Luminous Landscape Forum

Raw & Post Processing, Printing => Colour Management => Topic started by: narikin on April 17, 2018, 06:22:35 pm

Title: Building profiles especially for shadow separation?
Post by: narikin on April 17, 2018, 06:22:35 pm
I'm struggling to get good shadow separation on prints, despite using great files, and decent manufacturer and/ or custom profiles.  Pictures of buildings in the sun look great in the light, but the shadow side is all plugged up, and I'm having to apply a massive shadow correction curve, to get anything out of it.

Is there an x-rite profile target or setup that could help with a custom profile especially focused towards the bottom end of the tonal scale? Wishful thinking maybe, but wanted to ask.

I use x-rite with an Isis XL, M0, d50. Fully color managed pipeline, with Epson P20000. Files are phase one IQ100, 16bit, low ISO, with great dynamic range. Spectacular on screen, but really plugged up in when printed. At least until a huge correction curve is applied.

At the moment 3,3,3 and 8,8,8 and 14,14,14 are indistinguishable in print. Any tips, or advice to get better shadow definition greatly appreciated!

I tend to keep the files in 16bit and just send that to the printer to work out. Is that a mistake? should I drop it to 8 in a controlled way in Photoshop?

Thanks
Title: Re: Building profiles especially for shadow separation?
Post by: digitaldog on April 17, 2018, 06:32:15 pm
You can't profile what isn't in the print. One issue is the linearity or lack thereof of the Epson driver. For example, with my 3880, RGB 0/0/0 maps to Lstar of 7.2.
If this is really important, maybe a 3rd party driver (what some incorrectly call a RIP in all cases) that has better behavior in the deep shadows.
Edit: Luster paper.
Title: Re: Building profiles especially for shadow separation?
Post by: Doug Gray on April 17, 2018, 07:04:39 pm
I'm struggling to get good shadow separation on prints, despite using great files, and decent manufacturer and/ or custom profiles.  Pictures of buildings in the sun look great in the light, but the shadow side is all plugged up, and I'm having to apply a massive shadow correction curve, to get anything out of it.

Is there an x-rite profile target or setup that could help with a custom profile especially focused towards the bottom end of the tonal scale? Wishful thinking maybe, but wanted to ask.

I use x-rite with an Isis XL, M0, d50. Fully color managed pipeline, with Epson P20000. Files are phase one IQ100, 16bit, low ISO, with great dynamic range. Spectacular on screen, but really plugged up in when printed. At least until a huge correction curve is applied.

At the moment 3,3,3 and 8,8,8 and 14,14,14 are indistinguishable in print. Any tips, or advice to get better shadow definition greatly appreciated!

I tend to keep the files in 16bit and just send that to the printer to work out. Is that a mistake? should I drop it to 8 in a controlled way in Photoshop?

Thanks

What colorspace are the images in? 14,14,14 maps to L* of between 4 and 5 with sRGB and ProPhoto and less than 2 for Adobe RGB. The smaller RGB numbers are even lower. You won't see anything printed in that range unless you use BPC with Rel. Col or Perceptual.

This is actually an area I'm working on. I also have an iSis with an old 9800. It's quite non-linear in shadows but I've made some 9800 specific targets that makes it perform quite well.
Title: Re: Building profiles especially for shadow separation?
Post by: Mark D Segal on April 17, 2018, 08:07:54 pm
I'm struggling to get good shadow separation on prints, despite using great files, and decent manufacturer and/ or custom profiles.  Pictures of buildings in the sun look great in the light, but the shadow side is all plugged up, and I'm having to apply a massive shadow correction curve, to get anything out of it.

Is there an x-rite profile target or setup that could help with a custom profile especially focused towards the bottom end of the tonal scale? Wishful thinking maybe, but wanted to ask.

I use x-rite with an Isis XL, M0, d50. Fully color managed pipeline, with Epson P20000. Files are phase one IQ100, 16bit, low ISO, with great dynamic range. Spectacular on screen, but really plugged up in when printed. At least until a huge correction curve is applied.

At the moment 3,3,3 and 8,8,8 and 14,14,14 are indistinguishable in print. Any tips, or advice to get better shadow definition greatly appreciated!

I tend to keep the files in 16bit and just send that to the printer to work out. Is that a mistake? should I drop it to 8 in a controlled way in Photoshop?

Thanks

The most important thing of all you haven't talked about is what paper you are printing on. If you are printing on a matte paper the lowest L* rating you are likely to obtain with the best of them is about L* 14. So anything below L*14 would indeed be indistinguishable as you have observed; as Doug mentioned, Relative Intent with BPC enabled would make the best of this, because it maps the Maximum Black of the image file to the Black point of the paper and produces a tonal separation range thereafter which should be quite linear if the profile is good. (I have done quite a bit of work on this problem and written up the results of my tools and testing procedures on this website, with more to come.) If, however, you are using a fine luster paper such as Epson Legacy Baryta, Ilford Gold Fibre Silk, or a number of others in the same family of papers you should be able to achieve good tonal separation from about L*2 and upward. Usual human visual perception of differences in black shading appearance begins to kick-in from about L*3 or L*4 onward.

