Date   

Color Correction on LCD displays

Marco Cappuccio <marco@...>
 

Is anyone doing serious color correction on laptops or LCD displays? I
have found the color gamut of all LCD displays to be smaller than that of
a quality CRT. In addition, the angle of view dramatically changes the
perceived color. As a result I won't use a LCD display or laptop for any
color work. But the need to perform color corrections while traveling is
becoming greater. Has anyone found a workable solution?

-Marco Cappuccio


Re: Printing Overseas

Chris Murphy <lists@...>
 

Our images are set with the dot gain settings that our printers
have provided us (12%).
Which printer? The local one? Or the one in Hong Kong?

1.) The local printer's dot gain won't apply to the one in Hong Kong, so
if it's coming from the local printer, it isn't valid.

2.) If the Hong Kong printer doesn't have a lot of experience working
with negative platemaking, then their dot gain value is probably wrong
also. I have yet to see a negative platemaking process with film that has
dot gain that low even printing on a wonderful glossy coated paper which
I assume you aren't. That dot gain sound like CTP dot gain, or a positive
platemaking process dot gain with 8% added by someone who thinks negative
platemaking has 8% higher gain than positive platemaking.

The two processes are completely different. So I'm inclined to think that
this 12% figure is artificially low. Is there any chance they are doing a
copydot scan of your film, then making new positive films and using a
positive platemaking process?


Our printers have told us they can provide us with a colour profile for
their presses but they are a Mac shop and we are running PCs so I'm not sure
that their profiles would be of any use to us.
Not a problem. Just add the extension .icm to their profiles and they
will work fine. Where they go depends on your Windows operating system,
because the Joys of MicroSoft have decided to put ICC profiles in a
different location in each of their five operating systems.

If you do a search for .icm you should find a folder full of them and
that's a pretty safe bet that's where they should go on your machine.
Newer Windows OS's will recognize either .icm or .icc.


We have asked our printer to provide printed samples and files for these
samples so that we might compare their work on screen to our own and get a
better sense of how we can improve our colour correction to get what we
want. Is there a more accurate way to do this?
Sure. Have them print some small patches (1/2" x 1/2" is fine) in a trim
area of 100% cyan, magenta, yellow, black; and overprints red, green,
blue; and a 3 color black. Those colors you have someone measure xyY
values and plug those into Photoshop's Ink Colors pop-up as Custom. Also
have them print (it can all be done at the same time of course) at least
a 50% dot area so you can measure the dot gain and plug that into Curves
for each channel (so you need 50% cyan, magenta, yellow, and black).
Photoshop lets you enter 13 points for each channel, but you generally
don't need that much data. Minimum I'd get something in the middle, like
40% or 50% - then something at each end like 8% and 80% or 90% (or both).

Have them send you the prints and then you can have someone locally get
you these measurements (ask your printer for their ink rep, he might be
inclined to take the 15 minutes with his spectrophotometer he owns to get
these measurements for you).

Keep in mind that this is going to be based *their* positive platemaking
process. If you want to use negatives and find out their press behavior
based on negative platemaking, then you will have to send them films with
this test, and have them print it (and pay them for it) and then get it
measured.

Separation information is condition specific. That includes RIP,
imagesetter, film, plates, ink, temperature, humidity, blanket pressure,
ink/water ratio - all those variables make a different with separation
information. What that means is, you can't take accurate separation
information and then change one of those variables (such as using
negative platemaking when the basis for the separation info is for
positive platemaking).



Is this a case where colour profiling software is the answer, given that our
printer has a profile built to his press?
Perhaps. I'd be curious to see if it actually works, and the way you do
that is you make all of your separations using their profile, and then
get a contract proof that they will accept that simulates their press
behavior. Now the first time around this might mean sending them digital
files, and for them to have to *SHIP* you your contract proof. But at
least you'll see if their profile is good enough and working. At that
point it's their job to match the contract proof.

It's not so simple to just buy some software, push some buttons, and this
problem is fixed. It takes more effort than that. If all our problems
could be fixed by buying software and pushing buttons, man oh man would
we all be a lot more sane than we are.

Anyway, there are a couple of problems that need to be solved as I see it
whether you decide to adopt some form of color management or not:

1.) You need a way to get contract proofs that they will accept and
match. I would focus on this first because no matter what else you do, if
you are sending them negative based Matchprints and they aren't matching
them good enough, then that's a dead end. You need to find some kind of
proofing system they will accept, and that they can match.

2.) Maybe it's possible to use their profile, maybe not; but once you
find a contract proofing system that you both can work with (and
naturally you have reasonable access to) you can target that system for
your separations and then it doesn't matter if you use their profile or
not. If you get good looking proofs, and they've agreed to match them,
then that's that.

