Ad
Ad
Ad
Pages: « 1 ... 3 4 [5]   Bottom of Page
Print
Author Topic: ABW or Photoshop  (Read 11612 times)
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1675


WWW
« Reply #80 on: June 05, 2013, 06:41:30 AM »
ReplyReply

To clarify things a bit, here are my settings for the print tests that I did.  All prints were done with the Adobe Color Print Utility so that I could send untagged images through to the Epson driver.  The paper setting is 'Ultra Premium Photo Paper Luster' for Museo Silver Rag with a platen gap setting of 4 (I have found that this is the best paper setting for this paper).  For the "Epson Standard" data (probably a bad name), this was printed using 'No Color Management.'  I will be printing out a patch strip using color management and a high quality profile that I created with ArgyllCMS that I use for normal printing.  This profile includes a 51 step B/W patch set (maybe over kill but who knows).  I should have the data over to Andrea as soon as the patch set has dried down sufficiently.

Alan
Logged

MHMG
Sr. Member
****
Offline Offline

Posts: 623


« Reply #81 on: June 05, 2013, 06:53:40 AM »
ReplyReply


Assuming the strip was kept untagged and the ACPU set to no color management, what need to be confirmed is whether in the Epson driver panel,
under the "mode" menu it was selected "Epson std" or "Off (no color management)" before printing (they are mutually exclusive).


If I recall correctly, Adobe's rationale for providing the ACPU (Adobe color print utility) was to circumvent many problems people were having particularly on new Macs when trying to print untagged profiling targets. I know Alan is on a PC, so perhaps he can confirm this ACPU behavior on his OS, but at least on my MACs running OS10.8, ACPU forces the NCA mode automatically in the Epson driver (i.e., choices are grayed out on the menu) hence I concluded that the graph labeled Epson standard must be Epson NCA since Alan said he used ACPU to print the target.
Logged
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2868


« Reply #82 on: June 05, 2013, 07:26:43 AM »
ReplyReply

Just a thought. Mark McCormicks's comment on the different choices in Epson ABW and corresponding deflections from the L line makes me wonder whether for light levels like 50 Lux in musea any adaption on the curves is adequate enough for the task. It would not surprise me if less than 255 steps in a greyscale image may improve the representation of an image then. One way or another curves have to compress tone ranges within a given dynamic range when they open up other parts and the eyes are less discriminating at that light level.
More the opposite of a wide dynamic range, lots of light and the possibility that 16 bit printing improved the image compared to 8 bit.

On the probably required sRGB assigned input on Epson ABW mode: the HP Z3100/3200 PCL3 driver B&W mode has a choice between sRGB and AdobeRGB. The rest is the same, HP recommends driver color management too then. For the Epson it is interesting then that there is no equivalent greyscale gamma choice to assign that resembles the sRGB curve. I do not expect a conversion happens between color spaces and/or gammas assigned. The driver expects sRGB and uses media preset specific LUTs to create the tone range.


--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2012, 500+ inkjet media white spectral plots.
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #83 on: June 05, 2013, 07:37:02 AM »
ReplyReply

Ok,
thanks for all the clarifications,
now we know that the Alan "Epson std" labeled data set should be interpreted as "no color managed".
I don't know/use ACPU so this was not immediate for me.
Please, correct me if I'm wrong, but, from the printed results point of view, this should be the same as setting "printer manages color" in the PS print panel and "Off (no color management)" in the Epson driver, right?

Alan, do you have an explanation regarding what Roy pointed out for the strange higher values of the white L* values of the ABW-D3 data set?  Do you think it could be a measurement error?

Let me know when the new ICC data are ready, I will put it on graph.

Ciao Smiley
Andrea
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #84 on: June 05, 2013, 08:14:04 AM »
ReplyReply

On the probably required sRGB assigned input on Epson ABW mode: the HP Z3100/3200 PCL3 driver B&W mode has a choice between sRGB and AdobeRGB. The rest is the same, HP recommends driver color management too then. For the Epson it is interesting then that there is no equivalent greyscale gamma choice to assign that resembles the sRGB curve. I do not expect a conversion happens between color spaces and/or gammas assigned. The driver expects sRGB and uses media preset specific LUTs to create the tone range.

