UCR vs GCR...or Neither - RAW CMYK?


samarsh@...
 

--- In colortheory@y..., Dan Margulis <76270.1033@c...> wrote:

No, this is traditional UCR, black only appearing where
necessary to
achieve darkness, not to hold neutrality.
For anyone interested - I put together a quick example page.

http://members.ozemail.com.au/~samarsh/ucr_variants.html

This shows the specific separation controls for the Scitex
EverSmart Supreme oXYgen scanning software - including the
dubiously named 'OFF', UCR and GCR options.

Note the alien grey ramp for the OFF setting - which is what we
generally use for inhouse work and trade scans.

In Dans book, there are some images which show the effects of
GCR and total ink limits - which restrict the detail in the
shadows.

I have also included a quick screen shot showing the difference
in the magenta colour sep for a shadow region.

There are also links to low res TIF files of both Photoshops UCR
sep and the CreoScitex sep.

Even though the detail may not show on press, even with our
CTP setup - the OFF setting in the CreoScitex software
produces much more shadow detail in all separations.

As time goes on, I will do more tests - on coloured areas as well
as shadows and neutrals.

This is just to help visually explain my original confusing thread
on the 'off' setting.

Stephen Marsh.


Dan Margulis <76270.1033@...>
 

Dave Badger writes,

I have been using Photoshop's Medium Generation for a long time now
because
GCR is supposed to give you better saturation and depth in the darker
colors. I believe this to be true and think that, for example, cyan in reds
being replaced partially by black is a good thing.>>

I would think the opposite would be true. In principle the two results
should be the same. In practice, black is much stronger than cyan is, so
any variance in inking would create a color issue, especially since the
dominant magenta and yellow in the reds will have been reduced to
accommodate the black.

Using this setting means the black is more critical on press, so I would
think using UCR or Light GCR would be better for newsprint since their
black
plate would be less controlled or more dot gain.>>

Certainly inking generally is less in control in newspaper printing.
However, 1) the black ink used on newsprint isn't as powerful as in
commercial printing, therefore the impact of a density change isn't as
severe; 2) in commercial printing, if the black runs too heavily one of the
principal suspects always is that the pressman was attempting to compensate
for small type or type that is difficult to print because of thinness in
part of the strokes, such as Bodoni, Baskerville, or to a lesser extent,
Times Roman. Newspapers generally aren't subject to this effect because
their text face is invariably something like Excelsior or Ionic or Times
Europa that is specifically designed for legibility at the expense of
aesthetics.

So, while I'd agree with the general policy of using a skeleton black in
newspapers, I think there's more of a case for using a heavier black than
there is in other forms of offset.

This statement seems to indicate that people who use Photoshop's Medium
GCR setting are unsophisticated users. Yet you've said one form of UCR/GCR
is not necessarily better then another.>>

Those experienced in preparing images for CMYK, especially if the target is
a press, by and large certainly favor a lighter black than the Photoshop
default. One form of black generation isn't necessarily better, except in
an image-by-image context, and except in the context of how variable one
expectst the output device to be. A lot of the reasons for preferring a
lighter black vanish if the output is going to be inkjet rather than press.

Can you further explain why you favor a skeleton black and are there any
other downsides to GCR then those you listed above?>>

In addition to the issues mentioned earlier, the services of our friends in
the color management community become more important if you're using a
heavier black. Photoshop makes the wrong assumption that black has the same
dot gain as the other inks, whereas in real life it's usually higher. But
it's very much case by case, so a lot of users have incorrect black dot
gain values unless they've really gone to a lot of effort.

If you are using a skeleton black and your black dot gain value is too low,
this isn't great but it isn't a tragedy. If you're using a heavier black
that affects colored areas, you're really risking a muddy sep.

Dan Margulis


Dave Badger <dbadge@...>
 

on 10/6/01 3:00 PM, Dan Margulis wrote:

Dot gain and black generation are different animals. If you believe that an
image will correct or print better with a higher black component, that
decision would presumably the same on either 100# Kromekote or newsprint.
I have been using Photoshop's Medium Generation for a long time now because
GCR is supposed to give you better saturation and depth in the darker
colors. I believe this to be true and think that, for example, cyan in reds
being replaced partially by black is a good thing. Yet no black substitution
take place in lighter skin tones, so the start point of this setting seems
fine.

Using this setting means the black is more critical on press, so I would
think using UCR or Light GCR would be better for newsprint since their black
plate would be less controlled or more dot gain.

Printers who insist on one method or another generally are completely out
of touch. Right now, their clients can be divided into those who know what
they're doing, in which case they are probably using a skeleton black, and
those who don't, who are probably using the Photoshop default of a heavier
black. Yet from time to time both groups get good results.
This statement seems to indicate that people who use Photoshop's Medium GCR
setting are unsophisticated users. Yet you've said one form of UCR/GCR is
not necessarily better then another.