3.) Now, theoretically if their press profile is good enough, and your
tolerance isn't extremely high; you could just use their profile for both
separations and making an in-house digital proof. They probably won't
want to accept your proof as "contract" but you could send it as a
guideline saying the proof is based on a profiled inkjet (or whatever)
and their press profile. If their press profile is good, it will work.
There are bunches of people who are doing this with large format printers
on the Computer To Plate Pressroom list serve. So it can work, it's just
a little risky because there is no film, there's no contract proof - etc.
So this is one of those deals where it's easy and if it works, great; but
I wouldn't be doing this or recommending it for really really important
stuff. It sounds to me like it's important enough to be concerned about,
but not important enough to be rejecting the problematic results you've
gotten so far. So perhaps this is a viable option- at least at some point.



Is there any way I can take their
Mac profile and make it work with Photoshop on the PC (we are using
Photoshop 5) without spending a bundle on profiling software? Or do we need
to find another proofing system?
You definitely need to be looking at finding a more appropriate proofing
system that both they will accept *and* match *and* you have access to.
Three whammies.

I would not concern yourself with a profiling package at this point. Even
if you got one what would you do with it at that point? Fly to Hong Kong?
I mean if it's getting to that point of seriousness call me :) But
seriously, you can use their profile and see if it's good enough in the
meantime. Just add .icm and stick it in the right folder, restart
Photoshop, go into CMYK Setup, click on the ICC button at the top, and
then pick their profile in the pop-up list.


We would like to go to CTP printing but we are very nervous about not being
able to properly proof our colour.
That's the whole point behind contract proofs. A contract proof needs to
be something that you can get, and the printer will accept *and* match.
If you both sign off on it, then that is what they should be matching. It
doesn't matter how it's made. If it's made by a 4 year old with crayons
and you both agree to it, then that is the contract proof.


Chris Murphy


Re: Photoshop 6 question

Chris Murphy <lists@...>
 

Maybe I'm not understanding you correctly, but if you click on the "Save"
button while in the CMYK setup dialogue (using the built-in option) you can
save your setup and thus give it a name.
You have a bunch of images saved by Photoshop 5 with a profile embedded
based on a built-in separation setup. When you make a built-in separation
setup in Photoshop 5, you cannot name the separation and there is no
saving the separation. It saves itself automatically.

The save button is a totally different story that is only making this
complicated to understand and explain. So for a minute, totally ignore
that option because it really doesn't apply (you'll just have to trust
me).

Photoshop 5 and embedded ICC profiles:

1.) Photoshop 5 embeds ICC profiles into images no matter what CMYK Setup
model you use; built-in, ICC, or tables. If you embed profiles in images,
whatever setting in CMYK is current is saved as an ICC profile - even if
you didn't *create* an ICC profile. It's something that just happens.

2.) If you use the built-in model, there is no way to name those
settings. If you change dot gain from 20% to 22% and click the OK button,
the settings have been changed and Photoshop will use them. But you have
no option to specify a NAME for the resulting ICC profile that will be
embedded in your CMYK images. Photoshop generates this name itself based
on the settings.

3.) The automatic naming convention (that you have no control over), has
the following guideline:

<inkset> , <dotgain>

So if you have the default set (which is the "SWOP (Coated)" inkset and
20% dot gain), the profile name is: "SWOP (Coated), 20%"

If you do nothing but change the dot gain in built-in CMYK Setup from 20%
to 22%, then subsequent profiles have the name "SWOP (Coated), 22%"

If you use custom inks and curves in "built-in" in CMYK Setup, the
profile name will be: "Custom Inks, Curves"

*There is no way to change this.* You cannot come up with your own custom
naming convention for Photoshop 5 profile embedding when using
*built-in*. The way around this is:

1.) Make your built-in settings.
2.) Click on the Tables radio button in CMYK Setup - this loads the
built-in information as a table.
3.) Click the Save button and give it a logical name (this is the ICC
profile's filename and internal name you will see in embedded images, and
in pop-up lists).
4.) Save it into the ColorSync Profiles folder.
5.) Quit Photoshop and restart it so it will see the new profile.
6.) Go to CMYK Setup and click on the ICC radio button
7.) In the pop-up menu, select your newly created profile.

Now the same built-in information is being used for the same results,
except you have control over the name, and hence the name that gets
embedded. If you use built-in, you have NO control over the naming of the
profile that gets embedded.



When you open the tagged image
file, the mismatch dialogue box will display the name of your CMYK setup as
defined earlier (if you change your CMYK setup to something else).
In Photoshop 5, there is only one way to specify a custom name for
embedded profiles. Otherwise it uses the "<inkset>, <dotgain>" convention
I mentioned previously. If the profiles that are embedded in your images
don't follow this format, then they were embedded in a way other than
using built-in as the current CMYK Setup at the time the file was saved
from Photoshop 5.

What PS6 doesn't do is recognize that name as being the same, even though
it is the same CMYK setup file created in PS5.0/5.5. Sounds like a bug to
me...
If the letters aren't *exactly* identical in the profile names, then you
will get a mismatch. I just spend 15 minutes playing with this and I'm
not finding any bugs or unusual behavior. When I take a Photoshop 5 TIFF
with a profile embedded in it based on a custom inks/curves built-in
setting in CMYK Setup, and open it in Photoshop 6, I get a Profile
Mismatch that says:

Embedded: Custom Inks, Curves

Working: U.S. Web Coated (SWOP) v2


What does your Profile Mismatch dialog say for Embedded: and Working:?