Hello Ernst,
having the possibility in the HP B&W mode driver to chose between 2.2 gamma and sRGB gamma as the one expected to be present in the source image is nice.
This is something Epson could "copy & paste" for ABW Smiley

This lead to an important consideration in my opinion:
we don't know exactly whether ABW expect sRGB-gamma or 2.2-gamma. I was thinking the Epson ABW choice was for 2.2-gamma but probably it is sRGB-gamma.
In any case it is a fixed choice, so it could not be optimal for both the cases, and is affected from media settings too.
This means that the common convincement that "ABW provides better linearity", is a little bit dangerous and could lead to some noticeable printed deviations if blindly accepted without control, as we have seen.

In order to get out the best from plain ABW, printing and measuring a simple B&W test strip assigned to the desired colorspace and with given paper/printer/setting, for the two most common "normal" and "dark" settings, for example, could be very important at least to avoid big deviations and choose the better starting point.
We know that the results will be never optimal and may vary case to case, but at least we will have the information regarding how the still present non linearity is distributed across the L* range.

There could be lucky combinations, like my showed R3000+FibreSilk+ABW-dark in AdobeRGB, that are quite good out of the box, but we know that there could be far less optimal combination for sure.

To reach a full optimization (or to correct not satisfactory results that none of the preset can accommodate with a less lucky printer/paper/setting) the use of a correction profile like the one created by my beta-linearization tool (or by QTR with the needed workarounds) or, at least, a simple manual curve L* adjustment more or less empiric, is the only way and could greatly improve the final results.
It is maybe an extra-step in the workflow, I know, but given all the efforts and the level of scrutiny we invest in these activities it could be worth.

Ciao Smiley
Andrea
« Last Edit: June 05, 2013, 08:28:39 AM by NeroMetalliko » Logged
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1675


WWW
« Reply #85 on: June 05, 2013, 08:19:37 AM »
ReplyReply

Please, correct me if I'm wrong, but, from the printed results point of view, this should be the same as setting "printer manages color" in the PS print panel and "Off (no color management)" in the Epson driver, right?

Alan, do you have an explanation regarding what Roy pointed out for the strange higher values of the white L* values of the ABW-D3 data set?  Do you think it could be a measurement error?

Let me know when the new ICC data are ready, I will put it on graph.

Ciao Smiley
Andrea

Correct about the driver settings.  I will re-measure the ABW-D3 data set to make sure the value is correct.  Just printed the ICC profiled patches and will read these later today after they have had a chance to dry down.

alan
Logged

royvharrington
Newbie
*
Offline Offline

Posts: 7


« Reply #86 on: June 05, 2013, 12:46:21 PM »
ReplyReply

Andrea,

Thanks for the link to the linearization thread -- I had not read that before.
I see now that you've gone through all the math of color spaces etc.

My main comment on the linearization issue is I don't think that the straight-line is the best approximation
for the best perceptual match.  But neither is the S-curve concept.  The white end of the range
behaves (perceptually) very well, so a straight-line is fine for that end -- in fact simple luminance
scaling is built into the whole ICC profile concept.  Only the dark end is problematic and not
addressed in the ICC -- which brought up Adobe's BPC algorithms.  So the perceptual curve needs
to compress the shadows, make the mid tones stay in the same place, and the highlight go
straight to the dmin value.  You can look at my generic ICCs Gray Matte Paper and Gray Photo Paper
in the ColorSync Utility.  See the xTRC tag for the graphs -- this are not exactly the same as your
graphs since one of the axes is K values not both L values.  So an L=50 gets converted into the K
value that produces L=50 in the driver (i.e. print).  Anyway you can see the shape of the curves.

I guess one could debate whether this is the "ideal" transformation but it is the BPC algorithm and
does seem to produce a reasonably good match visually.
See:  http://www.color.org/AdobeBPC.pdf
There's a lot of hard to decipher stuff in this but the bottom line is scale in Luminance Y -- the
very last section 7.3 step 3.

