Hello Olaf,
Thank you very much for this post. Exactly what I was looking for. For
now, at least, I will continue to use AdobeRGB, but I think that it
would be interesting, if there is a good upcoming space as you mention,
for it to be communicated to software developers, camera manufacturers,
etc. once it is ready. This way it would be much easier for users
outside of Germany to implement it.
Best regards.
Darrian
-----Mensaje original-----
De: eci-en-admin(a)lists.transmedia.de
[mailto:eci-en-admin@lists.transmedia.de] En nombre de Olaf Drümmer
Enviado el: miércoles, 10 de diciembre de 2003 11:35
Para: Mailing List [ECI-EN]
Asunto: [ECI-EN] Re: ECI-RGB / Any advantages?
Hi,
I would like to draw attention to some aspects sometimes overlooked with
regard to this discussion and also to broaden the perspective a little
bit:
- focus on what's essential! eciRGB is definitely only one of several
reasonable options for a working color space RGB profile; so if your
image archive happens to be stored using Adobe RGB or what else, or you
are successful and happy using whatever you are using at the moment,
there typically is no need to switch to eciRGB just for the sake of it.
And: nobody _has to_ use eciRGB (though quite a few successful users are
using it). Stating eciRGB is a good option should not at all be
understood as stating everything else is rubbish.
- what's the original? This is one very essential aspect often
forgotten.
As I see it, the original is not what's out in the real world and
getting
photographed, nor is it what even the best camera captures. It is what
the photographer or the person doing retouching/color correction etc. of
the raw image wants it to be. Thus it becomes an issue not only what
colors occur in the real world, but also what colors can be printed.
- do we really always know where an image is going to be printed? Please
do not forget that images may be reused, the initial use being maybe a
display ad for a magazine general, then a similar campaign in newsprint,
then some digitally printed direct mailing pieces, then a huge poster at
some trade show, and so forth. eciRGB at least for many users in Europe
has proven to be a very suitable general purpose working color space.
Again: is it the only reasonable one? No. Does it have serious
deficiencies that some other profile does not have? Not that I know of.
- color science does have a decent understanding of what real world
colors are. Some admittedly do not occur that often but they may occur.
- Ugra in Switzerland has done a research project for Ifra (I am not
sure when the report is going to be available, probably first quarter
next year, they presented the preliminary results at the Ifra Color
Management Working Group during Ifra Expo October 2004 in Leipzig): Ugra
compared quite a few common RGB working color spaces and - using quite
an
impressive range of metrics - arrived at the conclusion that all in all
eciRGB comes in first, with a couple of other profiles coming in next
very close.
- one reason to not necessarily stay in a scanner profile color space
(unless you leave it un(re)touched) is that such profiles are not very
linear - thus color corrections, retouching etc. can result in
unnecessary loss of detail.
- there are always trade-offs in every area of our business: e.g. detail
vs. gamut (especially as long as 8 bit data is used). If maximum detail
is key and there is no need to be prepared for all possible real world/
pritable colors, some profile other than eciRGB may have advantages.
- 8bit Lab does have its limitations: less than a third of the possible
Lab value triples exist as real world color, i.e. one is only making
sensible use of roughly 5 millions out of 16 millions possible color
value triples. This _is_ a threat to keeping as much detail as possible.
- Jo is right in still asking for a summary of why eciRGB should be
considered and what's relevant to know about it. Here is a summary that
I
have compiled for upload to the ECI website (going to be available there
shortly):
What is eciRGB 1.0, and why should I care?
------------------------------------------
eciRGB 1.0 is recommended by the European Color Initiative (ECI) for use
as an RGB working color space and for color data exchange for ad
agencies, publishers, reproduction and printing houses.
There are a couple of aspects that led to the development and release of
eciRGB by the ECI in 1999:
- Up to Adobe Photoshop 5.5 the default for the working color space was
the monitor profile of the computer on which Photoshop was running. This
turned out to be a less than perfect approach, as every computer may
have
a different monitor profile, and a working color space should not depend
on what happens to be the monitor profile on a given computer, but
should
always be the same, at least within any given workflow.
- Furthermore a monitor profile typically does not cover certain ranges
of colors that can easily be produced on a printing press - the major
area of weakness of a monitor compared with a press is in the Cyan color
region.