Chris Murphy


Re: Optimal TAC, dotgain for b&w newsprint

Dan Margulis <76270.1033@...>
 

Stephen writes,

I guess and presume that none is 0%, light is 25%, medium is 50%,
heavy is 75% and maximum is 100% GCR...I am probably way off base
here, and I'm sure that I will be corrected by those who know better.>>

This is way off base. GCR is too complex to be described by a single
number. It's basically a combination of 1) what is the lightest gray
component at which black starts to appear; 2) how rapidly does black
increase in relation to CMY thereafter; 3) how the shadow is treated when
we start to get close to the ink limit.

I commonly use these settings:
SWOP Newsprint 30% Dot Gain
GCR separation type
80% Max Black
230% Max Ink (TIC/TAC/etc)
Heavy Black generation
0% UCA>>

Heavy black generation isn't standard although some use it successfully.
230 max ink is too low. Most newspapers say they'll accept 240, which means
they'll accept 250. So, I'd definitely move it up at least to 245.

I think that MAX GCR and 30-50% UCA *might* be the answer, but this
is too radical from my experience to trust to a live job.>>

This will definitely *not* be the answer. Maximum GCR is not usable, IMHO,
for serious reproduction of photographs. Its function is for the rare cases


I also must have the dot gain wrong, as images still need curves
adjustments to lighten the tones throughout the image - otherwise it
just runs too strong.

What you probably have wrong is the *black* dot gain, which is ordinarily
higher than that in CMY. Couple that with the heavy black generation you're
using, and the image will be too dark. The overall dot gain may be too low,
but the black is too low for sure.

Also for a 85-100 lpi screen, I have found that you can sometimes go
as low as 100 ppi when the image is S/S (same size).>>

This is image-specific. The more fine detail there is in the image, the
more conservative one has to be with resolution. 150 should be fine for the
overwhelming majority of images, but for images that are softer by nature
one can go lower, and 100 isn't unreasonable. I've seen plenty of
successful results where the resolution is actually *less* than the screen
ruling, but it takes a certain kind of image.

Dan Margulis


Wavelengths to RGB

Dan Margulis <76270.1033@...>
 

Alan writes,

In Photoshop's (and other software's) color pickers, hue values come in a
circle from 0 to 359 "degrees" through 0 again. From this one can create a
linear rainbow spectrum image, say 800 x 100 pixels, red on the left,
violet on the right and no change vertically. In the physics lab one can
also create a rainbow spectrum of the range of visible colors. If you scale
these two so their colors match at each end, the colors in the mid range
will not match.This is like an analogy, in hue, of a gamma matching problem
in brightness.>>

In reality almost any two methods of generating the spectrum would run into
the same problem.

Here's another way to describe this mismatch: I am 55 years old so I
learned "color theory" at art school
long before desktop computers and Photoshop even existed. We learned that
if you arrange the spectrum in a circle,
green is opposite red, etc. But in image editing software, cyan (180
degrees) is opposite red (0 or 359 degrees). The first of these seems truer
to my perception, with green offering the strongest contrast in hue to
red.>>

No, the strongest contrast would be a blue-green. Green and red share a
yellow component. Pure red is 255r0g0b. Pure green is 0r255g0b. While that
isn't particularly close to red, clearly if the blue value rises it goes
even farther away. 0r255g255b is the opposite of 255r0g0b.

<<This may just be early conditioning rather than unbiased perception, but
I doubt it because of what I will term "the Op Art experience": if you
paint equal width stripes of two fairly saturated colors, with equal
brightness, they jazz the eyes (appear to jump around) most when you choose
two colors that are opposites in the traditional theory rather than the
current software model (ie red-green and orange-blue rather than red-cyan
and
green-magenta).>>

This test is impossible, because pure CMY colors are much lighter than pure
RGB. Trying to compare red to cyan where the two are of equal brightness is
like comparing a basketball to a golf ball where both are the same size.

<<For some experiments in human color perception (both normal and various
types of color-blind), I want to create a number of images, with colors
defined in 24bit rgb (the human subjects will see them on a monitor).>>

If they are going to be seeing them on a monitor, there's a limit to how
much precision is necessary. But the real problem is using hue as the
basis. It is, as you say, linear; certain small variations in hue are
easily perceived and others of equal value aren't. The best currently
implementable way is to use L*a*b* increments as the base. This gives two
variables, a magenta-green and a yellow-blue axis. The value scale is
logarithmic, the increments are fairly well related to human perception,
and best of all, it corresponds closely to the most common forms of vision
impairment. For example, the most prevalent form of "color-blindness" is
actually a total inability to evaluate magenta vs. green, often coupled
with mild to severe impairment of yellow vs. blue. By setting up your
scales this way you can filter out a lot of problems caused by impaired
color perception on one axis.

