Printer color management best practise...

[Ethan Hansen]

Ethan, my workflow is a little different so I'm wondering if you
can offer some insight for me. I'm shooting NEF with a D70 and
using Photoshop Camera RAW to do the conversion. I'm printing with
an i960 using custom profiles. While I realize that ProPhoto RGB
essentially contains all visible color and is overkill, I'm using
it as my working space for a couple of reasons:

1) Both my camera and printer are capable of producing colors
outside of Adobe RGB, and it would seem to me that to get the most
out of this equipment I would want to use a working space that
fully contains the gamut of both devices. ProPhoto RGB seems to be
the only color space that does that (at least, out of the ones
installed on my system).

PP RGB is the only option Camera Raw offers that fits the bill here. It is what I use for the same reasons - at least when I am working on imager in ACR for which the final output is potentially something that surpasses Adobe RGB. I do most of my image processing using Capture One which allows using any color space for output. I'm partial to Bruce Lindbloom's Beta RGB here. It is smaller than ProPhoto RGB, hence fewer quantization errors appear. It is also large enough to encompass the output of both printers and DSLRs. Finally, it is a gamma 2.2 color space, which is better than PP RGB's 1.8 for B&W work (but that borders on a how many photons can dance on the head of a pixel discussion).

2) According to Bruce Fraser's Camera RAW book, Photoshop uses
ProPhoto RGB internally during the conversion, so by keeping the
file in that working space I avoid a conversion.

Bruce certainly is more knowledgeable about ACR than me. Still, something sounds strange here, and not having a copy of his book at hand, I can't check. A strength of ACR is that it carries out most adjustments in the lineary encoded raw output space. Many are performed on the luminance data alone. The internal camera profiles are then used to convert into one of the four output color space selections. I do not see where an intermediate pass through PP RGB would be beneficial. This is a problem with Capture One, which uses a funky "Phase One RGB" space to do all too many adjustments in. If ACR does indeed convert to PP RGB, then you certainly would avoid a second conversion.

Just checked something: I processed two versions of an image having colors that fell outside the sRGB gamut in ACR. I sent one to sRGB, the other to ProPhoto. Converting the PP RGB version to sRGB afterwards gave an image exactly identical to the sRGB one. This cursory experiment indicates Bruce is right.

Given those facts does ProPhoto RGB not make sense to use as a
working space? Unlike the orginal poster, I do NOT convert my
images to sRGB at any point. I also do all my editing in 16-bit
mode (even the plug-ins I use regularly support 16-bit now). So
basically my images start out in ProPhoto RGB 16-bit and stay that
way through the entire workflow, until printing when QImage will
use my paper-specific custom printer profile. This approach seems
to be working pretty well for me, but if there are ways to improve
it I'm always open to suggestion.

All sounds reasonable to me! If, however, you have an image that requires heavy-handed editing, there might well be merit to dropping to Adobe RGB first. This is particularly true for prints on matte papers where the output range is well within aRGB. A color space conversion here will do less damage than the potential for posterization in yellows and browns that can easily happen with large edits in PP RGB.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[kovacj]

I'm trying to validate some assumptions to confirm that the
workflow that I use is optimal. I have not been able to find any
reference to what input a print driver can accept (specificially
the Canon in my case) natively (i.e. without applying any
"automatic" color space conversion as I'd like to be able to
control the conversion myself in Photoshop). I am currently
operating on the assumption that the driver will natively only
support sRGB (I would like to be wrong about this). Does the
printer accept other input (e.g. CMYK) natively? Has anyone come
across this information?

The printer driver only supports RGB. If you send it CMYK data, the
driver first converts to RGB using a truly awful algorithm and the
converts the resulting mess back into CcMmYK. If you set the Print
Type to "NONE" as recommended above, the driver does not perform
any additional RGB color conversions. The output exceeds even the
bounds of Adobe RGB in some colors rather than being clipped to
smaller than sRGB. If your printer driver does not list the "None"
setting, you need to upgrade the driver - older drivers lacked the
necessary setting. Use a profile built for this setting and you can
control all the color parameters in Photoshop. You can view 3-D
models of the actual output range of several Canon models on our
web site at:
http://www.drycreekphoto.com/tools/printer_gamuts/

--
Ethan Hansen
http://www.drycreekphoto.com/

Hi Ethan,

When you say RBG data, that's a bit too vague IMO; is it sRGB, aRGB, ppRGB? Is it 8-bit or 16 bit?

