Re: Color management

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

On Thu, December 1, 2011 19:36, Karl Shah-Jenner wrote:
> David Dyer-Bennet read my excessively long post and commented:
>>Okay, the alarm bells are starting to ring here.  Epson doesn't sell
> profiles to photographers; they give profiles for Epson media and standard
> Epson printers.  If you want a profile unique to your own printer, or for
> a combination of ink, printer, and paper that crosses company boundaries,
> you can either make it yourself, or use one of a number of commercial
> outfits that make custom profiles.  But Epson isn't selling profiles
> anywhere I've ever heard of.  And thus there is no benefit to Epson from
> making this "hard"; what Epson benefits from is satisfied customers.
> Here in Oz they did sell profiles, they had a seperate firm develop them,
> but then they were onsold, I made the mistake of presuming it was the case
> the world over. .

Ah, of course; I should have thought of that as a possibility (and thus
had less "alarm bell" and more "oh, different").

>>While for extreme enlargements you can probably do better in your own
> software than in the printer, I find sending original-size images to my
> printer for ordinary print sizes to be the best practice.  There are very
> modest gains in sharpness from excess data, and skipping the resizing step
> in the workflow saves me from a range of possible mistakes (saving the
> wrong thing, possibly the small version over my master version).
> here again you're right to a point, the printers rip will be better often
> than a cheap, bought RIP unless it's for a specific non-photographic
> purpose
> like an architects RIP.  My point was as I alluded, sending excessively
> large images to a printer will just upset it.  I also still feel it's good
> practice to store printable images at the 'optimum' pixel count for given
> sizes

In theory, I certainly believe in optimizing to the point of insanity and
then storing that version for future use.  For the ultimate-quality
prints; mostly even my hardcopy isn't pushing for ultimate quality; either
I'm lazy, or else I've learned too much about printing (so that "ultimate"
is getting to be quite a lot of work).

> I'll confess I have Irfanview batch rip all my printables straight to
> folders labeled 8x10's, 6x4's, largescreen and email within subfolders.
> THis is done in a DOS batch process with one click so my brain disengages
> ..

It's tremendously useful to be able to disengage the brain from things
that don't actually need interactive intelligence.

>>And I don't even know what you mean by saying "I tell the Epson I'm using
> Ilford paper".  The only meaning I can see is that you're using an Ilford
> profile?  But that should be in Photoshop, not in the printer (at least
> that's the preferred workflow for everybody I've talked to about this).
> Whatever software package is used, the printer dialogue box will still
> present it's self and that's where people will select the profile (paper
> type).

Perhaps a lot of people will, but it's not what I've been taught is the
recommended workflow; I do the profile processing in Photoshop and turn
off the adjustments in the printer.

Anyway, I think we've established that we are indeed talking about output
profile selection, which was what I wasn't sure of.

>>> the way it looks by having a monitor profile 'adjust' the image to look
>> good, you have another calibration profile adjusting the printer.. now,
>> should one go wrong - which is it?  Could it be they're all adjusting
>> each
>> other to give you what you had in the first place? (and that's not even
>> going anywhere near the 'hidden' profiling done by the OS, browsers and
>> graphics cards)
>>Not if you're doing it right.  The opposite, in fact -- with proper
> profiling, what's going on is each device is coming as close as it can to
> properly representing the colors recorded in a file.  If you start nudging
> each one individually, *that's* when you end up chasing your tail.
> Might be just what I've seen, but the first step in any courses I've been
> on
> has seen the instructor telling the students we begin by profiling the
> monitor.  Say we follow this step, then we go to the printer.. Straight
> up,
> they've utterly ignored the OS and the graphics card's intervention -
> utterly.  No mention at all.  Next thing is people start working on the
> images based on what they see.. and when it doesn't print quite right the
> fiddling starts.

The monitor profile controls the OS and the graphics card's intervention.

> The graphics folks do every step according to densitometers and colour
> analysers, so they don't even need to see the images at any point, it's
> all
> colourmetrics to them - as soon as us photographers start making
> adjustments
> at any point along the path we're fouling up their carefully designed
> path.

Not in a properly color-managed workflow.  A good printing plant will
produce the "right" colors for the numbers in the file, just as a properly
profiled monitor/graphics card will display as close as it can to the
"right" color for the numbers in the file.

> Now sure, if you only fiddle one aspect, one variable in your path then
> you'd be minimising the chance of errors as you say, but..  Let's say at
> the
> monitor calibration point you have the OS saturating the colours up.. so
> you
> tweak the monitor to tone them down according to your Spider/Eye-One.
> That's
> two variables changed.  The Video card is also smoothing and dropping
> colour
> data, so now we're at 3 variables.

Nope; you set the monitor characteristics to neutral first, then the
profile controls the rest.

> At this stage though the image is unaltered, but we have 3 'profiles' at
> work already.  Now we go look at the image and start changing it to look
> 'better'.  We send it to print and it's not quite what we expected - what
> do
> we change?  We've already done the workflow profiling and it should be
> bang
> on.. so we start changing settings in the printer - now we're making 2
> changes based on 4 different system profiles, two of which we have access
> to, 2 we do not.

I'd never dream of changing the printer parameters at that point.  If
absolutely necessary, and it is sometimes, I'd add adjustment layers in
the image, named as applying to a particular printer/paper combination.

> I've still had no answer from anyone whether Quartz (in Apples) colour
> space
> and management can be turned off ..

