Re: Image Print
Re: Image Print
- Subject: Re: Image Print
- From: neil snape <email@hidden>
- Date: Tue, 17 Sep 2002 09:20:47 +0200
on 17/09/2002 04:48, DuWayne at email@hidden wrote:
>
>
RGB/CMYK it doesn't matter. Some one had asked why reference files are
>
CMYK/RGB and not lab - well the reason is that we are building lookup tables
>
and what the lookup table is doesn't matter as long as you do conversions
>
between the color spaces correctly. If I have a CMYK color and I know the
>
source space, the source space profile will give me the lab value. If I have
>
a lab value, the RGB output profile will know what color to send to the rip
>
engine for conversion to plane data. The opposite also works. ICC profiles
>
don't have any "magically" control of the printer.
This how color worlds are built by the calculation through the CMM. As with
all calculated data with interpolation and interpretation there are and will
be some errors dependant on precision.
>
No ones software has any "black magic" in it. All the pieces(postscript,
>
icc, printer encoding) are very well defined for any one to look at. The
>
big trick is connecting the pieces becuase all devices/technology add a lot
>
more variables & errors than anyone every thought of. It just comes down to
>
who can write a more complex formulas to deal with more variables, trap
>
errors and not create other problems. Look a simple gray balance -
>
photolabs/printing presses have been doing it correctly for years its not
>
difficult. Yet try 6-8 non-linear pigmented metamerizing inks with variable
>
volume & density droplets hitting coated paper - Now either all the
>
developers went stupid on us or its really difficult to control.
The OEM manufacturers also have a lot of responsibility for and with this.
Now that Epson have come up with a combo that eliminates many of the past
problems they give the hardware a chance to get the results closer to what
we'd expect.
Black magic can also be so called "secrets" that can't be dispelled. I
always find it makes my day when someone like Bruce Lindbloom , Henrik
Holmegaard , Phil Green, Chris Cox, Thomas Knoll, Robin Myers, etc dispel
some secrets!
>
I try to stay away from saying or doing "linearisations", becuase it is the
>
stupidest thing you can do to a printer. Forcing a printer to print a
>
certin dot area say 50% cyan when you send it 50% cyan in a file is dumb.
>
First your basing your measurement on an arbitray "n" value to make the
>
calculation. Second it makes an assumption that cyan ink in the printer is
>
the same color as your therotical cyan in your given source space and that
>
the halftones are also close to your therotical 50% cyan. We are not
>
matching dot area we are trying to match color. If you want halftones dots
>
to match and don't care about match color from device to device- then use
>
dot area calculations - forget icc an go back to plain old lookup
>
tables/transfer curves.
Hmmm. Linearisation puts the printer in a balanced repeatable fashion, a
known state if you like. Linearsation is another table as are the tables in
an ICC profile as they are in a CRD. Some people who are stuck on
densiometric values from a press can correlate the numbers onto an inkjet
but it's just to keep the terms familiar. Gamut by proper linearisation is
increased not decreased. When too much ink is laid down at a certain point
the Chroma values reverse. Trick is to maximise the gamut by linearisation,
something that the Epson drivers didn't have. I hear that the Epson drivers
don't lay down enough ink, which is also a problem with certain other
printers I'm told. This is very important and probably why ImagePrint will
fetch a vastly superior print than the Epson driver. Other rip makers will
or should do the same.
>
The reason linearisation doesn't work is becuase if the printer needs to
>
print 70% cyan ink to match the color of 50% therotical cyan in your given
>
source space you have killed a lot of gamut. Yes you have to get ink on
>
paper reasonably clean(not soggy), but simple hard linearisations is not the
>
answer it will do more harm. It takes a complex "print recipe" that takes
>
into account more factors to print good on a given media.
See above, visual linearisation is possible (hard lin?) but a measurement
based table has nothing to do with color repro just pure gamma step
calculation. It doesn't effect gamut extremities, rather brings the scale
into a range onto which the ICC profile will better distribution of the
grid point location.
>
> This would be firmware on Epsons mixing the Ultachrome blacks I'm
>
guessing.
>
The firmware controls the mechanics of the printer how to fire dots - how to
>
move etc...not what you think.
What does the actual separation of the LC, LM (Blacks in the case of the
Epson Ultrachromes) then?
>
CMM will never go to using spectral information, becuase they don't have to
>
and there is no benfit to do so. Spectral data can be used to build
>
profiles but it has taken to long to get the ICC spec to where its today,
>
there is no way that they will make changes like that. Most of the profiling
>
software on the market can't compute an icc profile correctly from a single
>
lab measurement per patch with out errors and flaw. Do you really think that
>
they will be handle 20-30 spectral mesurements per patch (50,000 + pieces of
>
data) and compute a reasonable profile especially they way inkjet inks
>
bronze and metamerisize?
ICC CMM's are not designed for this as far as I know. It's true that the ICC
architecture has superseded it's original goals by far. It's been with the
help of very crafty color scientists that the great results we have today,
sparing with an old framework, in the need of improvement. These very people
are so good at dealing with these issues it'll be a pleasant surprise to see
just what they can do in the future!
>
As for a recent developer, Colorbyte has been around for years writing the
>
software for a lot of the hardware in the market. Umax scanners, Howtek
>
scanners, Iris printers.
Yes I know. Rather reluctant to adopt open ICC standards though. Will it
continue as before?
>
Everyone at Colorbyte is pretty cool about looking a samples, responding to
>
emails, looking at problems and they have even visited a bunches of
>
customers throughout the country. They are a whole lot better than a lot of
>
manufacturers I have dealt with, they seem to care about there product.
>
Just make sure you know what your doing before taking up their time.
That's good to hear. We need more third party suppliers to take heed. Too
bad for anyone that doesn't meet an expectation of knowing what they're
doing.
Neil Snape email@hidden
http://mapage.noos.fr/nsnape
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.