Isn't it puzzling that for devices that like cameras, the supported image formats are well documented (e.g. RAW, TIFF, sRGB, aRGB), but I have yet to come across a similar spec for any printer (driver). How does one know if a printer driver supports ppRGB (for example) for the translation to its native format (e.g CcMmYK)?

If a printer's Gamut is wider than can be represented by a given color space (e.g. aRGB) - then that color doesn't get printed, right? I don't know where the color information would come from?

I've done a bit of reading into Windows ICM 2.0, I think that the information that I'm looking for is centered around the PCS (Profile Connection Space) - the following link is a good start: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/icm/icm_786l.asp . I wonder if RGB (any RGB) is used when printing from Photoshop, or does Photoshop pass information to the driver in the PCS?

Much to learn, I don't remember film being quite this much fun...
Thanks again,
Jarek

 

[Jeff Kohn]

Ethan, thanks for the reply and the excellent information...

PP RGB is the only option Camera Raw offers that fits the bill
here. It is what I use for the same reasons - at least when I am
working on imager in ACR for which the final output is potentially
something that surpasses Adobe RGB. I do most of my image
processing using Capture One which allows using any color space for
output. I'm partial to Bruce Lindbloom's Beta RGB here. It is
smaller than ProPhoto RGB, hence fewer quantization errors appear.
It is also large enough to encompass the output of both printers
and DSLRs. Finally, it is a gamma 2.2 color space, which is better
than PP RGB's 1.8 for B&W work (but that borders on a how many
photons can dance on the head of a pixel discussion).

I've heard some others mention Beta RGB but didn't really know much about it, is it something I can get somewhere? You mention gamma 1.8 versus 2.2, which raises a red flag for me since I'm working on PC's where it's my understanding that the "native" gamma is 2.2 Should I be concerned about this? Assuming I stick with ACR for conversions, do you think there would be any advantages/disadvantages to having ACR convert to ProPhoto RGB and then immediately converting to Beta RGB before I start editing?

2) According to Bruce Fraser's Camera RAW book, Photoshop uses
ProPhoto RGB internally during the conversion, so by keeping the
file in that working space I avoid a conversion.

Bruce certainly is more knowledgeable about ACR than me. Still,
something sounds strange here, and not having a copy of his book at
hand, I can't check. A strength of ACR is that it carries out most
adjustments in the lineary encoded raw output space. Many are
performed on the luminance data alone. The internal camera profiles
are then used to convert into one of the four output color space
selections. I do not see where an intermediate pass through PP RGB
would be beneficial. This is a problem with Capture One, which uses
a funky "Phase One RGB" space to do all too many adjustments in. If
ACR does indeed convert to PP RGB, then you certainly would avoid a
second conversion.

I oversimplified a bit, looking back at the book what Bruce actually says is that after demosaicing, white balance and calibration adjustments are done, "the remaining operations are carried out in the intermediate linear-gamma version of ProPhoto RGB."

Just checked something: I processed two versions of an image having
colors that fell outside the sRGB gamut in ACR. I sent one to sRGB,
the other to ProPhoto. Converting the PP RGB version to sRGB
afterwards gave an image exactly identical to the sRGB one. This
cursory experiment indicates Bruce is right.

Given those facts does ProPhoto RGB not make sense to use as a
working space? Unlike the orginal poster, I do NOT convert my
images to sRGB at any point. I also do all my editing in 16-bit
mode (even the plug-ins I use regularly support 16-bit now). So
basically my images start out in ProPhoto RGB 16-bit and stay that
way through the entire workflow, until printing when QImage will
use my paper-specific custom printer profile. This approach seems
to be working pretty well for me, but if there are ways to improve
it I'm always open to suggestion.

All sounds reasonable to me! If, however, you have an image that
requires heavy-handed editing, there might well be merit to
dropping to Adobe RGB first. This is particularly true for prints
on matte papers where the output range is well within aRGB. A color
space conversion here will do less damage than the potential for
posterization in yellows and browns that can easily happen with
large edits in PP RGB.

Thanks, glad to know I'm on the right track (for the most part at least). There's so much information floating around out there about color management, workflow, etc, and much of it is conflicting so sometimes it can be hard to wade through it all and figure out what actually works. One last question, what would your definition of "heavy-handed editing" be? For most images I'll do some layer-based post-capture sharpening, maybe a minor adjustment with levels/curves or large-radius USM for contrast enhancment. For the most part that's it although sometimes I'll need to do Noise Reduction, "creative" sharpening, or in the case of portraits for instance some cloning/healing.