----------

The other issue is the AdobeRGB vs sRGB for Mac drivers.  With Apple's Cocoa API
the idea was to always manage the color with ColorSync.  So the default was to always convert
image output to a standard profile and let the driver handle anything special.  This works great for
displays and virtually all the screen stuff in OSX is nicely color managed.

Up through Leopard 10.5  Generic RGB (a gamma 1.8 profile) was the "standard" profile for Mac print drivers.
After that (Snow Leopard) Apple decided that sRGB (close to a gamma 2.2) would be the standard.
As a driver writer you can actually select which one you want but default of sRGB is the norm.
The Epson driver lets you choose this if Printer Manages Color is on, and Color Matching is set
to Epson Color Controls. This all makes third party papers and custom ICCs problematic -- its
just not geared to the idea very well.  But over the last few versions of OSX, CS, Epson, they
have managed to make a special program ACPU, a Photoshop Manages Color option, hidden flags
in the Print settings, and Epson driver NoCM option (forced) to allow custom ICCs for color. 

B&W is a bit strained in this though -- ACPU doesn't work, the hidden flag doesn't work, and
the NoCM is forced/ABW disabled. Apple also created a new profile called Generic Gamma 2.2 (horrible
name since its not a gamma function) that mimics the shape of the sRGB profile but in grayscale.

Looks like Epson has matched the sRGB for ABW river -- but it would be interesting to see what
they do with matte papers -- straight-line, compressed shadows.

Roy
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #87 on: June 05, 2013, 05:19:22 PM »
ReplyReply

My main comment on the linearization issue is I don't think that the straight-line is the best approximation
for the best perceptual match.  But neither is the S-curve concept.  The white end of the range
behaves (perceptually) very well, so a straight-line is fine for that end -- in fact simple luminance
scaling is built into the whole ICC profile concept.  Only the dark end is problematic and not
addressed in the ICC -- which brought up Adobe's BPC algorithms.  So the perceptual curve needs
to compress the shadows, make the mid tones stay in the same place, and the highlight go
straight to the dmin value.  You can look at my generic ICCs Gray Matte Paper and Gray Photo Paper
in the ColorSync Utility.  See the xTRC tag for the graphs -- this are not exactly the same as your
graphs since one of the axes is K values not both L values.  So an L=50 gets converted into the K
value that produces L=50 in the driver (i.e. print).  Anyway you can see the shape of the curves.

I guess one could debate whether this is the "ideal" transformation but it is the BPC algorithm and
does seem to produce a reasonably good match visually.
See:  http://www.color.org/AdobeBPC.pdf
There's a lot of hard to decipher stuff in this but the bottom line is scale in Luminance Y -- the
very last section 7.3 step 3.

Hello Roy,
many thanks for your explanations and for the direct link to the Adobe doc.
I have looked at it and it was interesting.

I know that the linear approach is only one of the possible choices,
in my case, it was only the simplest and logical to start with,
and even if not perfect for all the reasons already pointed out, the results I got were highly better than in absence of it, at least in my personal opinion.
Nothing stop the possibility to develop/try further different targets in the future, if desired.

I find interesting your idea of an "asymmetric" target, if you let me call it so. Smiley
It is very interesting even the solution to make two different general target for Matte and Photo papers, in order to compensate the different characteristics.

As per what I can understand, the resulting shapes are not casually similar to what I consider "slightly compressed in the shadows and overall a little bit dark" that are very common to see in my first-pass strip test.
The funny thing is that, usually, I compensate it when I build a linearization curve based on the straight line target. Smiley

I'm on PC so I don't know ColorSync, I have tried to dump the grayTRC values and to plot the numbers as they are inside your Matte and Photo paper ICC.
I have attached the graph because I think it could be useful, please, feel free to add any comment if you like.

