Color Correction by the Numbers for Ink Jets


hfdomke <missouri@...>
 

Dan Margulis's writes on p. 23 and 24 in "Professional
Photoshop 6" that for writing curves, the Highlight should be
5C2M2Y and the Shadow should probably be something like
80C70M70Y70K. This is geared for prepress and many of us
also print on high quaility ink jet printers that want RGB
numbers.

If I go into Photoshop's color picker and enter those numbers
and convert them to RGB I get these numbers:
Highlight: R248 G250 B255
Shadow: R0 G1 B12

If I can pick my black point and white point dropper values to use
with curves (such as in a program like Nikon Capture 2) would
these be reasonable number to enter if my output device will be
a 6-ink Epson Inkjet printer like the SP-890, 2000P or 75000? (I
use all three). What about if I was outputting to an Iris printer?

Thanks for the help!
Henry


Lee Varis
 

hfdomke wrote:

Dan Margulis's writes ...the Highlight should be
5C2M2Y and the Shadow should probably be something like
80C70M70Y70K.
If I go into Photoshop's color picker and enter those numbers
and convert them to RGB I get these numbers:
Highlight: R248 G250 B255
Shadow: R0 G1 B12

If I can pick my black point and white point dropper values to use
with curves ... would these be reasonable number to enter if my
output device will be a 6-ink Epson Inkjet printer
You can not extrapolate accurate RGB numbers from CMYK aim points. The
numbers you cite are highly dependent on what RGB & CMYK workspace
profiles you are using. I'm guessing that Dan's numbers are based on a
different kind of CMYK setup than what you've got set in your PS Color
Settings (you don't mention what you are using there- when I do this in
my set up I get 251R 249G 248B for white). Also the CMYK gamut is
smaller in most areas of color and several RGB colors may have the same
CMYK equivalents (dependent on rendering intent settings that affect
gamut compression) - the numbers you end up with won't necessarily match
the ideal RGB starting numbers for these CMYK aim points. I would prefer
to find neutral black and white points for RGB images, something like:
R240 G240 B240 for white (sometimes as high as 250 ea.) and R5 G5 B5 for
black (sometimes I like values as high as 10 for black) These numbers
are for highlight and shadow where you want to hold some sense of detail
or texture - a specular highlight should still top out at 255 ea. with 0
ea. for deep black shadows. Most of the time you want something neutral
in these areas and neutral in any of the standard PS workspace profiles
will be equal values in RGB. In CMYK if your numbers in these extreme
areas are off a few points it won't be that noticeable but your RGB
numbers seem to be definitely on the cool side and may force a cool bias
in other areas of the image if you use them to auto correct.

I also question the accuracy of the color picker to do conversion
calculations anyway - try creating patches of color in RGB and then
actually convert to CMYK to test this out. I find that the resulting
numbers are sometimes off a little bit from the color pickers
"prediction" (though not so much with these "end point" numbers). I'd be
interested to hear from someone at Adobe how Photoshop arrives at the
color equivalent numbers in the color picker.

--
Regards,

Lee Varis
varis@...
http://www.varis.com
888-964-0024


Stephen Marsh <samarsh@...>
 

Henry writes:

Dan Margulis's writes on p. 23 and 24 in "Professional
Photoshop 6" that for writing curves, the Highlight should be
5C2M2Y and the Shadow should probably be something like
80C70M70Y70K. This is geared for prepress and many of us
also print on high quaility ink jet printers that want RGB
numbers.
Then as has been suggested in a recent post by Andrew Rodney - you convert
from your CMYK tag or assumed workspace profile set-up in colour settings -
to the profile for your inkjet, on a copy of the original file. This RGB
file is then output to the Epson profile which simulates your press or acts
as your contract proof, or pre-proof guide or whatever.

If the CMYK numbers in the file match the assigned/assumed profile - then
you will get true neutrals.

This will not happen with standard aimpoints - you have to follow the LAB
values in the profile when making your curve moves in CMYK. This does not
really matter for most people, but if you are expecting a neutral conversion
out of CMYK then the right numbers and profile have to be used. What most
people seem to do is hit standard aimpoints such as 5/3/3 or whatever, even
if this is not providing neutral values via LAB readings for the profile.
This colour may proof or print on press as neutral - but to the profile and
any mode conversions, it is not neutral (since the CMYK numbers are not
synced to the file).

If I go into Photoshop's color picker and enter those numbers
and convert them to RGB I get these numbers:
Highlight: R248 G250 B255
Shadow: R0 G1 B12
This demonstrates my point that if you alter CMYK numbers so that they are
out of sync for the profile - and the profile is expected to deliver neutral
colours in LAB or RGB - then you may have another thing coming.

This may or may not be an issue - not many convert CMYK to other modes as a
workflow rule (apart from channel blend tricks etc), and when they do they
do not care that the 'closed loop non icc numbers' do not match the profile
for neutrals, since they may care about the actual final CMYK values in the
final film or plate more than having them match the profile.

Some users do like to keep the neutral values in the file matching the
profile - so that conversions out of CMYK are accurate to the profile,
probably more so for neutrals throughout the tonal range but not so much in
the highlights/shadows (where more generic aimpoints might be used, even if
they disregard the profile).

This is not easy to comment on, there are a lot of ways that people edit
when in CMYK.

If I can pick my black point and white point dropper values to use
with curves (such as in a program like Nikon Capture 2) would
these be reasonable number to enter if my output device will be
a 6-ink Epson Inkjet printer like the SP-890, 2000P or 75000? (I
use all three). What about if I was outputting to an Iris printer?
You would ideally need to profile each device using the print settings,
ink/stock/resolution etc - and I do not mean ICC profile, although this
would be ideal.

This is just like finding total ink limits on a press. Head on over to this
page and download this file. Stuffit expander required for decompression:

http://www.thelawlers.com/essays.html

http://www.thelawlers.com/FTP/Gray%20step%20chart.sit

This is a 100 step grayscale chart in vector EPS format, each patch varies
by 2-3 RGB levels when converted to RGB. This image can also act as a gray
balance test image when converted to RGB or CMYK, but would also ideally
include a grayscale gradient as well to see if 'crossover' is an issue on
inkjets. I am not suggesting gray balance at this point (that's ICC profiles
and CM) - I am just trying to help you find suitable RGB aimpoints for your
common devices, if you wish to optimise your RGB before print.

Rasterize into Photoshop as RGB at say 120 or 240 ppi, assign your WS RGB
(workspace, i.e. Adobe RGB etc) and save as PSD.

Now convert or print/convert from workspace to Epson RGB or however you
normally print for the Epson or IRIS - just convert to the appropriate
profile for the intended output.

Visually evaluate the step chart so that you can find the patch where the
min and max tonal values start/end.

Use these as guides and adjust after running these aimpoints on test images
with real image content, the shadow detail will probably vary more on
natural images than the step chart - but you should have a max limit to
shoot for.

I think there are two different workflows here...

a) Using an inkjet to simulate a press or proof

b) Alternate workflow for optimising images in RGB for an RGB output

Method (a) is trying to exactly reproduce the CMYK file within the limits of
the profiles/device etc on an inkjet.

Method (b) does not care about matching the CMYK file, it is about making
the inkjet perform to its best.

Regards,

Stephen Marsh.