Thanks again,

--
Jeff Kohn
Houston, TX
http://www.pbase.com/jkohn

 

[kovacj]

Hi Pat,

Color is for wusses, B&W is where it's at...

Just kidding. What started this quest for knowledge is trying to determine if I could actually print a decent monochome image on my S900 (I also have access to an S9000). After some experimentation, I was able to get some respectable results by converting the images to Duotone and then back to RGB - but only if I used the workflow described in the initial message in this thread.

I have noticed very subtle differences if color prints as well (blues are where it's at for comparisons IMO) with different workflows, but the differences are much more apparent in the monochome(ish) prints.

I'm trying to understand - at the very lowest level - how/when the image data is converted and/or translated from the time I hit the print button to when the ink starts to fly.

A funny turnaround from the film days (bad old days!) - B&W was easy and color was hard.

Cheers,
Jarek

 

[Ethan Hansen]

I think all this is coming down to what can the printer driver
accept as input and what does it do with the data (i.e. how does it
work under the covers) - i.e. if ICM is not enabled, what role does
the default profile play?

On your Canon, if you set the Print Type to "None" and disable ICM, the default profile is irrelevant.

To answer your question re. Ilford Classic Pearl paper (the black
box, not the red one), the recommended setting is HGPF - check out
C1 at...

[SNIP]

This is more than a recommended setting. It is the condition the profile was built for. If you configure the printer driver differently, you guarantee the profile will be inaccurate. A printer profile simply provides a snapshot of how a particular printer behaved when the profiling target was printed. If anything changes - printer, ink, paper, or driver settings - the conditions no longer reflect those the profile was built for. The major paper manufacturers, with the glaring exception of HP, and both Canon and aEpson with their ink have been very good about maintaining surprisingly stable manufacturing tolerances. Epson created much bad press for themselves a few years back when they moved ink manufacturing from Japan to China. The ink properties changed as well. Following that flap, Epson and Canon have been more careful to keep a steady course.

Color geek addendum: Based on our data of some thousands of printer/paper combinations we have measured through our custom printer profiling service, the least variable part of the equation is the paper. We see average batch-to-batch variations of 0.3-0.6 Delta E-2000 (one Delta-E approximates the minimum visible color difference under ideal lighting) on papers made a year apart. The inks drift by 1.0 - 1.5 dE-2000 over the same timeframe. This makes a single profile perform adequately for most photographers for at least several years. We offer a subscription profiling service for photographers with more demanding needs than this.

The average printer-to-printer variation, by comparison, is 4-7 dE-2000. This is somewhat misleading, as printers end to differ most in a selected color range. If just that range is viewed, the difference jumps up to 8-15 dE-2000. This level is easily visible, and explains why you find some photographers prasing the accuracy of canned profiles while others complain about a persistant color cast when using the same setup. Both are right - their printers are just different.

A final comment, and I'll shut up. I mentioned HP's paper was all over the place. We have seen variations in their Premium Photo Plus Paper of 6 dE-2000. The papers in question were purchased a month apart from the same retailer. Samples from other customers showed similar variability.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[Ethan Hansen]

Hi Ethan,

When you say RBG data, that's a bit too vague IMO; is it sRGB,
aRGB, ppRGB? Is it 8-bit or 16 bit?

It's RGB. Just numbers. When you disable ICM and set the print type to None, the printer driver makes some internal assumption of how a particular RGB number translates to C(c)M(m)YK(k) + O, R, G, or B for some desktop models. A profile built using these settings determines the mapping used. The printer itself has no conception of what a color space is - it deals with numbers. The role of the profile is to give physical meaning to these numbers.

Isn't it puzzling that for devices that like cameras, the supported
image formats are well documented (e.g. RAW, TIFF, sRGB, aRGB), but
I have yet to come across a similar spec for any printer (driver).
How does one know if a printer driver supports ppRGB (for example)
for the translation to its native format (e.g CcMmYK)?

The difference is that a camera is an input device, while a printer is an output device. Camera makers have had the importance of color management slowly beaten into their heads. The original Nikon D1, for example, did not tag its output with any color space. Many photographers complained about the infamous "magenta cast" to certain skin tones. There was no problem with the camera color, just with the interpretation. A camera profile fixed things up, while simply assuming the D1 shot in Adobe RGB made many people look like terminal alcholics.