- Staying in the scanner profile's color space often makes no sense, as
a
scanner's color space typically is not very uniform (a characteristic
offered for example by the Lab color space) this makes them unsuitable
for any corrections or editing.
- Lab in principle looked like a good alternative candidate, but hardly
any widely used software (until very recently) supported editing and
retouching for 16 bit Lab data or import of 16 bit Lab images (and while
PostScript/EPS supports up to 12 bit Lab data, in PDF 16 bit image data
have only been introduced very recently in version 1.5, earlier versions
only accommodating a maximum of 8 bit), and 8 bit Lab data was far too
inefficient, as only less than a third of all possible Lab color values
actually occur in real world color data: which means that out of the 8
bit only a bit more than 6 bit would actually be used, which simply
isn't
enough. Even with 16 bit Lab the issue arises that it uses twice the
amount of data - while the increase in file size often will be moderate
due to compression, the amount of RAM needed definitely doubles.
- while there were and still are other reasonable offerings around in
terms of RGB working space ICC profiles ECI wanted to see one that
... has a gamut that covers all colors that can be printed on today's
printing presses - whether sheet fed or web offset, gravure or newsprint
- but not much beyond (in order to not to waste precision for bits that
never really get used)
... produces a neutral gray whenever the values for Red, Green and Blue
are equal
... approximates uniform distribution of color values, i.e. equal
difference between two color values in eciRGB mirrors an perceived equal
difference when these colors are seen by the human eye
... is based on a Gamma of 1.8 and a light source of 5000K.
This profile was created and extensively tested in 1998/1999 and
provided
to the public free of charge in 1999. Most ECI members have meanwhile
based their internal workflows on eciRGB as the preferred (and often the
only) RGB working color space, and have been more than satisfied with
the
advantages achieved.
Is eciRGB the only working color space that makes sense?
Definitely no. Other experts and user groups in the industry as well as
vendors have been working on this issue and have come up with
alternative
options that may work as well as eciRGB. If you happen to have been
working with let's say Adobe RGB or ColorMatch RGB until now, it would
not be a good idea to convert your image database to eciRGB just for the
sake of then having eciRGB data. Nevertheless, when you are about to set
up a new print oriented workflow, the ECI is confident that there are
hardly better options than eciRGB (though possibly quite a few that are
comparably good).
How does eciRGB compare to sRGB?
sRGB has seen excessively wide use, as especially HP and Microsoft were
promoting it as the standard RGB profile (in the end often going so far
that they - whether directly or indirectly - propose not to use any ICC
profiles anymore but instead to simply implicitly store and exchange any
RGB data as sRGB. More and more digital cameras, low end inkjet printers
and even monitors now simply assume sRGB. While this may look like a
smart move, it must not be forgotten, that sRGB as a color space has
serious weaknesses - there are a lot of colors today's printing presses
as well as other output devices like photo printers, large format
printers and many inkjets (not to speak of up to date monitors or
digital
cameras) can produce that cannot be stored in sRGB. If quality is
important, sRGB is not an option.
Should I get rid of my scanner profiles right away and use eciRGB
instead?
No, definitely not. Simply assigning eciRGB instead of the specific
profile e.g. for a scanner is one of the worst things you can do. Always
use the correct source profile for the data at hand, and then convert
from that color space into eciRGB.
Note 1: What's a working color space anyway? The short answer is, that
such a color space is good for working on an image (or rather any color
object), whether color correcting or retouching it or editing it in any
other way. A working color space can be made such that it is independent
of idiosyncrasies of a specific device (e.g. it can be made uniform to
preserve detail equally well in each area of the color space). At the
same time it can be made such that it serves a class of color spaces
well, e.g. that it comprises all or most of the colors that can be
output
with all monitor and printing devices that are available today.
Note 2: In ISO there is currently work underway - as ISO standard
"ISO/CD
22028-2 Photography and graphic technology - Extended colour encodings
for digital image storage, manipulation and interchange - Part 2:
Reference Output Medium Metric RGB colour image encoding (ROMM RGB)" -
to
possibly arrive at a working color space that is even better than eciRGB
or other widely used working color spaces.
Sorry for the lengthy post, folks!
Olaf Druemmer
Vice chairman of the European Color Initiative
_______________________________________________
ECI-EN mailing list
ECI-EN(a)lists.transmedia.de
http://lists.transmedia.de/mailman/listinfo/eci-en