In technical terms, I'd prefer the spectrum images to be logarithmic
rather than linear. But don't worry if that sounds too complicated;
almost any source would be better than what I have found so far!>>

Another argument for the L*a*b* method. A good source of information on
this, with a lot of links to esoteric stuff, is
http://www.inforamp.net/~poynton/ColorFAQ.html

Dan Margulis


Re: Optimal TAC, dotgain for b&w newsprint

Christine Holzmann <tekila@...>
 

To all,
I just want to apologize for not having yet responded to some of your queries/comments. I am still going to add some comments in this area, as it is a REALLY important topic to me as it is to all of you.......I am just on deadline right now and don't have a split second to spare until tomorrow or so..........:(


Crissi
_______________________________________

Firstly, I know that talking about colour is 'off topic' to this
post,
so I will respond to the original message regarding monotone work.

I have a question for Christine or anyone elese who would care to
comment.

--- In colortheory@egroups.com, Christine Holzmann <tekila@m...>
wrote:

TAC = 260
Black ink limit = 80
I prefer to use UCR for newsprint....you get the skeleton black
plate that way with less muddiness.
Christine, I do both mono & colour news separations each week for
local and city papers - I am hoping to pick the lists collective mind
on this subject, and this seems like a good opening...

No paper we deal with in Sydney seems to offer sep tables or ICC
profiles for their material suppliers. When they do give specs, they
are contradictory or are aimed at the traditional separator or drum
scanner operator.

I am the first to admit that I know just enough to be dangerous, but
to me MEDIUM GCR is not 70% replacement?

As far as I know, Photoshop does not deal in numerical values by
default, it has none, light, medium, heavy and maximum GCR levels.

I guess and presume that none is 0%, light is 25%, medium is 50%,
heavy is 75% and maximum is 100% GCR...I am probably way off base
here, and I'm sure that I will be corrected by those who know better.

For our printers, grey balance and registration seem to be a common
problem - not to mention consistent dot gain.

I commonly use these settings:

SWOP Newsprint 30% Dot Gain
GCR separation type
80% Max Black
230% Max Ink (TIC/TAC/etc)
Heavy Black generation
0% UCA

This cures most registration and colour balance problems, as most of
the neutral tones are in the black plate.

I think that MAX GCR and 30-50% UCA *might* be the answer, but this
is
too radical from my experience to trust to a live job.

I also must have the dot gain wrong, as images still need curves
adjustments to lighten the tones throughout the image - otherwise it
just runs too strong.

Also for a 85-100 lpi screen, I have found that you can sometimes go
as low as 100 ppi when the image is S/S (same size).

Some newspapers use JPEG internally in their FTP, so images get
degraded further, whether you like it or not (but considering the
output, this does not really matter).

Can anyone comment on my choices for colour newsprint work, or give
some rationale to why other settings should be used?

Image content varies, and is too numerous to describe. I realise that
there is no magic 'single setting' for all images, but there must be
some rules to follow (I will shortly purchse Dan's book, as it seems
that the answers to many of my questions will be found there).

Thanks,

Stephen Marsh.



To unsubscribe from this group, send an email to:
colortheory-unsubscribe@egroups.com
--
DESIGN EDITOR @ The Citizen News
http://www.thecitizennews.com

DIGITAL/GRAPHIC ARTIST
http://www.crissi.com


Re: Photoshop 6 question

Andrew Engelhardt
 

Just to clarify for everyone, we're running Photoshop on Windows so I'm wondering if anyone else using Windows has encountered this. "Just for fun", I tried creating a new CMYK table in PS5.5 and loading it into 6. Same thing happened when I opened an image, the profile mismatch pops up saying> Embedded: test_table, curves >Working: test_table. Am I doing something wrong in my color settings set-up? Is this possibly a Windows bug ? It seems to me that this is sort of defeating the purpose of CM if it's telling me the exact same set-up is wrong and is going to cause no end of confusion for our operators. Any suggestions appreciated.

Thx again.

Andrew Engelhardt



Andrew Engelhardt writes,

>>Maybe I'm not understanding you correctly, but if you click on the "Save"
button while in the CMYK setup dialogue (using the built-in option) you can
save your setup and thus give it a name.>>

That's correct. You have now named the setup and the embedded profile will
reflect that name.

>>When you open the tagged image file, the mismatch dialogue box will
display the name of your CMYK setup as defined earlier (if you change your
CMYK setup to something else).>>

Yes, it will recognize the profile name, the question is why is the
mismatch box appearing. I've tried several permutations of this and it all
works normally for me (that is, I don't get a mismatch message in PS 6) so
I'd have to suggest looking for something specific to your operation.

Dan Margulis

To unsubscribe from this group, send an email to:
colortheory-unsubscribe@...