A printer, by contrast, takes your input and spaltters ink on the page. Nooprinter, marketing claims to the contrary, prints in a working color space such as sRGB or Adobe RGB. These color spaces are mathematical abstractions; printers are not. A printer uses its own, unique color space. If you view the models I have provided links to in the posts above, you will see the difference. A editing color space is defined by the coordinates of the red, green, and blue primaries, the white and black points, and a gamma curve that determines how to get from one to the other. None of this matters to a printer.

The only way to see what colors a printer can actually reproduce is to measure it. This also strongly depends on what paper is used - once again, see the plots. The same printer that can print colors falling well outside the limits of Adobe RGB on glossy paper can't come close to the range of even sRGB on matte surfaces.

If a printer's Gamut is wider than can be represented by a given
color space (e.g. aRGB) - then that color doesn't get printed,
right? I don't know where the color information would come from?

Yep. The Spinal Tap printer has yet to be invented - you can't set a printer to 11. Here again is where you need accurate profiles. You can use Photoshop's Soft proofing to see how the final print will appear and what colors fall outside the printer's gamut. The profile governs how these colors will be rendered. You can choose the rendering intent that works best with the image content and, if the mapping is not to your tastes, perform manual edits. A primer on using profiles with Photoshop is on our web site at:
http://www.drycreekphoto.com/Learn/profiles.htm

I've done a bit of reading into Windows ICM 2.0, I think that the
information that I'm looking for is centered around the PCS
(Profile Connection Space) - the following link is a good start:

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/icm/icm_786l.asp . I wonder if RGB (any RGB) is used when printing from Photoshop, or does Photoshop pass information to the driver in the PCS?

The PCS is the intermediate space between your editing color space and the printer space. Neither RGB nor CMYK numbers specify a particular color. A color space is necessary to bring meaning to the numbers. See http://www.drycreekphoto.com/Learn/color_management.htm for an overview. The PCS is usually LAB or XYZ. Numbers in these device-independent spaces do specify an exact color. The image data are converted from the image color space to the PCS space. A second conversion is made to the printer color space. If you use ICM, the image is sent to the printer driver in the image color space and the driver and/or ICM performs the conversion. If you disable ICM, the driver does not convert anything - you do this beforehand in Photoshop.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[kovacj]

Hi Phil,

I was careful and noted the settings on edges of the prints - that said, I'm probably nowhere near 5-9's reliable...

I did try "double profiling" and found the opposite - images came out cooler and quite a bit darker - the coolness may be a by-product of my using a warm Duotone instead of strainght grayscale.

Cheers,
Jarek

 

[Ethan Hansen]

I've heard some others mention Beta RGB but didn't really know much
about it, is it something I can get somewhere?

See Bruce Lindbloom's web site at: http://www.brucelindbloom.com/BetaRGB.html

The advantage of Beta RGB is it is "just big enough" to fit essentially all colors that can be captured or printed. It also has the standard attributes an editing space should: gray balance (equal amounts of red, green, and blue make a neutral color) and perceptual uniformity (edits can be applied evenly to the entire space rather than being concentrated in a single region). These two points are why you do not want to edit in a printer, camera, or monitor color space: click-balancing creates a color cast rather than eliminating it, and edits are not applied uniformly.

The limits of ProPhoto RGB were set to safely include the maximally saturated yellow E6 film could record. This required pushing the red and green primaries out so far they are imaginary colors. (Side note: editing color spaces are defined by the red, green, and blue primaries. This makes a triangular-shaped color space. The color gamut of actual devices - cameras, scanners, film, printers, monitors, etc. are irregularly shaped blobs. Enclosing a large blob within a triangle requires pushing the points of the triangle far out in space.)

Beta RGB won't hold the extreme E6 yellows. That is not a serious concern for me, as no digital camera will record these shades nor can any printer reproduce them. Sacrificing these unobtainable and irreproducible colors to get a better behaved working space is worth it. Frankly, I have not edited enough oddball images in Camera Raw to have come to a firm conclusion about whether or not a conversion to Beta RGB is a benefit or a detriment.

You mention gamma
1.8 versus 2.2, which raises a red flag for me since I'm working on
PC's where it's my understanding that the "native" gamma is 2.2
Should I be concerned about this? Assuming I stick with ACR for
conversions, do you think there would be any
advantages/disadvantages to having ACR convert to ProPhoto RGB and
then immediately converting to Beta RGB before I start editing?

The monitor and working space gammas are independant of each other. The rationale behind a gamma 2.2 editing space is that it more closely matches the innate gamma of the LAB color space used as the intermediary for color space conversions. If the gammas are matched, there will be fewer conversion artifacts (show up as posterization or banding). As I mentioned above, in the larger scheme of things, this is way down on the list of what one needs to worry about.