I don't touch Apple stuff myself.

> My method is just starting with the printer, calibrating that and then
> having the monitor 'best represent' what the printer produces.  I honestly
> couldn't think of a better way to do it as a photographer.
>>Your approach also makes your files unique to you, not usable by anybody
> else.  Because they encode all the adjustments you've made to your
> personal environment.
> Here I apologise as I laugh and contemplate the discussions I've heard at
> pro printers and photo kiosks as arguments break out over profiles, colour
> spaces and flared tempers go head to head over who's right and who's wrong
> about image files.

I've printed the same files on two of Ctein's Epson printers (a 9800, and
I think the little one for the first proof was a 2400), my R800, and
Costco's 7800 (using Ilford rather than Epson paper, but with a custom
profile for that) and gotten beautiful matching.

A lot of people mess up to the point of applying profiles twice and things
like that, but if you do it right, it works much better than anything

> My image alterations are minimal (most of the time there's no alterations
> done from camera to print, since the printer is 'profiled' at day 1) and
> sure, I cannot hand an image to a graphic dude without them having to do
> some work, but then that's the way it always was, the relationship between
> graphics companies using photographs in publications - they did that side,
> I
> did the picture taking side.  If they have a clue then it's a
> straightforward process for them to make the image print ready and look as
> good as it can..

Ah, that may also be important.  A photo being prepared for print gets the
full exhibition-quality printing treatment, meaning everything I know how
to do.  This rarely ends up being less than 2 image layers and 4
adjustment layers, usually with layer masks.  Half an hour to (obscenely
large numbers here).

> Me, I don't want to know about 4 colour print process - I'll always hate
> the
> results, just as any E6 shooter who saw their lovely slide projected with
> all that lovely gamma was dissapointed when they saw their darker tones
> descent to black.  Projected  V reflected is never what we want.

I don't think I've had anything of mine reproduced in color offset -- back
when I did yearbook and magazine stuff we were B&W.  But even photographic
prints from slides weren't very good, unless you went to exotic processes
AND expert printers.  Basically, if you wanted prints you were much better
off shooting negatives (as wedding photographers did for example).  I know
the printing industry was mostly too ignorant to work from negs
effectively, so magazine work and such was always slides, at considerable
cost to the results.

> Again though, loading other folks tagged images into my PC   I can happily
> show them the *same image* loaded into 4 different programs all looking
> utterly different.  version 1 applies a colour space alteration, version 2
> applies a colour space + profile, version 3 ignores colour space and
> profile, version 4 applies a profile and ignores colour space.

Most of them are wrong; they're bad viewers.  I can look at images in
Photoshop and Irfanview and Firefox and they all look the same, which is
as it should be.

> but back to printing - the end goal of all photos -  the graphics dudes
> prints will look wrong to me as he has his colour management set for a 4
> colour printer.  and that's the gist of colour management for
> all-platform-total-100%-portability, restricting the colour space to the
> lowest common denominator to ensure ALL output looks the same!  If you do
> it
> the 'right' way this is what you're aiming for

No, that's not how it works at all.  However, if low-gamut processes are a
part of your plan, you do have to consider them, and decide if you want to
make separate versions for that, or just accept limitations overall.

I ignore them, since my uses are photo-quality prints plus online.  Offset
printing isn't relevant to me any more.

> I do not want that, I want the full gamut of the 8 or more ink set printer
> available to me.  So by not employing a software profile I have no tags
> clipping the image colour range and my images will preserve the full gamut
> of whatever the camera was capable of producing.

You've got a large basic misunderstanding of color management going here. 
Also possibly a misunderstanding of where the real choke-points in color
space are.

Here's how my workflow goes:

RAW file is converted in ACR or Bibble Pro into Prophoto RGB color space
16 bit channels.  That's a huge space, but with 16-bit channels doesn't
compress things particularly.

That is then manipulated in Photoshop or Bibble Pro until I realize I'm
making it worse instead of better :-).

If I'm printing locally, I print from Photoshop (I basically use Bibble to
prepare things in bulk for the Web) directly, using an appropriate printer

If I'm sending to a lab to print, I follow their instructions, which
nearly always are to convert to sRGB color space and jpeg level 10.  This
is a smaller color space than Prophoto RGB -- but big enough for the final
print.  The big space is needed while working the image, not so much for
the final result.

> THis method is not as strange as it initially seems, it's 'working
> backwards' against the graphics method, but don't forget, graphics
> workflows
> are designed to start at the monitor stage as they input values to create
> their colours.

Modern color-managed workflows are intended to handle everything, but
photos are their primary actual use, and the implementers know this.

>>This is where we were before color management became available.  I have
>> to
> say that color management works monumentally better than this for me.
> I've never encountered anyone who started at the printer and worked back
> before, and my recollection of events was that photoshop was well
> established before any of us had access to a colour printer (or even a
> colour PC) at home

I was using HP color printers for at least two generations before I got
Photoshop (it existed, at least in the Mac world, but I didn't jump into
it until I think version 3 in the Windows world).  I used Corel Photo
Paint for my first few years working with scans (I started around 1993 I
David Dyer-Bennet, dd-b@xxxxxxxx;

[Home] [Share Photos] [Photos] [Bryce] [Rail] [Epson Inkjet] [Scanner List] [Kitty's Korners] [Gimp Users] [Gimp for Windows]