Quote
The other issue is the AdobeRGB vs sRGB for Mac drivers.  With Apple's Cocoa API
the idea was to always manage the color with ColorSync.  So the default was to always convert
image output to a standard profile and let the driver handle anything special.  This works great for
displays and virtually all the screen stuff in OSX is nicely color managed.

Up through Leopard 10.5  Generic RGB (a gamma 1.8 profile) was the "standard" profile for Mac print drivers.
After that (Snow Leopard) Apple decided that sRGB (close to a gamma 2.2) would be the standard.
As a driver writer you can actually select which one you want but default of sRGB is the norm.
The Epson driver lets you choose this if Printer Manages Color is on, and Color Matching is set
to Epson Color Controls. This all makes third party papers and custom ICCs problematic -- its
just not geared to the idea very well.  But over the last few versions of OSX, CS, Epson, they
have managed to make a special program ACPU, a Photoshop Manages Color option, hidden flags
in the Print settings, and Epson driver NoCM option (forced) to allow custom ICCs for color. 

B&W is a bit strained in this though -- ACPU doesn't work, the hidden flag doesn't work, and
the NoCM is forced/ABW disabled. Apple also created a new profile called Generic Gamma 2.2 (horrible
name since its not a gamma function) that mimics the shape of the sRGB profile but in grayscale.

Looks like Epson has matched the sRGB for ABW river -- but it would be interesting to see what
they do with matte papers -- straight-line, compressed shadows.

I cannot say it for sure, but given the not changeable sRGB-like companding curve adoption we have discussed, I doubt Epson implemented in ABW an internal different general target for Matte and Photo papers as you have done. Obviously there are some differences in the various media settings, but probably the general concept/target behind it is the same (but maybe I'm wrong here).

Thanks again.

Ciao Smiley
Andrea
Logged
royvharrington
Newbie
*
Offline Offline

Posts: 7


« Reply #88 on: June 05, 2013, 06:18:14 PM »
ReplyReply

...
I find interesting your idea of an "asymmetric" target, if you let me call it so. Smiley
It is very interesting even the solution to make two different general target for Matte and Photo papers, in order to compensate the different characteristics.

As per what I can understand, the resulting shapes are not casually similar to what I consider "slightly compressed in the shadows and overall a little bit dark" that are very common to see in my first-pass strip test.
The funny thing is that, usually, I compensate it when I build a linearization curve based on the straight line target. Smiley

I'm on PC so I don't know ColorSync, I have tried to dump the grayTRC values and to plot the numbers as they are inside your Matte and Photo paper ICC.
I have attached the graph because I think it could be useful, please, feel free to add any comment if you like.

I cannot say it for sure, but given the not changeable sRGB-like companding curve adoption we have discussed, I doubt Epson implemented in ABW an internal different general target for Matte and Photo papers as you have done. Obviously there are some differences in the various media settings, but probably the general concept/target behind it is the same (but maybe I'm wrong here).

Thanks again.

Ciao Smiley
Andrea

Hi Andrea,

I'm not sure why you make a point of "asymmetric" -- that's given.  Any line between dMax L=16 and dMin L=96 is
asymmetric - since the end points are.   And every different dMax would yield a different curve/target for any system.

My only point is that a straight-line in L-space isn't as good visually as a straight-line in Y-space.  You probably
can't see the difference with photo paper because the dMax is so dark.  But with matte paper it is noticeable.
Maybe the graphs (which you did plot correctly) are misleading.  They just show what you get for a
correction curve given a simple straight-line Y-space mapping for two difference dMax values.

Overall the effect is taking into account the loss of dMax mostly in the shadows (compressing) rather than
over the whole range.

Roy
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #89 on: June 06, 2013, 02:28:51 AM »
ReplyReply

I'm not sure why you make a point of "asymmetric" -- that's given.  Any line between dMax L=16 and dMin L=96 is
asymmetric - since the end points are.   And every different dMax would yield a different curve/target for any system.