[SNIP]

Thanks, glad to know I'm on the right track (for the most part at
least). There's so much information floating around out there about
color management, workflow, etc, and much of it is conflicting so
sometimes it can be hard to wade through it all and figure out what
actually works. One last question, what would your definition of
"heavy-handed editing" be? For most images I'll do some layer-based
post-capture sharpening, maybe a minor adjustment with
levels/curves or large-radius USM for contrast enhancment. For the
most part that's it although sometimes I'll need to do Noise
Reduction, "creative" sharpening, or in the case of portraits for
instance some cloning/healing.

By heavy-handed, I meant images where you need to do large tonal or color edits. These usually fall into the category of trying to salvage an image from a much less than ideal source. For example, a severely backlit shot that has saturated colors (particularly yellow or brown) in the underexposed area. Bringing this up in PP RGB can create posterization where the same edits performed in Adobe RGB would not.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[kovacj]

Hi Ethan,

It's RGB. Just numbers. When you disable ICM and set the print type
to None, the printer driver makes some internal assumption of how a
particular RGB number translates to C(c)M(m)YK(k) + O, R, G, or B
for some desktop models.

That doesn't fly - the embedded color space is required to give the numbers context; i.e. are you saying that R=0xFF, G=0, B=0 in sRGB is the same red as R=0xFF, G=0, B=0 in aRGB? Surely the printer would have to interpret the number differently based on the color space?

But... it's all moot anyways as PCS is the common image currency between the printing application and the driver; i.e. a printer(driver) doesn't accept RGB data, it accepts PCS data - I think you confirm that in the last para of your previous message. I guess that's the key that I was looking for.

Many, many thanks for the info..
Jarek

 

[kovacj]

Hi Jeff,

I don't think you're correct here for a couple of reasons. First,
if printers only supported sRGB I don't see how there would be much
point in using ICC profiles, everything would just be sRGB. Second,
I've used an Eye One Photo to create my printer profiles, and
resulting profiles have colors in them that exceed not only sRGB
but even Adobe RGB. If the printer was only capable of working in
sRGB I don't see how this would be possible. Bottom line is, I
think you are limiting yourself by converting your images to sRGB.

I was pretty sure that I couldn't be correct in my assumption as I recall reading that the S900's gamut was wider than aRGB in the reds (I think). I didn't know how the "guts" of the driver handled the different type of data, and I couldn't think of a better way to ask the question - sorry if I caused any confusion.

The answer that I was looking for is - I think - an intermediate "format" called the Profile Connection Space (PCS). The bit of reading that I've been able to do on the Microsoft site re. color management seems to point in this directly (I've only scratched the surface).

I am going to repeat my Duotone printing experiment and post the details, including the image, steps taken, and scanned results (should be good enough to show differences - I hope - othewise you'll all just think I'm nuts).

Cheers,
Jarek

 

[Ethan Hansen]

It's RGB. Just numbers. When you disable ICM and set the print type
to None, the printer driver makes some internal assumption of how a
particular RGB number translates to C(c)M(m)YK(k) + O, R, G, or B
for some desktop models.

That doesn't fly - the embedded color space is required to give the
numbers context; i.e. are you saying that R=0xFF, G=0, B=0 in sRGB
is the same red as R=0xFF, G=0, B=0 in aRGB? Surely the printer
would have to interpret the number differently based on the color
space?

No. Read what I wrote. What I said is that RGB numbers by themselves have no meaning until a color space is associated with them. The printer driver does not interpret the colors based on anything except hard-wired algorithms that convert a particular RGB triplet into a predefined CMYK (+etc) output. The color intelligence is supplied by a color engine. This can be the Windows ICM engne, Adobe's ACE, Apple's ColorSync, Kodak's, etc. The point I'm trying to get across is that when you disable color management (see your talk about turning off ICM above) and set the Canon print type to "None", the driver is doing no RGB color conversions. None. Zip. Nada.

But... it's all moot anyways as PCS is the common image currency
between the printing application and the driver; i.e. a
printer(driver) doesn't accept RGB data, it accepts PCS data - I
think you confirm that in the last para of your previous message.
I guess that's the key that I was looking for.

