Still trying to figure the grey level of an 18% grey card (was: If I shoot a grey card, should this

I had several responses on this, all giving different answers!! I have
also seen three definitions of the meaning of 18% grey:

Definition 1: An analysis of 'average scenes' (whatever that may be)
showed they were 18% reflective. Can't imagine there is such a thing as
an average scene that would be in any way useful!

Definition 2: the mid point of a 5-stop scale. I can see that 100/18 is
around 2.5 stops. This sounds more plausible, but if you work on a
6-stop range, the point moves to 12.5% so there's little justification
for the figure.

Definition 3: It's just an arbitrary figure to which most exposure
meters are calibrated, by convention (and some do something else).

Perhaps the truth is a combination of these. But there must be a value
(between 0 and 255) that represents the same shade of grey as an 18%
card. I know that there is the gamma question, but surely this is to
essentially remove the non-linearity of the display tube, not to distort
the grey-scale itself.

A mid-grey that is scanned and then printed should look the same, no
matter how screwed up it is on the display. I would have thought that
the tube should *try* to display the same grey that is being scanned and
being printed (within the constraints of reflective vs transmissive
viewing).

So, the question remains: if I create a grey of 127 or 128, *should* it
look like an 18% grey card? If not, what level of grey should?

The root of my question is the following problem:
When I send mono work out for printing, I attach a grey scale of equal
steps (0, 16, 32, 48,.....240, 255). It is easy to recognise 0 and 255
when printed, but how do I tell them that their rendition of mid-grey is
too light or too dark?

I would need a full wedge to prove linearity across the range, and to
ensure that there is no black or white clipping. But I really want a
separate grey patch, labelled "this should look like and 18% grey card
when printed". Perhaps this is 127/128, or perhaps it's something else
(and that is the question!) The only reason to choose 18% grey is that
they would certainly have a reference card available to compare.

I'd appreciate any further comments on this perplexing (but essentially
simple) issue.

TIA

 

Have you tried scanning a grey card?

 

A grey card comes out at about RG&B 160 on a reasonably calibrated
monitor.

 

A few years ago there was a discussion in this group along the same line you
seem to be asking. I'm sure it made a few go out and buy a gray card and
understand a bit better about nuetral gray and it's visual perception under
varying light sources and intensities.
If you printer and paper were capable of producing a sample that could
reflect 18% gray then your problem would be solved.

 

In message <>,

That would be (0.18^(1/2.2))*255 = ~117. Most cameras and RAW
converters target ~127, though.
--

 

In message <>,

I would think that the answer depends on the paper used. Paper is in
its own little world of low contrast, and blacks that are grey in the
real world. I don't think prints even attempt to center around a grey
card.
--

 

In message <PAd0c.38058$%>,

Huh? Monitors don't have RGB values. They are passive analog devices,
displaying voltages created by the gfx chips. An average exposure will
be about 127 with most cameras and RAW converters. I've seen 117 as
well (which is what 18% grey should be) in one context, (but I don't
remember where).
--

 

Quite wrong. If you would take the time to compare a grey card to the greys
available in Photoshop you will find - that if your monitor is fairly well
calibrated you will get about 160. Don't talk theory when it takes vary
little work to see that your are flat out wrong. If you don't have a grey
card - buy one.

 

In message <kog0c.55216$>,

You're not making one bit of sense. Inability to communicate and
inability to detect failed communication and initiate a retry is an
epidemic disease, and you are infected.

Where is this "160" figure coming from?

A grey card's level in a room depends totally on ambient lighting.
--

 

It shouldn't have anything to do with the monitor!

 

What has it got to do with the paper used? Any media variations should
be taken care of in the printer profiling and the driver. You shouldn't
be adjusting your image to each media used, you should be adjusting the
printer transfer characteristic.

 

Completely correct.

You cannot map a grey card to any level in your digital picture
or on your monitor. A grey card simply reflects 18% of the light.
A monitor does not reflect light - it emits light.


/Roland

 

In message <Xns949E960127CB3klotjohan@130.133.1.4>,

So what the hell was he talking about? I can see why people killfile
him; he writes, but doesn't read.
--

 

In message <>,

People who don't really read and just look for keywords to agree with or
pounce on generally get that communicative feeling, even if
communication isn't happening.

Anyway, any solid-shaded grey object comes out at 127 in photoshop with
most cameras and software, not 160.

Maybe a sicilian or a black Irish (my paternal grandmother was one of
these, and my sister has a dark complexion), or someone with a South of
France tan, but most caucasians are about a stop more reflective than
18% grey.
--

 

In message <>,

Doifferent paper and ink combinations have different dynamic ranges,
regardless of what the software tries to do.
--

 

An 18% grey card has a reflectivity of 18%.

Now, how it is used in exposure determination is something else. This
is to a large degree up to judgement and desire of photographer when you
use manual metering. All of your 'definitions' could apply.

 

You are a complete fool - and a stubborn, blind fool at that. Since you
cannot learn I suggest you go live in my killfile with the rest of the
idiots. Good bye loser.
To others who may be curious: The "160" figure comes from comparing the
colour of a grey card with the colours on a monitor and on prints made from
swatches of those colours. Mr Sheehy has obviously never done a single test
in this matter - and would rather make noise than know the truth. Ignore the
dork.

 

If the monitor is calibrated -- even only moderately calibrated you will get
approximately the same grey at 160 as the grey card -- isn't that clear? Try
it. Open photoshop, make an 8x10 file and fill it with 160 grey (160R, 160G,
160B) then hold up your grey card -- they should be close - very close. Now
print the file -- your printer will probably add a cast to the file but it
will still be close.

 

[]

[]

What would be a reasonable question might be:

- if I expose for white = 255, what grey level should an 18% grey card be?

Of course, the answer depends on what gamma correction the camera software
has used when making the 8-bit image from the original sensor data.....

Cheers,
David

 

True. A more subtle question is "what does white mean?". If you expose
to map white to 255, does that mean the white of white paint, or a white
tablecloth, or a white shirt? All are diffuse reflectors that reflect
about 90% of the incident light, but they are likely to have somewhat
different colour. Also, do you expose for your white reference in
sunlight, or in shadow, if the scene contains both?

Or do you expose with some headroom above flat white in order to be
able to capture details from specular reflections, such as the sun
reflecting from shiny metal? It's not practical to avoid all
overexposure of such reflections, but it's often desirable to have them
at least a bit brighter than matte white.

The digital photography question "what pixel code should a grey card be"
has the chemical photography equivalent "what negative/transparency
density should a grey card be exposed and developed for. In both
domains, there is no absolutely correct answer - the best exposures in
different circumstances end up with somewhat different grey card
exposures. There are only guidelines to get someone started in the
right direction.

Dave