Hello Roy,
Sorry for the misleading naming, the word "asymmetric" I (somewhat improperly) used was a concept to be intended in the sense that the shadows are slightly compressed but, as you point out, the highlights should not and are going to Dmin directly.
So, using the same philosophy, an "S" shape could be considered symmetric (as concept) because the shadows and highlights are both compressed in the same way (with the obvious caveat of being anchored to black and white paper values, which are usually not equally spaced from 0 or from 100, but this was not the point). Similarly, even a straight line could be considered symmetric, as philosophy, in the sense that there is no conceptual different treatment between shadows and highlights.
In pure math term the word is not appropriate as you pointed out.

Quote
My only point is that a straight-line in L-space isn't as good visually as a straight-line in Y-space.  You probably
can't see the difference with photo paper because the dMax is so dark.  But with matte paper it is noticeable.
Maybe the graphs (which you did plot correctly) are misleading.  They just show what you get for a
correction curve given a simple straight-line Y-space mapping for two difference dMax values.

Yes thanks, I had understood the point, it could be interesting to evaluate the effect of a Y straight line as target instead of the L* one, 
in special way for the more Dmax limited Matte papers, I will keep it in consideration.

Quote
Overall the effect is taking into account the loss of dMax mostly in the shadows (compressing) rather than
over the whole range.

Yes, that was exactly the reason why I tagged this approach as "asymmetric", as previously explained, so not in negative mean, but just because of the conceptual idea behind it.

Many thanks for all the useful opinion you share here,
it is really appreciated.

Ciao Smiley
Andrea
Logged
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2868


« Reply #90 on: June 06, 2013, 05:35:25 AM »
ReplyReply

Roy, Andrea,

Is there any chance you could construct a kind of appearance model into the curve creation that would address the display lighting level in combination with the available dynamic range? Say choices for bare display, framed behind ordinary framing glass and 120 Lux, 450 Lux, 2000 Lux? Skip the 50 Lux of musea I would say. There is not much published on display light levels and B&W tone ranges.  I also did a search on the Zone System and display conditions but nothing to build on is found then. A lot more on the image creation phase and different light levels in the scene.
Interesting PDF is this one but also for the input side:
Perceptually Based Tone Mapping for Low-Light Conditions, Adam Garnet Kirk

In ICC profile creation there are choices for different color temperatures in viewing conditions but normally not for lighting levels. Some comments on that could be made for B&W prints too, the use of OBA papers and different light levels/color temperatures for example.

In the Colorsync list in a discussion called "Colormetric Accuracy in the Field" some examples were quoted of the huge difference between the viewing light in the creation phase of prints/art and the light at display. Stanley Smith and Graeme Gill comments.

--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2012, 500+ inkjet media white spectral plots.
« Last Edit: June 06, 2013, 05:44:43 AM by Ernst Dinkla » Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #91 on: June 06, 2013, 08:51:16 AM »
ReplyReply

To clarify things a bit, here are my settings for the print tests that I did.  All prints were done with the Adobe Color Print Utility so that I could send untagged images through to the Epson driver.  The paper setting is 'Ultra Premium Photo Paper Luster' for Museo Silver Rag with a platen gap setting of 4 (I have found that this is the best paper setting for this paper).  For the "Epson Standard" data (probably a bad name), this was printed using 'No Color Management.'  I will be printing out a patch strip using color management and a high quality profile that I created with ArgyllCMS that I use for normal printing.  This profile includes a 51 step B/W patch set (maybe over kill but who knows).  I should have the data over to Andrea as soon as the patch set has dried down sufficiently.

Thanks Alan for sending me the new data related to the Argyll ICC measurements,
you have not specified which is the expected colorspace of the source images during the profile creation (i.e. if you used -S sRGB or -S AdobeRGB in Argyll colprof),
so I have a similar dilemma as the previous data set here, regarding how to interpret the K values to transform it in L* values for x axis reference.

Looking to the graph I could dare to guess you used AdobeRGB in Argyll colprof, but please correct me if this is not the case.
In the meantime I have done the graphs for both AdobeRGB (2.2 gamma) and sRGB (sRGB gamma).
You can find it in attachment for further evaluation.