Wrong. The driver does not accept data in a device-independent color space. It takes RGB input. Again, read what I wrote - ceaseless repetition is getting tiring. A Profile Connection Space is just that - a connection. The connection in this case is between the color space of your image and that of the printer. Where the conversion takes place is up to you. If you want the printer driver to handle the conversion (hint: if you use Photoshop this is dumb), ICM takes over. You can also convert in Photoshop to use the superior Adobe Color Engine and enable BPC. In either case, you are slinging RGB numbers around - that is what the printer driver takes.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[William Carson]

I've downloaded BetaRGB.icc from the brucelindbloom site. Where do I put it? In photoshop Profiles? Does it showup as a conversion option like sRGB and adobe RGB? In otherwords, how do I use it? Thanks for all this useful info in this discussion.
Will

I've heard some others mention Beta RGB but didn't really know much
about it, is it something I can get somewhere?

See Bruce Lindbloom's web site at:
http://www.brucelindbloom.com/BetaRGB.html

The advantage of Beta RGB is it is "just big enough" to fit
essentially all colors that can be captured or printed. It also has
the standard attributes an editing space should: gray balance
(equal amounts of red, green, and blue make a neutral color) and
perceptual uniformity (edits can be applied evenly to the entire
space rather than being concentrated in a single region). These two
points are why you do not want to edit in a printer, camera, or
monitor color space: click-balancing creates a color cast rather
than eliminating it, and edits are not applied uniformly.

The limits of ProPhoto RGB were set to safely include the maximally
saturated yellow E6 film could record. This required pushing the
red and green primaries out so far they are imaginary colors. (Side
note: editing color spaces are defined by the red, green, and blue
primaries. This makes a triangular-shaped color space. The color
gamut of actual devices - cameras, scanners, film, printers,
monitors, etc. are irregularly shaped blobs. Enclosing a large blob
within a triangle requires pushing the points of the triangle far
out in space.)

Beta RGB won't hold the extreme E6 yellows. That is not a serious
concern for me, as no digital camera will record these shades nor
can any printer reproduce them. Sacrificing these unobtainable and
irreproducible colors to get a better behaved working space is
worth it. Frankly, I have not edited enough oddball images in
Camera Raw to have come to a firm conclusion about whether or not a
conversion to Beta RGB is a benefit or a detriment.

You mention gamma
1.8 versus 2.2, which raises a red flag for me since I'm working on
PC's where it's my understanding that the "native" gamma is 2.2
Should I be concerned about this? Assuming I stick with ACR for
conversions, do you think there would be any
advantages/disadvantages to having ACR convert to ProPhoto RGB and
then immediately converting to Beta RGB before I start editing?

The monitor and working space gammas are independant of each other.
The rationale behind a gamma 2.2 editing space is that it more
closely matches the innate gamma of the LAB color space used as the
intermediary for color space conversions. If the gammas are
matched, there will be fewer conversion artifacts (show up as
posterization or banding). As I mentioned above, in the larger
scheme of things, this is way down on the list of what one needs to
worry about.

[SNIP]

Thanks, glad to know I'm on the right track (for the most part at
least). There's so much information floating around out there about
color management, workflow, etc, and much of it is conflicting so
sometimes it can be hard to wade through it all and figure out what
actually works. One last question, what would your definition of
"heavy-handed editing" be? For most images I'll do some layer-based
post-capture sharpening, maybe a minor adjustment with
levels/curves or large-radius USM for contrast enhancment. For the
most part that's it although sometimes I'll need to do Noise
Reduction, "creative" sharpening, or in the case of portraits for
instance some cloning/healing.

By heavy-handed, I meant images where you need to do large tonal or
color edits. These usually fall into the category of trying to
salvage an image from a much less than ideal source. For example, a
severely backlit shot that has saturated colors (particularly
yellow or brown) in the underexposed area. Bringing this up in PP
RGB can create posterization where the same edits performed in
Adobe RGB would not.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[Jeff Kohn]

I've downloaded BetaRGB.icc from the brucelindbloom site. Where do
I put it? In photoshop Profiles? Does it showup as a conversion
option like sRGB and adobe RGB? In otherwords, how do I use it?
Thanks for all this useful info in this discussion.
Will

I'm sure Ethan or somebody will correct me if I'm wrong, but I think the idea is to use it as your working space in Photoshop. On Windows XP you should just need to put it in the Windows\system32\spool\drivers\color folder (you may also need to right-click on it and choose "install", not sure if that step is necessary). Then you should be able to choose it as your RGB working space from the Color Settings dialog in Photohops.