Printing Overseas

Dan Margulis <76270.1033@...>
 

John McKercher writes,

Our problem is that on the occasions that we have had stuff printed in
Hong
Kong, we have been disappointed with our output. We will get our
matchprints
looking good on our end, run our film and then ship it for printing. We
explain to the printers that we want them to run heavy on the yellow but it
always seems that our colour is lacking when we get the results. They take
our negs and print from them even though they do most of their printing
from
positives and are not that familiar running film and printing from
negatives. Our images are set with the dot gain settings that our printers
have provided us (12%).>>

12% is ordinarily too low for printing from negs, and usually results in
muddy-looking reproduction. However, with fine paper and good printing
conditions, it is possible that it's right.

Our printers have told us they can provide us with a colour profile for
their presses but they are a Mac shop and we are running PCs so I'm not
sure
that their profiles would be of any use to us.>>

You can use their profiles on a PC (you just have to give them an .icm
suffix), but the problem is you would then have no method to generate a
proof for yourself. Also, if this printer isn't even able to cope with
supplied negs expecting them somehow to have an accurate profile is
somewhat like expecting a fifth-grader to have an accurate understanding of
calculus. And finally, in the unlikely event that the profile is accurate,
it will only be accurate for supplied *positives*.

We have asked our printer to provide printed samples and files for these
samples so that we might compare their work on screen to our own and get a
better sense of how we can improve our colour correction to get what we
want. Is there a more accurate way to do this?>>

This is normally the way to go, except here those printed samples were
probably generated from positive film, hence useless to you if you are
supplying negs. Nor can you use your own files and printed samples, because
they were probably screwing around with them to try to match your proof,
and it doesn't represent their normal printing. What you want them to give
you is proofs from your own images, printed the way they feel comfortable
printing them. If it looks too muddy or whatever, fine, you adjust your own
methods to that.

We would be happy to produce positives for them to print from but there
is
no one here in town that can make matchprints from positive film and we
don't really have the experience with other proofing methods that we could
be sure to trust them.>>

To simplify: you are currently giving the printer something they don't get
good results with. If you give them what they want, you currently have no
ability to proof it. Something has to give.

The brute-force solutions are,
a) find somebody who can make proofs from positive film; this is less
common than it used to be, but it isn't a rare request.
b) pull positives and negatives simultaneously, have proofs made from the
negs and send the pos to the printer.
c) pull positives and have your current supplier contact them into
intermediate negs and make proofs from those. This can be done accurately,
but the proofs will be wrong-reading.

The preferable method is to become more comfortable with digital proofing.
Modern proofers can be configured to match almost any output. This way, you
supply a digital proof with your negs. You are going to need to go through
this education process anyway if you ever go CTP.

Dan Margulis


Re: Photoshop 6 question

Christine Holzmann <tekila@...>
 

>Hi all, I have a CMYK table that I created in PS5/5.5 with the "built-in"
model. I loaded this table (saved as a .api) into my CMYK settings in PS6
and now when I open CMYK files created in PS5 I get a profile-mismatch
although the profile name appears the same. Any ideas why?
Photoshop 5 didn't have a way for you to name the built-in settings. So
what it does is it comes up with its own name, and then embeds that into
your CMYK images. When you load it into Photoshop 6 it gets yet a
different name depending on the filename it was saved as from Photoshop 5.

There are a couple of ways to deal with this depending on your workflow:

1.) Let the profile mismatch happen each time and just dismiss it. (i.e.
ignore problem)

Chris,
Regarding the option you gave below about dismissing the message, then using assign profile to assign the current working space...........
if my CMYK settings in PS 6 are exactly the same as PS 5 (I loaded PS 5 settings into PS 6 after having done a "save as" in PS 5 CMYK settings dialogue box), and I open up a CMYK image in PS 6 that had been saved in PS 5 and get the profile-mismatch setting, can I not simply decide to convert to PS 6 working space and that will be just as good??
(I REALLY hope my question makes sense...:)


Crissi
_____________________________________________________________________________


2.) After opening and dismissing message; go to Image:Mode:Assign Profile
and Assign your current working space (your loaded profile). Now when you
save the image, the working space profile and name will be embedded in
the image. Next time you open it you won't get a profile mismatch.


There is a more automated way to do this involving AppleScripts (to batch
embed the same profile into a bunch of images) but it would be a lengthy
post to explain this in detail.


Chris Murphy

To unsubscribe from this group, send an email to:
colortheory-unsubscribe@egroups.com
--
DESIGN EDITOR @ The Citizen News
http://www.thecitizennews.com

DIGITAL/GRAPHIC ARTIST
http://www.crissi.com


Re: Printing Overseas

Andrew Rodney <andrew@...>
 

on 1/10/01 8:04 PM, John McKercher at john@hartleyandmarks.com wrote:

Our printers have told us they can provide us with a colour profile for
their presses but they are a Mac shop and we are running PCs so I'm not sure
that their profiles would be of any use to us.
No problem, the profiles will work just fine on your PC. Just be sure they
place the .icm extension in the name and you're all set to go.