After the useful Mark and Roy comments and for additional convenience I have provided both the plot twice,
they are the same in fact, only with a different dotted line reference:
"LIN dotted ref" means that the dotted straight line is my usual one (connecting ink black to paper white)
"ABS dotted ref" means that the dotted straight line is the ideal L* ramp from 0 to 100.

Regarding the L* linearity:
it is visible that the 2.2-gamma plot, even if not perfectly matching my LIN dotted target, it is less dark (compressed) than the gamma-sRGB one in the shadows.
In addition, comparing the same two gamma plots to the ABS dotted target we can spot the kind of shape Mark and Roy have well explained.
For this reason I guess that if you use this ICC in AdobeRGB images you should be probably satisfied (at least regarding the tone contrast, I cannot say nothing regarding the color gamut). This is something you was not able to achieve in ABW for 2.2-gamma, as we have previously seen.

Regarding the gray neutrality: we can say that Argyll has done an excellent job here, the gray tone is perfectly neutral and smoothly rolling off to paper white tone. this result is superior to the ABW one, and I think that this could be even visible in the real print maybe.

My final thought is that, assuming your work is in Adobe RGB, for B&W print you are overall best served from this ICC (using the same settings you have used here obviously): you get a better L* shape and much better gray neutrality than in ABW. You loose only something in Dmax, but this is a known issue.

I hope this was useful,
let me know your opinion.


Ciao Smiley
Andrea
« Last Edit: June 06, 2013, 08:53:14 AM by NeroMetalliko » Logged
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1675


WWW
« Reply #92 on: June 06, 2013, 09:34:41 AM »
ReplyReply

Thanks for plotting this!  To answer your question I used AdobeRGB for the color space.  I have found that this profile does give excellent B/W prints with the only drawback that the Dmax is not quite as good as with the Epson ABW driver.  Of course as Mark has noted this may not be too important as there is a question whether it would be visibly perceptible.  It looks like this is a good vindication of my use of the 51 step B/W patch set when I make the profiles.  Now I'm going to have to go back and make some test prints under all these different conditions to look at how shadow detail is handled!!!!   Grin
Logged

NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #93 on: June 06, 2013, 09:44:22 AM »
ReplyReply

Is there any chance you could construct a kind of appearance model into the curve creation that would address the display lighting level in combination with the available dynamic range? Say choices for bare display, framed behind ordinary framing glass and 120 Lux, 450 Lux, 2000 Lux? Skip the 50 Lux of musea I would say. There is not much published on display light levels and B&W tone ranges.  I also did a search on the Zone System and display conditions but nothing to build on is found then. A lot more on the image creation phase and different light levels in the scene.
Interesting PDF is this one but also for the input side:
Perceptually Based Tone Mapping for Low-Light Conditions, Adam Garnet Kirk

In ICC profile creation there are choices for different color temperatures in viewing conditions but normally not for lighting levels. Some comments on that could be made for B&W prints too, the use of OBA papers and different light levels/color temperatures for example.

In the Colorsync list in a discussion called "Colormetric Accuracy in the Field" some examples were quoted of the huge difference between the viewing light in the creation phase of prints/art and the light at display. Stanley Smith and Graeme Gill comments.

Hello Ernst,
thanks for raising this complex argument and for pointing out to this technical publication, I will look at it because it seems very interesting, even if I don't know if it could be practical useful in real terms.

Frankly, I think you are overestimating my knowledge Smiley

The only answer I can try to formulate at the moment is that, at least in principle, it would/could be possible to compensate for every kind of behavior, within the limits set by a Lab curve-styled correction, if we know how to model/represent these different conditions you are interested in.

In simple terms, if I know that in a given light condition the perceived tone curve is shapeable in a given way, we can try to build a compensation profile based on this shape in order to print the image specifically for that kind of condition.
This could be more or less complex to realize, maybe, but, in theory, I guess it should be technically feasible.

The culprit is "how" to model each light condition in this way?
Honestly, at the moment I completely ignore how to do it and where, if any, these information can be found.