The same pluses and minuses apply: with a heavier black,
one risks muddier images, and they're harder to correct in Photoshop. OTOH
a heavier black will hold neutrality better, and also make it easier to
print the job in register, or rather make a misregistered job look more
acceptable.
Can you further explain why you favor a skeleton black and are there any
other downsides to GCR then those you listed above?


Dave Badger


Dan Margulis <76270.1033@...>
 

Stephen writes,

I use one of those expensive Scitex scanners you mention - and I would
have to agree with your take on their default tables black generation. I
have not been able to make a custom black generation curve in Photoshop
that even comes close.>>

The question is why you would want to. The whole point of variant black
generations is that theoretically the colors remain the same--in principle
Heavy GCR gives the same result in print as UCR. In practice it doesn't,
but the difference between Photoshop's Light GCR and Scitex's is going to
be very small in the overall scheme of things.

In part this is due to how hard the black generation curve is to edit -
and that Photoshop _only_ gives the option of UCR or GCR separation
methods.>>

*All* separations, saving those that are CMY only, are UCR or GCR.

The Scitex scanner in default CMYK tables settings scans into
CMYK with GCR and UCR turned OFF. You get 'raw' CMYK.>>

There isn't any such animal. The scanner records R,G, and B data which
becomes, with mild variation, C, M, and Y. The black data is calculated
from the other three--there are no CMYK-native scanners.

It seems that CMY are used throughout the image to describe neutral
tones. I guess this is similar to using GCR with no black generation.>>

No, this is traditional UCR, black only appearing where necessary to
achieve darkness, not to hold neutrality.

Now I use GCR light black or UCR most of the time, unless more black can
help (when doing Photoshop seps). This is very much an individual image and
output based decision - gone are the days when one setting suited all (now
that I know better). I have not reached this level on the scanner yet.>>

Your policy is a good one, but I doubt that it's worth worrying about
extending it to the scanner. When you run across something that needs a
nonstandard black, do it in Photoshop-it's a lot easier. Four years ago, in
reviewing the new Scitex line, I wrote:

"Basic scanning may be easy, but the exceptions are lethal. Scitex's has
several sets of preset CMYK parameters for different printing
conditions--but, unbelievably, none for the most common of all, SWOP. Users
can create these parameters for themselves--if they happen to be one of the
twenty or so individuals on the planet capable of figuring out Scitex's
exceedingly opaque implementation of black generation."

Dot gain is never explicitly mentioned in the scanner software.>>
Dot gain and black generation are different animals. If you believe that an
image will correct or print better with a higher black component, that
decision would presumably the same on either 100# Kromekote or newsprint.

But from what you are saying, knowing the amount of UCR or
GCR is pointless without knowing the black generation method,
as in the start point for black?>>

In order to conceptualize the kind of black that's being generated one
needs to know the start point of the black and the how the slope of the
curve varies. A percentage is only going to be relevant at one point on the
curve. In darker areas, the percentage of black will be higher and in
lighter areas it will be lower.

I thought that the use of the term 50% GCR was an old drum scanner or
traditional separation terminology.>>

It's sort of like the percentages used in the "Amount" field of the Unsharp
Mask filter or the dot gain percentage in CMYK setup. The user doesn't need
to know what the percentages mean, just that a higher percentage increases
the effect. Unlike these two, however, which *do* have meanings that an
interested user could find out, "50% GCR" is meaningless without further
explanation, except that it probably generates less black than 60% GCR.

Many printers still list this in their separation spec sheets - instead
of the more 'regular' or 'standard' Photoshop terminology. But just like
SWOP, they do not mention a start point
- only the amount of UCR or GCR.>>

Sure, and many others express their column width in pica ems, or indicate
that they have 2% dot gain.

Dan Margulis


samarsh@...
 

Dan writes:

I suspect that the answer probably depends on how many
printers use
expensive scanners and scan directly into CMYK. If so, they're
almost
certainly going with the manufacturer's default, which is
ordinarily a
light black, UCR style, but Scitex products will produce a darker
black
than Screen products which in turn produce a darker black
than anything
from Heidelberg/Linotype/Hell. If they're using Photoshop to
make the seps,
odds are that they're going with a heavier GCR, because most
printers don't
know much about the topic--the knowledge, if anywhere, is in
prepress, and
it's fairly rare there.
Dan, this is a good opening for bringing up a subject which has
been bugging me for a while.

I use one of those expensive Scitex scanners you mention - and I
would have to agree with your take on their default tables black
generation. I have not been able to make a custom black
generation curve in Photoshop that even comes close.