--
Jeff Kohn
Houston, TX
http://www.pbase.com/jkohn

 

[Dominic.Chan]

I'm sure Ethan or somebody will correct me if I'm wrong, but I
think the idea is to use it as your working space in Photoshop. On
Windows XP you should just need to put it in the
Windows\system32\spool\drivers\color folder (you may also need to
right-click on it and choose "install", not sure if that step is
necessary). Then you should be able to choose it as your RGB
working space from the Color Settings dialog in Photohops.

You don't need to both; i.e., by right-clicking to install a profile from any folder, the file will automatically be copied to the required folder.

 

[William Carson]

Hello and thanks for replying. I was able to successfully install the profile. I am using a mac OSX so when I dropped the profile on the (path) library> colorsync> profiles> I was asked to type the administrater password before it would allow it.
Will

I'm sure Ethan or somebody will correct me if I'm wrong, but I
think the idea is to use it as your working space in Photoshop. On
Windows XP you should just need to put it in the
Windows\system32\spool\drivers\color folder (you may also need to
right-click on it and choose "install", not sure if that step is
necessary). Then you should be able to choose it as your RGB
working space from the Color Settings dialog in Photohops.

You don't need to both; i.e., by right-clicking to install a
profile from any folder, the file will automatically be copied to
the required folder.

 

[spanky181896]

Hi Jarek

I'd seen that before in fact it's on my comp disc. Where we got mixed up was you are using the Gallerie 'Classic' papers while I use the Gallerie 'Smooth' line... yes, you are using the right driver settings...

Do you not notice any 'pooling' in the classic papers? They say Classic pearl is the better! (I think the texture hides the 'pooling' better than the smooth gloss's plain flat surface. I need to get some ink soon so I'll probably get a pack of the classic pearl.

I'll tell you another good paper I tested (I used the same torture test method only I put sellotape across the surface to seal it) A good paper that hardly faded at all and never washed off under a running tap was Ilford Printasia Photo Glossy Paper, a 238gsm glossy paper that seems like a good mix between a nanopoous and a swellable paper, taking the best bits fro each type. There's also a Printaisa Premium Glossy Photo Paper...

I found I got slight pooling with this paper, it left the printer 'tacky', it didn't wash off when rubbed under running water and it was a really good non-fading paper! I used the HGPF media setting but maybe a PP Pro, or other setting may cure it?

My best

 

[kovacj]

Do you not notice any 'pooling' in the classic papers? They say
Classic pearl is the better! (I think the texture hides the
'pooling' better than the smooth gloss's plain flat surface. I need
to get some ink soon so I'll probably get a pack of the classic
pearl.

No, no pooling in the Classic Pearl provided the correct settings and profile are used - they are not insta-dry like the Smooth Pearl paper (the ink is tacky so don't let the prints stack in the output tray), but they are also not insta-fade like the Smooth Pearl paper either.

I'll tell you another good paper I tested (I used the same torture
test method only I put sellotape across the surface to seal it) A
good paper that hardly faded at all and never washed off under a
running tap was Ilford Printasia Photo Glossy Paper, a 238gsm
glossy paper that seems like a good mix between a nanopoous and a
swellable paper, taking the best bits fro each type. There's also a
Printaisa Premium Glossy Photo Paper...

Thanks for the tip, I don't recall seeing this paper in the photo stores in my neck of the woods, but I'll look again.

Cheers,
Jarek

 

[kovacj]

First things first - I apologize for the ceaseless repetition, I can only try and offset this by thanking you for the time and effort that you've put into this thread; I am very grateful for your help and I can assure that, from this end in any case, your efforts are not wasted.

Here's what I think I know so far:

• at some lower level in the driver - there is a translation of RGB values to how to spit the CcMmYK ink to the paper - the RGB values are assumed to be in the printer's color space; whether or not they are in the right space is moot, that's how they are interpreted.

• a media-ink-printer profile describes the RGB color space that the printer expects. A media-ink-printer profile can be used by a color engine (e.g. ACE) to convert the image from a working space - such as aRGB or ppRGB - to the printer's RGB color space.

• the lower-level in the driver does not make use of profiles.

• at some higher level in the driver - functionality is provided to convert an image from some RGB color space to the RGB color space described by the active media-ink-printer profile. It's better to do this step in Photoshop, but if you let the driver do the conversion, it's at this stage that the driver uses the media-ink-printer profile (the driver likely delegates the conversion task to the Microsoft ICM (which uses PCS), but that's irrelevant).

• the prefered method is to use Photoshop (or Qimage, etc..) to convert an image from it's working color space to the RGB color space described by the media-ink-printer profile - it is then important to disable all color management in the printer driver so that the RGB values are not modified as they are in the printer's RGB color space after the conversion; converting again (i.e. "double-profiling") will likely assume the input is sRGB and change the RGB values again (this woudl be bad).