There is one profiling arrangement you can try that may improve your shadow tone separation and that would be to get profiles made in M3 luminance condition (polarization). X-Rite profiling technology is not equipped to handle this. Barbieri is, but the equipment costs a mint. Chromix (https://www.chromix.com/colorvalet/print?-session=SessID:63F80A570a2f62994FJRo18B356C)  is equipped to make M3 profiles for customers, so I would suggest if you are prepared to invest $99 for making such a profile, get in touch with them, have one made and test it. I have tested a highly reputed RIP years ago when printing to matte paper and I didn't see any difference it made to the shadow detail issue that's worth writing home about, but things evolve over the years, so if you can download one of them in trial mode without committing money, that may be worthwhile, but I wouldn't hold my breath. Printer drivers are pretty darn good these days. The M3 route with Chromix may be more prospective for you.
Title: Re: Building profiles especially for shadow separation?
Post by: narikin on April 18, 2018, 09:07:31 am
What colorspace are the images in? 14,14,14 maps to L* of between 4 and 5 with sRGB and ProPhoto and less than 2 for Adobe RGB. The smaller RGB numbers are even lower. You won't see anything printed in that range unless you use BPC with Rel. Col or Perceptual.

This is actually an area I'm working on. I also have an iSis with an old 9800. It's quite non-linear in shadows but I've made some 9800 specific targets that makes it perform quite well.

Doug - thanks. I use ProPhoto (16bit) always.

Interesting that you made some specific targets to adapt to non-linear shadows, (due to printer driver, I presume?) that's kind of what I was wondering exists. If you think its translate-able to my setup, I'd love to try one. (plz PM) But if not, then no reason to waste your time. Thanks.

The most important thing of all you haven't talked about is what paper you are printing on.  [snip]

Chromix (https://www.chromix.com/colorvalet/print?-session=SessID:63F80A570a2f62994FJRo18B356C)  is equipped to make M3 profiles for customers, so I would suggest if you are prepared to invest $99 for making such a profile, get in touch with them, have one made and test it. Printer drivers are pretty darn good these days. The M3 route with Chromix may be more prospective for you.

Mark - apologies, should have mentioned, final prints are usually Legacy Platine, (or Canson Platine) a very good PK paper, as you know, so should get great shadows. I proof on Epson Premium Proofing Semi Matte, and yes, do see a good improvement in shadow tone, (and all colors) when I step up to the expensive stuff.

M3 polarisation profile sounds interesting. Thanks for this advice. I might well try that, worth the $99 investment, with 60" Platine paper at $320/roll. These prints are big at 60x80", so it costs me a lot every failure, and I'd be happy for a $99 buck investment to save just a couple of prints and its paid for itself.

You can't profile what isn't in the print. One issue is the linearity or lack thereof of the Epson driver. For example, with my 3880, RGB 0/0/0 maps to Lstar of 7.2.
If this is really important, maybe a 3rd party driver (what some incorrectly call a RIP in all cases) that has better behavior in the deep shadows.
Edit: Luster paper.

Andrew - thank you. "you can't profile what isn't in the print" is kind of what I was asking- if you flip that meaning around to mean: make & print a profile target that focus especially on shadow tones, print and profile that, to reveal and compensate for driver errors.  But I get your actual intended meaning, that it is near impossible to print anything like this with an Epson Driver that can't tell the difference. (Is Canon or HP any better?)

RIPs - I tried Imageprint, and found it a waste of time and money. I have QImage, but rarely use it, to be honest. Might it help? I don't own or licence any of the really expensive RIPs though.


Title: Re: Building profiles especially for shadow separation?
Post by: digitaldog on April 18, 2018, 11:32:23 am
Andrew - thank you. "you can't profile what isn't in the print" is kind of what I was asking- if you flip that meaning around to mean: make & print a profile target that focus especially on shadow tones, print and profile that, to reveal and compensate for driver errors.  But I get your actual intended meaning, that it is near impossible to print anything like this with an Epson Driver that can't tell the difference. (Is Canon or HP any better?)

RIPs - I tried Imageprint, and found it a waste of time and money. I have QImage, but rarely use it, to be honest. Might it help? I don't own or licence any of the really expensive RIPs though.
What I mean is, if the current driver blocks up shadows due to linearity issues, nothing will magically fix that other than better tonal separation via a differing driver setting or driver itself.
Title: Re: Building profiles especially for shadow separation?
Post by: Mark D Segal on April 18, 2018, 11:39:22 am
What I mean is, if the current driver blocks up shadows due to linearity issues, nothing will magically fix that other than better tonal separation via a differing driver setting or driver itself.

