> The morale is, just because the colours in a Macbeth colour checker is
> correct, it doesn't mean that real life images will be.
> Consequently if your image looks pleasing, who cares what the Macbeth colour
> checker looks like, unless your task is to shoot a reproduction of this
> particular target...

THomas  and Stephen I thank you both  for your comments , and will throw in
a bit more discussion here.
 What I has been trying to achieve is a "pleasing " condition good enough as
to supply the best color reproduction of paintings. I have shot a second
round of paintings( the first ones were done before having the Card) and
color wise, I can say it was good, but not the best result, and I want to
improve on this). It follows that I am using very even strobe lighting .

My basic procedure consist in using a grey card to check exposure as per the
camera's histogram. Once the exposure is correclty set( usually compensating
for 1/3 or 1/2 stop underexposure , typical of the D60), I procedd to set my
custom white balance with a real white card.

>From this point on , I go on with the shooting of the paintings, no changes
in exposure or camera settings, right till the end. Open the files in PS,
assign sRGB and convert to ColorMatch to star color corrections .

WHile doing this with the Gretag Macbeth Card, the resulting initial file is
rather pale, no contrast etc, things I can easily solve, but achieving a
"pleasing" result is something I can say of every  file ( painting) I open
and "correct" with pleasing colors but results may end up really different
from reality when they go to the press, and I don't have acces to the
paintings after the shooting, in order to stablish visual comparisons..
Hence my plan to look for a systematic approach to color control, which ,
IMHO, should be easy to standardise if the set conditions don't change.

Which is why I have gone through extensive testing with the card, just to
find out similar  results: If I run a full set of Selective COlor
adjustments, by  means of which I get  very close to target values, and If I
run a simple proofing test ( using basic Swop  coated profiles), lots of
colors fall out of gamut. THis is when I got a mental block about the card.

In referance to target colors there is something  Stephen already wrote
about, but I want to add something more and ask something more:

None of my attempts to get -even close- to the target values was possible
with my D60 files rgeardless of any assigned profile I could try with to
open the untagged original files. What was  I doing wrong??
What I refer to as "target" values are those that come in a table of RGB
values along with the Chart. Andrew Darlow suggested me a reading in the
Bruce Lindbloom site.

http://www.brucelindbloom.com/

To make it short, after checking the info about the many tests performed by
Lindbloom himself on the Color Chart, I decided to write to him, asking for
help to understand why I was failing so BIG in my attempts,
THe response was more than dramatic, he fully agreed with me on the
discrepancies of the values reported on the GM card( the card comes with
tables of different values, corresponding to  what they call a "Pure" TIFF
file, but there is no mention whatsoever of any ICC profile of this target
file .)

With permission, here is  part of the reply sent by Mr.Lindbloom:


>Jorge, You ask an excellent
> question, one that I also wondered about. Where did those printed RGB values
> come from (the ones included with the ColorChecker chart)? I know the vice
> president of technology at the Munsell division of GretagMacbeth (who makes
> the charts) and I asked him that question last summer. Here is my question
> followed by his response:
> 
> My question:
>> 6) These values are radically different from either of the RGB sets
>> described above. What RGB color space are they in? Without an RGB
>> definition, these values are meaningless. (In fact, this particular set of
>> values is completely bogus in my opinion, no matter what color space is
>> used.)
> 
> His response:
>> These values were published by a 3rd party without our permission but were
>> included later by Munsell. THEY ARE TOTALLY WRONG AND SHOULD NOT BE
>> INCLUDED. It's purpose was to optimize a scanner by setting the white patch
>> to 255 and the black to 0 to get the best dynamic range from an old scanner.
> 
> So there you have it. THE RGB VALUES PRINTED WITH THE COLORCHECKER
>SHOULD BE *IGNORED*. I have great confidence in the RGB values I have on my
> site, for each of 15 different working spaces.

In fact, the values reported by Lindbloom on his site can be kept on
anyone's computer since he offers a free  Excel table to download and you
can check real values in 15 different Color Spaces.
I got proof of the veracity of the values reported by him thanks to an
independent download I was able to do from another site, offering a Tiff
file of the GretagMacbeth chart for calibration purposes, and its values are
impressively close to values reported by Lindbloom.

http://www.drycreekphoto.com/Learn/monitor_calibration.htm

I would love  to find out how to round up all this info into something
useful for me and everyone else, or receive advise about how to forget about
all this and proceed otherwise, running an optimised  alternative workflow
that can help me  standardise this sort of basic shootings, with simple and
reproducible  conditions.

> You can only place the grey after the black or white. When you click with
> either the white or black eye dropper any other adjustment to the (R+G+B)
> curves is erased. Check the individual channels after setting white, black
> and grey (in that order although you can do black before white).
Let me start by saying that I am talking about calibration with Levels in
PS, in orde to save the resulting values and be able to load them
afterwards.
I always leave the grey for last, and it falls elsewere, this is, at or
below the 120 point. And by doing only this, the result is still pale, so I
would still have to keep moving the grey slider ( in Levels) to achieve
"pleasing" results.

> Jorge, on the four middle gray patches between white/black, the L of LAB
> idealised reading is close to (light to dark, left to right):
> 
> 81L, 66L, 50L, 35L or in A98 200rgb, 159rgb, 119rgb, 83rgb (quickly measured
> of my digital colourchecker action based on published measurements).
> 
> It would be more critical to get the neutral patches neutral (zero AB LAB
> readings or R=G=B) - I would not worry about the brightness too much.
> 
> In a linear space the 128 value would be midway, but it is my guess that
> most do not use gamma1 edit spaces.

I 'll check on this values, and the reasoning of not working on linear space
may settle the issue,  but what then should be a standard value in gammas
1,8 or 2,2??


Excuse the long rant and hope I am not bothering you all. I have been really
trying to define a rationale and simple method to include the GRetag MAcbeth
into the calibration of cameras and/or files ,specially since the Canon guys
insist on NOT using profiles, their effectiveness being of real use under
the shooting conditions the profiles was built with. We should then have to
deal with as many profiles as shooting conditions we can be dealing with in
our work, and then calibration under each situation is just as useful,
cheaper and updated to real conditions every time.

   All the best and thanks in advance,.

  Jorge Parra
   APA/ASMP
w

===============================================================
GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE

Reply via email to