• in the case of the printer driver for the Canon S900 (v1.6.1) you have to select the right settings to completely disable any color conversion; i.e. Color Adjustment = Manual, Enable ICM deselected, and Print Type = None. Some amount of color correction is enabled by default.

If the above assumptions are correct, then the process is in fact quite simple, thought by no means obvious IMO. As the saying goes, the devil's in the details, that certainly applies here.

Much thanks again for all you help.
Jarek

 

[Ethan Hansen]

• at some lower level in the driver - there is a translation of RGB

values to how to spit the CcMmYK ink to the paper - the RGB values
are assumed to be in the printer's color space; whether or not they
are in the right space is moot, that's how they are interpreted.

Yes, except that you are giving the driver too much credit. It has a hardwired conversion between RGB and CMYK values. There is some level of intelligent dithering depending on what the neighboring pixels are, but this is a second order effect.

• a media-ink-printer profile describes the RGB color space that

the printer expects. A media-ink-printer profile can be used by a
color engine (e.g. ACE) to convert the image from a working space -
such as aRGB or ppRGB - to the printer's RGB color space.

• the lower-level in the driver does not make use of profiles.

Yes to both.

• at some higher level in the driver - functionality is provided to

convert an image from some RGB color space to the RGB color space
described by the active media-ink-printer profile. It's better to
do this step in Photoshop, but if you let the driver do the
conversion, it's at this stage that the driver uses the
media-ink-printer profile (the driver likely delegates the
conversion task to the Microsoft ICM (which uses PCS), but that's
irrelevant).

The driver uses the system level color engine - ICM on Windows, ColorSync on a Mac. The primary reason I recommend using the Adobe engine is that it intelligently scales black points. This problem occurs in most images: the darkest black the printer can produce is lighter than the black point of the color space the image is in. Without Black Point Compensation (BPC), all tones in your image that are darker than the ink black get clipped to black. There goes your shadow detail. BPC addresses an omission in the ICC specifications, which is why more simple-minded color converters, e.g. ICM, do not include it.

• the prefered method is to use Photoshop (or Qimage, etc..) to

convert an image from it's working color space to the RGB color
space described by the media-ink-printer profile - it is then
important to disable all color management in the printer driver so
that the RGB values are not modified as they are in the printer's
RGB color space after the conversion; converting again (i.e.
"double-profiling") will likely assume the input is sRGB and change
the RGB values again (this woudl be bad).

QImage includes some form of BPC. I do not know how it works, nor if it is up to the quality one gets from the Adobe engine. You are right about not double profiling. I do not know what color space is assumed for printing - it may well be sRGB.

• in the case of the printer driver for the Canon S900 (v1.6.1) you

have to select the right settings to completely disable any color
conversion; i.e. Color Adjustment = Manual, Enable ICM deselected,
and Print Type = None. Some amount of color correction is enabled
by default.

If the above assumptions are correct, then the process is in fact
quite simple, thought by no means obvious IMO. As the saying goes,
the devil's in the details, that certainly applies here.

The whole flow takes some caree to configure the first time through. You can store combined printer settings in what Canon confoundingly calls "profiles". These are not color profiles; they are handy groups of media and printer settings. Save one for each paper type you use, and life is easier. In a similar manner, you can save Soft Proof settings (color profile + rendering intent + BPC options) in Photoshop. At that point, all printing takes is recalling the correct Soft Proof in Photoshop, choosing the appropriate "profile" in Canon's driver, and printing. Selecting a PS Soft Proof setup automatically fills out the Color Management output options in the PS Print with Preview dialog. There may well be a method to do the same in QImage, but I do not use that program frequently enough to do more than swear at the UI.

--
Ethan Hansen
http://www.drycreekphoto.com/

 

[Ethan Hansen]

Do you not notice any 'pooling' in the classic papers? They say
Classic pearl is the better! (I think the texture hides the
'pooling' better than the smooth gloss's plain flat surface. I need
to get some ink soon so I'll probably get a pack of the classic
pearl.

No, no pooling in the Classic Pearl provided the correct settings
and profile are used - they are not insta-dry like the Smooth Pearl
paper (the ink is tacky so don't let the prints stack in the output
tray), but they are also not insta-fade like the Smooth Pearl paper
either.

Also bre careful of water spotting. The Classic papers are not waterproof with many dye inks.
--
Ethan Hansen
http://www.drycreekphoto.com/