In addition, as you know, these kind of strictly targeted corrections have the disadvantage to be maybe good in the theorized conditions, but at the risk to become worse if something change, so the real usefulness is in perfectly controlled constant environmental condition only.
This mean that if you print an image for a given light condition (exhibition) and during their life time this picture is then moved/sold exposed in different conditions, it could look very bad.
This is something to keep in consideration, because maybe a more general-purpose conservative approach, even if never optimized for all conditions, in long terms is more appropriate.  

In any case this is an intriguing argument indeed,
I'm sure that there is more expert people here who could add very valuable opinions on this regard and this will really appreciated.

Ciao Smiley
Andrea  

« Last Edit: June 06, 2013, 09:48:58 AM by NeroMetalliko » Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #94 on: June 06, 2013, 10:01:23 AM »
ReplyReply

Thanks for plotting this!  To answer your question I used AdobeRGB for the color space.  I have found that this profile does give excellent B/W prints with the only drawback that the Dmax is not quite as good as with the Epson ABW driver.  Of course as Mark has noted this may not be too important as there is a question whether it would be visibly perceptible.  It looks like this is a good vindication of my use of the 51 step B/W patch set when I make the profiles.  Now I'm going to have to go back and make some test prints under all these different conditions to look at how shadow detail is handled!!!!   Grin

Glad you found it useful,
please, keep in consideration that if you print the same AdobeRGB image with the ICC and with ABW-D1 (or D2) it could be that, by comparing it with the ICC one, depending on the kind of image and light conditions, in the deep shadows you could get a slight apparent advantage in the ABW printed images, due to the showed slight "bump" present in that zone, derived from the sRGB-gamma expectation of the ABW driver.

Let we know your findings.

Ciao Smiley
Andrea
Logged
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2868


« Reply #95 on: June 06, 2013, 10:29:45 AM »
ReplyReply


Frankly, I think you are overestimating my knowledge Smiley

In addition, as you know, these kind of strictly targeted corrections have the disadvantage to be maybe good in the theorized conditions, but at the risk to become worse if something change, so the real usefulness is in perfectly controlled constant environmental condition only.
This mean that if you print an image for a given light condition (exhibition) and during their life time this picture is then moved/sold exposed in different conditions, it could look very bad.
This is something to keep in consideration, because maybe a more general-purpose conservative approach, even if never optimized for all conditions, in long terms is more appropriate.  

Ciao Smiley
Andrea  


You should not think that I could read that article but it gives an impression of the intentions :-)

There are good reasons to optimize a print for a specific display condition and that should happen at the last stage at printing time without affecting the "mother" file. A printer profile acts like that. Qimage's print filter too, which is another candidate for the job if the Qimage curves tool was not so clumsy. Describing the precise display condition in the selections will help, the light to darker settings in for example the ABW driver are too vague (Mark's explanation helps though) and I avoid similar choices in the HP drivers. The Lux numbers I quoted are estimations of average display conditions by Kodak, Wilhelm (ISO) and what is considered a level for color critical work. See the explanation of exposure levels in PDF files of Aardenburg-Imaging.

The choices will not replace print proofs for serious work but could offer a better starting point.

--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2012, 500+ inkjet media white spectral plots.



« Last Edit: June 06, 2013, 10:31:41 AM by Ernst Dinkla » Logged
MHMG
Sr. Member
****
Offline Offline

Posts: 623


« Reply #96 on: June 07, 2013, 09:13:53 AM »
ReplyReply


In simple terms, if I know that in a given light condition the perceived tone curve is shapeable in a given way, we can try to build a compensation profile based on this shape in order to print the image specifically for that kind of condition.
This could be more or less complex to realize, maybe, but, in theory, I guess it should be technically feasible.

The culprit is "how" to model each light condition in this way?
Honestly, at the moment I completely ignore how to do it and where, if any, these information can be found.

