
>
5. Analogies are misleading - having seven sensors is not like having
>
a
>
>
ruler marked in sevenths - it is like have seven parts of a ruler that
>
you
>
>
try to glue back together so that the reconstructed ruler is undetectable
>
>
from the original ruler.
>
>
6. This discussion must be starting to wear on the interests of this
>
>
forum. So I will give you the last word but suggest we continue this
>
>
off-line, if you are interested.
Oh, I imaging it has provided entertainment, and hopefully education, to a
number of people, the others simply stopped reading the thread.
Thanks again for your input,
C. David Tobie
Design Cooperative
email@hidden