In part this is due to how hard the black generation curve is to
edit - and that Photoshop _only_ gives the option of UCR or GCR
separation methods.

To mimick the defualt scanner sep, I would need to make two
Photoshop conversions, one using CMY only and another to
make a custom black plate to plug into the CMY file!

The Scitex scanner in default CMYK tables settings scans into
CMYK with GCR and UCR turned OFF. You get 'raw' CMYK.

By evaluating the separation setup curves in the scanner for the
off setting - one can get a basic understanding of the process.

It seems that CMY are used throughout the image to describe
neutral tones. I guess this is similar to using GCR with no black
generation. This is the closest description to the CMY behaviour
that I can come up with, and is not a good example.

Black curve seems to start at around 40% and seems irrelevant
to the 'linear' neutral CMY curve (where cyan is run higher).

Shadow aimpoints are very high in all four colours, often
delivering TAC of 360%.

This is out of my 'desktop' experience and must be a throwback
to more traditional scanning or separation. This 'raw' CMYK is
very different to Photoshops UCR or GCR separations.

Basically - there seems to be no way to make the built in settings
of Photoshop behave in a similar way - since UCR or GCR is
always in use for a separation.

Now back to UCR or GCR.

The Scitex software does use the % term for GCR. There is no
mention of none, light, medium, heavy or maximum black
generation - as in Photoshop.

As a side note - my 'antique' copy of Macromedia xRes had a
similar approach.

The conversion from RGB to CMYK was 'raw'. After the file was in
CMYK, then the UCR/GCR command was run...you could even
apply this command to a CMYK file created in other software.

A dialog box was presented with the choice of UCR or GCR and
their related controls, including start point for the grey
component, black limit, TAC and percent of UCR or GCR (with
UCA as an option as well).

All this was applied as a post separation move - instead of
during separation...which seems weird to my Photoshop
experience, but it may have some uses. I have not kicked these
options around much - since xRes gathers actual and virtual
dust.

As for your comment on the knowledge being is pre press, and
that it is usually lacking - I have to agree.

Before reading your works, I used GCR med black generation as
a general rule - but did experiment with light black generation.

Now I use GCR light black or UCR most of the time, unless
more black can help (when doing Photoshop seps). This is very
much an individual image and output based decision - gone are
the days when one setting suited all (now that I know better). I
have not reached this level on the scanner yet.

But most users that I have seen use GCR Med and the other
defaults in Photoshops built in settings as a standard - if they
are not using a profile such as SWOP v2 or US Flatsheet v2 etc.

Rather than just use the Photoshop terminology, here's
what they have to say:

"Current recommendations suggest that a safe range of GCR
to use is between
30% and 60%. A 50% GCR setting removes 50% of the gray
component normally
produced by the chromatic color and compensates by adding
an equivalent
amount of black."

This naive description is roughly equivalent to me telling you
that I would
like to meet you at 2 p.m. today on the corner of First and Main,
without
telling you in what city. Any sane method of GCR will use a
relatively low
percentage of black in light greys and a relatively high one in
darker
greys. The SWOP definition is meaningless, worthless. And if
SWOP doesn't
know what it's talking about, how can we expect the individual
web printers
to?
Well, if using xRes then I guess I would enter 50% as the
percentage of UCR or GCR - with a start point of around 25%. It
seems that xRes does work this way - there are no light,
medium, heavy or max options.

The same for my Scitex separation method - if GCR is chosen
then you have no choice but to enter a percentage value for UCR
or GCR, and a black start point %, black limits, TAC etc.

Dot gain is never explicityly mentioned in the scanner software.

But since xRes is history, then the question is a lot harder for
Photoshop - which does not use these terms. Instead we have
UCR, or GCR - none (cmy) light, med, heavy and max black
generation.

But from what you are saying, knowing the amount of UCR or
GCR is pointless without knowing the black generation method,
as in the start point for black?

I take it that the black start point in more traditional separation
software is what Photoshops named black generation settings
do, or perhaps the custom black generation curve option?

I thought that the use of the term 50% GCR was an old drum
scanner or traditional separation terminology. Since I entered
pre press through typesetting - I just presumed that this was a
gap in my knowledge, that would have been covered if I originally
trained in repro instead of type.

Many printers still list this in their separation spec sheets -
instead of the more 'regular' or 'standard' Photoshop
terminology. But just like SWOP, they do not mention a start point
- only the amount of UCR or GCR.

As for the original post - I work for an Australian commercial
printer and we will take pretty much any separation you can throw
at us (for good coated stock). We use CTP and only have one
four colour flatsheet litho press - so things are pretty tight in
process control. Before working here, I would have considered
the seps we produce mud - but everyone seems happy...

I am glad that I am not the only one having problems
understanding all this...I thought it was just me. <g>

Regards,

Stephen Marsh.