In addition, as you know, these kind of strictly targeted corrections have the disadvantage to be maybe good in the theorized conditions, but at the risk to become worse if something change, so the real usefulness is in perfectly controlled constant environmental condition only.
This mean that if you print an image for a given light condition (exhibition) and during their life time this picture is then moved/sold exposed in different conditions, it could look very bad.
This is something to keep in consideration, because maybe a more general-purpose conservative approach, even if never optimized for all conditions, in long terms is more appropriate.  

In any case this is an intriguing argument indeed,
I'm sure that there is more expert people here who could add very valuable opinions on this regard and this will really appreciated.

Ciao Smiley
Andrea  



A subject near and dear to me. Over thirty years ago, my senior research thesis at Rochester Institute of Technology explored the complex issues of human brightness perception. I got into the subject somewhat by chance by asking a professor why color slide film (remember slides?) had to be processed to a much higher contrast level than reflection prints in order to look good when projected in a dark theater. That one question set me on a lifelong course of study in the subject of color and tone reproduction in prints and photographs and ultimately to print preservation because once a printmaker achieves his/her vision of a "perfect print" why wouldn't we want to preserve the artist's original intent as long as possible?

I sought out experts at Eastman Kodak which quickly led me to the work of Bartleson and Brenneman on lateral adaption, (which in turn ultimately explained my original question about slide projection in a dark theater). However, as Andrea and others have observed, it's generally a challenge to "bake" into any given image reproduction a custom tone curve aimpoint because the printmaker has little control on how the final print will be displayed. As a good summary for some of the fascinating human visual system complexities which color scientists are working on with regard to both color appearance models and image appearance models (image appearance being even more complex) take a look at this information: http://www.cis.rit.edu/fairchild/PDFs/AppearanceLec.pdf


If you'd like to see a practical demonstration of effect of surround brightness on image appearance, the monitor checker target I developed to validate monitor calibration has three layers that can be toggled on and off to add a light, middlegray, or dark field surrounding the calibration target. The instruction layer in the image file will help you figure out how to use this test target. Photoshop is needed to use it:

http://aardenburg-imaging.com/cgi-bin/mrk/_4878ZGxkLzBeMTAwMDAwMDAwMTIzNDU2Nzg5LyoyMTA=

Note in particular how the choice of the dark surround improves one's ability to observe shadow detail (the dark numbers going from L*= 0 to L* = 9 in the target). Zooming in closer to the letters, i.e, magnifiying them in one's area of view thus affecting viewing surround conditions again, can also be seen to have an impact on visual appearance. This means that even one's choice of final print size will have an affect on what tone curve and viewing conditions are optimal for any given image.

The Bartleson-Breneman adaptation effect is often used by museum lighting designers to compensate for low overall (e.g., 50 lux) lighting levels, for example. Allow the viewer time to adjust to a room that has very low light levels (less than 10 lux). Now train a focused spot illuminating just the print at 50 lux. It will appear much brighter and more luminuous than if one just set the whole room to 50 lux average even illumination on white walls. This is the Bartleson-Brenneman effect being used in practice to create superior display environment for valuable prints and paintings that require lower overall illumination levels for best preservation practices.

I mention all this as a practical/informative exercise that printmakers can try in their own studios, the goal, IMHO, being to produce prints that look better under a wider range of real world lighting conditions and tailored to suit one's personal tastes. I haven't done an exhaustive technical study of the subject, but my sense is that modern ICC profiles and the CIELAB color model are designed to produce optimal color and tone viewing when print illumination levels approach 2500 lux on white or gray walls lit with broad and even lighting. One can drop to about 500 lux and still find the reproduction quite good. But as one goes down from there, especially as one gets below 200 lux for middle-aged people with reduced ability to discriminate color in low lighting levels, then compensation with a lighter tone curve aim point and/or some attention being paid to viewing surround conditions, overall print size, and even mat border size/color starts to come into play for a better print viewing experience.

regards,
Mark
http://www.aardenburg-imaging.com
« Last Edit: June 07, 2013, 09:34:50 AM by MHMG » Logged
Pages: « 1 ... 3 4 [5]   Top of Page
Print
Jump to:  

Ad
Ad
Ad