We have asked our printer to provide printed samples and files for these
samples so that we might compare their work on screen to our own and get a
better sense of how we can improve our colour correction to get what we
want. Is there a more accurate way to do this?
Why don't you simply have them send you the profiles (email), convert some
files from RGB to CMYK using the profiles and see if they would output a
test for you. IF the profiles are good and IF your displays are properly
calibrated, you should get a good soft proof on screen.

Andrew Rodney


Re: Questions

Dr Peter Goldman
 

Greetings Group from "Land That Time Forgot",
 
>What is the difference between finding the d-man/d-min on a negative and
>setting the white and black point on curves?

>Are we seeing any parallax problems with the new digital cameras?

>Do the digital cameras use the NTSC work space because of the video
>connection? The D1 stats list NTSC/Pat video and the fuji S1 Pro lists
>NTSC Video Out.

>I hear a lot about the D!. Has anyone had experience with output from
>the Fuji S1 Pro?  Gary Owens
I have been following this thread with great interest.  Apropos of Gary Owen's
query, does anyone have any experience with the "new" Canon D30 which I have
been eyeing lustily.  Specifically, which RGB color space does it utilize for file download and
are there any particular problems / tricks referable to using this particular Canon body.
 
Best regards from "The Dark Continent",
Peter
Peter L. Goldman, M.D., F.A.C.S. (neurosurgery)
LCDR, MC, USNR (Ret.)
Yachting Photographer
Member of the Yachting Journalists' Association (U.K.)
Tel.     +27-(0)21-554-1110
Fax     +27-(0)21-554-2576
Cell     +27-(0)82-562-6688
e-mail:    photodoc@...; yachting_photographer@...
 


Re: Optimal TAC, dotgain for b&w newsprint

samarsh@...
 

David, I use an 'unscientific' eyeball/info palette method - which
gets me by without any real shocks when the paper comes off the press.

I use a custom action to convert colour originals to to greyscale
mode
snapshots - I then simply select the 'best' looking one from 5 or 6
different conversion methods.

These actions use 30% newsprint dot gain settings in the conversion
from RGB to greyscale - and the Greyscale setup in Photoshop 5.5 is
set to Black Ink.

But these images are still too dark. So I use curves in the action to
reduce the tones for printing.

I set non detail highlights to around 3-5% and shadows to 80-85%. The
mid/quarter/three quarter tones are usually pulled down as well.
Usually the 50% point ends up being around 35%.

This image is also previewed without black point compensation if
using
the ICC CMYK setup. Tones may appear light on screen, but with the
info palette you can see what your true tones are.

I no longer get prints which are too dark, and the paper must love my
'grey' images which are not too black for them to print. Perhaps I am
too conservative with my settings and things could be darker...

Take what you want from this post, I'm sure the forum will spot any
holes in my approach.

Hope this helps,

Stephen.


Re: Optimal TAC, dotgain for b&w newsprint

samarsh@...
 

Firstly, I know that talking about colour is 'off topic' to this
post,
so I will respond to the original message regarding monotone work.

I have a question for Christine or anyone elese who would care to
comment.

--- In colortheory@egroups.com, Christine Holzmann <tekila@m...>
wrote:

TAC = 260
Black ink limit = 80
I prefer to use UCR for newsprint....you get the skeleton black
plate that way with less muddiness.
Christine, I do both mono & colour news separations each week for
local and city papers - I am hoping to pick the lists collective mind
on this subject, and this seems like a good opening...

No paper we deal with in Sydney seems to offer sep tables or ICC
profiles for their material suppliers. When they do give specs, they
are contradictory or are aimed at the traditional separator or drum
scanner operator.

I am the first to admit that I know just enough to be dangerous, but
to me MEDIUM GCR is not 70% replacement?

As far as I know, Photoshop does not deal in numerical values by
default, it has none, light, medium, heavy and maximum GCR levels.

I guess and presume that none is 0%, light is 25%, medium is 50%,
heavy is 75% and maximum is 100% GCR...I am probably way off base
here, and I'm sure that I will be corrected by those who know better.

For our printers, grey balance and registration seem to be a common
problem - not to mention consistent dot gain.

I commonly use these settings:

SWOP Newsprint 30% Dot Gain
GCR separation type
80% Max Black
230% Max Ink (TIC/TAC/etc)
Heavy Black generation
0% UCA

This cures most registration and colour balance problems, as most of
the neutral tones are in the black plate.

I think that MAX GCR and 30-50% UCA *might* be the answer, but this
is
too radical from my experience to trust to a live job.

I also must have the dot gain wrong, as images still need curves
adjustments to lighten the tones throughout the image - otherwise it
just runs too strong.

Also for a 85-100 lpi screen, I have found that you can sometimes go
as low as 100 ppi when the image is S/S (same size).

Some newspapers use JPEG internally in their FTP, so images get
degraded further, whether you like it or not (but considering the
output, this does not really matter).

Can anyone comment on my choices for colour newsprint work, or give
some rationale to why other settings should be used?

Image content varies, and is too numerous to describe. I realise that
there is no magic 'single setting' for all images, but there must be
some rules to follow (I will shortly purchse Dan's book, as it seems
that the answers to many of my questions will be found there).

Thanks,

Stephen Marsh.


Re: Photoshop 6 question

Dan Margulis <76270.1033@...>
 

Andrew Engelhardt writes,

Maybe I'm not understanding you correctly, but if you click on the "Save"
button while in the CMYK setup dialogue (using the built-in option) you can
save your setup and thus give it a name.>>

That's correct. You have now named the setup and the embedded profile will
reflect that name.

When you open the tagged image file, the mismatch dialogue box will
display the name of your CMYK setup as defined earlier (if you change your
CMYK setup to something else).>>

Yes, it will recognize the profile name, the question is why is the
mismatch box appearing. I've tried several permutations of this and it all
works normally for me (that is, I don't get a mismatch message in PS 6) so
I'd have to suggest looking for something specific to your operation.

Dan Margulis


Printing Overseas

John McKercher <john@...>
 

We are book publishers based in North America and we do some of our colour
book printing in Hong Kong. When we are printing here in North America, we
use Matchprints in order to colour proof our work (which we have been happy
with). At the moment we are not using colour profiling software. We get the
colour we want by using our monitors, our experience and the matchprints
that we run that are fairly close to what we are looking for on press. Often
on press, we end up bumping up the yellow in order to get that "punch" in
the colour that we like for our books.

Our problem is that on the occasions that we have had stuff printed in Hong
Kong, we have been disappointed with our output. We will get our matchprints
looking good on our end, run our film and then ship it for printing. We
explain to the printers that we want them to run heavy on the yellow but it
always seems that our colour is lacking when we get the results. They take
our negs and print from them even though they do most of their printing from
positives and are not that familiar running film and printing from
negatives. Our images are set with the dot gain settings that our printers
have provided us (12%).

Our printers have told us they can provide us with a colour profile for
their presses but they are a Mac shop and we are running PCs so I'm not sure
that their profiles would be of any use to us.

We have asked our printer to provide printed samples and files for these
samples so that we might compare their work on screen to our own and get a
better sense of how we can improve our colour correction to get what we
want. Is there a more accurate way to do this?

We would be happy to produce positives for them to print from but there is
no one here in town that can make matchprints from positive film and we
don't really have the experience with other proofing methods that we could
be sure to trust them.

Is this a case where colour profiling software is the answer, given that our
printer has a profile built to his press? Is there any way I can take their
Mac profile and make it work with Photoshop on the PC (we are using
Photoshop 5) without spending a bundle on profiling software? Or do we need
to find another proofing system?

We would like to go to CTP printing but we are very nervous about not being
able to properly proof our colour.

Thanks in advance,
John McKercher
Hartley & Marks Publishers
john@hartleyandmarks.com


Wavelengths to RGB

alanmartin@...
 

I am trying to find a link between the view of scientists
(the physics of light and the biology and psychology of human
perception of color) and the expertise your members have in
color matching.

More specifically, I am hoping to find a way of matching
physical Hue values (ie the frequency or wavelength of a color)
to the rgb values of a typical monitor.

The Problem
In Photoshop's (and other software's) color pickers, hue values
come in a circle from 0 to 359 "degrees" through 0 again.
From this one can create a linear rainbow spectrum image,
say 800 x 100 pixels, red on the left, violet on the right
and no change vertically.
In the physics lab one can also create a rainbow spectrum of
the range of visible colors.
If you scale these two so their colors match at each end, the
colors in the mid range will not match.
This is like an analogy, in hue, of a gamma matching problem
in brightness.

Here's another way to describe this mismatch:
I am 55 years old so I learned "color theory" at art school
long before desktop computers and Photoshop even existed.
We learned that if you arrange the spectrum in a circle,
green is opposite red, etc.
But in image editing software, cyan (180 degrees) is opposite red
(0 or 359 degrees). The first of these seems truer to my perception,
with green offering the strongest contrast in hue to red.
This may just be early conditioning rather than unbiased perception,
but I doubt it because of what I will term "the Op Art experience":
if you paint equal width stripes of two fairly saturated colors,
with equal brightness, they jazz the eyes (appear to jump around)
most when you choose two colors that are opposites in the
traditional theory rather than the current software model
(ie red-green and orange-blue rather than red-cyan and
green-magenta).

It seems that current software uses this model because it fits
in with the technology: monitors use rgb so they space the pure
r,g,b colors equally round the circle.

That's the background problem.
Here is the practical need I am seeking help with:

For some experiments in human color perception (both normal and
various types of color-blind), I want to create a number of images,
with colors defined in 24bit rgb (the human subjects will see them
on a monitor). Later I shall also make printed images, but that
translation from rgb to cmyk will be less difficult - I shall be
aided by reading the messages in this eGroup no doubt.

The images will include some gradual spectra and also some images
with single-color juxtaposed blocks. In both cases I need to
translate from wavelength or frequency numbers to rgb numbers.
That way I can juxtapose colors which are true harmonics of each
other, in the same way that a major chord in music uses frequencies
with a simple mathematical ratio.
I am not fussy as to the form... it could be just a lookup table,
or an rgb image of a spectrum with markings and wavelength numbers
along the bottom, or even (best of all) a tiny program where you
enter a wavelength or frequency number and get your rgb numbers
as an answer.
So far my searches on the web have been fruitless. I found an image,
as described, but it was an 8 bit indexed gif (yuk) with only three
wavelengths marked (too crude).

In technical terms, I'd prefer the spectrum images to be logarithmic
rather than linear. But don't worry if that sounds too complicated;
almost any source would be better than what I have found so far!

Sorry to bother you with a query that's way off your usual subject
matter!

Alan Martin alanmartin@one.net.au


Re: Photoshop 6 question

Darren Bernaerdt <darren@...>
 

Chris,

Maybe I'm not understanding you correctly, but if you click on the "Save"
button while in the CMYK setup dialogue (using the built-in option) you can
save your setup and thus give it a name. When you open the tagged image
file, the mismatch dialogue box will display the name of your CMYK setup as
defined earlier (if you change your CMYK setup to something else).

What PS6 doesn't do is recognize that name as being the same, even though
it is the same CMYK setup file created in PS5.0/5.5. Sounds like a bug to
me...

Darren Bernaerdt


Hi all, I have a CMYK table that I created in PS5/5.5 with the "built-in"
model. I loaded this table (saved as a .api) into my CMYK settings in PS6
and now when I open CMYK files created in PS5 I get a profile-mismatch
although the profile name appears the same. Any ideas why?
Photoshop 5 didn't have a way for you to name the built-in settings. So
what it does is it comes up with its own name, and then embeds that into
your CMYK images. When you load it into Photoshop 6 it gets yet a
different name depending on the filename it was saved as from Photoshop 5.


Re: Photoshop 6 question

Chris Murphy <lists@...>
 

Hi all, I have a CMYK table that I created in PS5/5.5 with the "built-in"
model. I loaded this table (saved as a .api) into my CMYK settings in PS6
and now when I open CMYK files created in PS5 I get a profile-mismatch
although the profile name appears the same. Any ideas why?
Photoshop 5 didn't have a way for you to name the built-in settings. So
what it does is it comes up with its own name, and then embeds that into
your CMYK images. When you load it into Photoshop 6 it gets yet a
different name depending on the filename it was saved as from Photoshop 5.

There are a couple of ways to deal with this depending on your workflow:

1.) Let the profile mismatch happen each time and just dismiss it. (i.e.
ignore problem)

2.) After opening and dismissing message; go to Image:Mode:Assign Profile
and Assign your current working space (your loaded profile). Now when you
save the image, the working space profile and name will be embedded in
the image. Next time you open it you won't get a profile mismatch.


There is a more automated way to do this involving AppleScripts (to batch
embed the same profile into a bunch of images) but it would be a lengthy
post to explain this in detail.


Chris Murphy


Photoshop 6 question

Andrew Engelhardt
 

Hi all, I have a CMYK table that I created in PS5/5.5 with the "built-in" model. I loaded this table (saved as a .api) into my CMYK settings in PS6 and now when I open CMYK files created in PS5 I get a profile-mismatch although the profile name appears the same. Any ideas why?

Thx
Andrew Engelhardt
Digital Prepress & Imaging
Marketing Dept.
London Drugs Ltd.
12831 Horseshoe Pl.
Richmond B.C. Canada
V7A 4X5
(604) 272-7602
mailto: aengelhardt@...



old monitor

Dan Margulis <76270.1033@...>
 

Kenji writes,

My older Mitsubishi Diamond Pro monitor seems as if I can't get it
calibrated anymore. The problem is that there is a magenta/red cast in
the shadows and little bit of artifacting in the shadows. The overall
gamma and brightness doesn't allow me to get anywhere near calibrated
these days.It won't go dark enough...it feels too flat and washed out
in the shadows.>>

So?

If you're interested in quality color, you can't rely on your *perception*
of neutrality, which is affected by too many variables. If you look at a
single image, and the shadows seem to you too red, perhaps they in fact are
too red, or perhaps there is something else in the picture that's causing
you to *perceive* that they're too red. So, you don't trust your
perception, you look at the Info palette and find out whether it's really
neutral.

As for the contrast, obviously you are seeing less of it on this monitor.
But the whole purpose of calibrating a monitor is to be able to predict
what an image will eventually look like. Whether it's going to become a
chrome or print on newsprint, it won't match the contrast you see on your
monitor, no matter how good your monitor is. So, everybody makes a mental
adjustment.

So I have two questions. Is it maybe a problem with the video card, or
is it probably the monitor? And is it worth taking it in to a repair
shop? Can they get it anywhere near to being able to use it for digital
photography needs?>>

Based on your description, it seems usable now. Such problems typically are
with the monitor and not the card. As for repairing monitors, this is
generally a losing proposition unless they are under warranty. Monitors are
like people, as they get older more and more things start to go wrong.

Dan Margulis