That makes sense Andrew, but I would think a primary task would be to analyze (a) whether there really is a linearity problem (seems likely in this case) and (b) if confirmed, whether it is caused by profile or driver. I don't know how an Epson 20000 driver performs, but my first inclination would be to suspect profiling and/or software settings rather than inherent issues with the driver/printer. The 20000 is designed for fine art printing amongst other things and includes 4 levels of gray inks. One would expect it to reproduce quite linear tonality very far down the L* range on luster/gloss media.
Title: Re: Building profiles especially for shadow separation?
Post by: Doug Gray on April 18, 2018, 11:58:38 am
That makes sense Andrew, but I would think a primary task would be to analyze (a) whether there really is a linearity problem (seems likely in this case) and (b) if confirmed, whether it is caused by profile or driver. I don't know how an Epson 20000 driver performs, but my first inclination would be to suspect profiling and/or software settings rather than inherent issues with the driver/printer. The 20000 is designed for fine art printing amongst other things and includes 4 levels of gray inks. One would expect it to reproduce quite linear tonality very far down the L* range on luster/gloss media.

I was able to significantly improve the I1P profiles by  creating an iSis target specific to the 9800. The 9800 is very nonlinear near L*=8 and L*=85. The default profile targets, and even larger ones such as BAs or custom patch sizes up to 5k didn't sufficiently deal with these 9800 anomalies.

OTOH, if the P20000's driver is well behaved there might not be as much to gain. It needs to be characterized.

Because the iSis produces very consistent readings at a reasonable speed, it should be a fairly straightforward process. Narakin has both the printer and an iSis XL so we shall find out. I've PM'ed him.
Title: Re: Building profiles especially for shadow separation?
Post by: digitaldog on April 18, 2018, 12:00:19 pm
One can try differing media settings within the Epson driver but they only go so far. This may help in visually examining if the settings do anything useful:
http://digitaldog.net/files/InkDensity.zip (http://digitaldog.net/files/InkDensity.zip)
Title: Re: Building profiles especially for shadow separation?
Post by: nirpat89 on April 18, 2018, 12:38:48 pm
I was wondering instead of searching for that perfect profile, why not add a correction curve in your final image.  Digital negative makers (like myself) routinely calibrate their processes by such correction curves to map a linear full tonal range in the image to the shortened non linear range of something like pt/pd or salt printing chemistry.  They particularly take care of the clumped up shadows in salt prints due to 'self-masking" effect. 

One can use the soft-proofing function to guess a good curve that will open-up those shadows using an appropriate image like a step wedge in the area of interest. 


:Niranjan.
Title: Re: Building profiles especially for shadow separation?
Post by: smilem on May 31, 2018, 02:01:07 am
Tremendous help would be the possibility to linearize the printer each ink channel separately. This could be done using PrintFAB driver for Windows if they added the linearize feature I was testing in 2011.

Without linearization and proper ink limits no wonder shadows are plugged. On printers like Canon Pro 9000, 6700D etc. if you use third party ink then the best you get on any paper is about L20 all because the ink curves are wrong.
Title: Re: Building profiles especially for shadow separation?
Post by: arobinson7547 on June 01, 2018, 08:29:54 am
I think smilem nailed it - pure and simple and 100% on the mark. I 'could' only add that the individual channels being ink limited and linearized is a first and very necessary step. but RIGHT AFTER that is the Total Ink Limit of ALL the individual channels, printed together.

At a point, as that ink is laid on the Paper. Two things are going to happen (or 'are' happening at the same time) The ink is going get 'as dark as it's GOING to get; (max density) AND usually before that point is reached the ability to distinguish the detail between steps is going to diminish and eventually disappear. (Even as the density is still increasing on a Matte Stock, for example, the ink load will cause finer detail to 'fill in' and be lost.

My conclusion is the ideal circumstances is Ink limited and linearized channels and a Total Ink Limited that is raised enough from the Bottom (maximum density) to retain fine detail/separation. So you have to sacrifice density for detail (someone wearing a black jacket with a pattern; Maximum density make the jacket dark and rich but losses the detail. Raising the Bottom, lightens the shadow BUT retains and brings out the texture/detail of the Black Jacket)


At least, that the way "I" see it.


Speak Now or forever hold your peace <smile>
Title: Re: Building profiles especially for shadow separation?
Post by: narikin on June 01, 2018, 09:05:35 am
I was wondering instead of searching for that perfect profile, why not add a correction curve in your final image.  Digital negative makers (like myself) routinely calibrate their processes by such correction curves to map a linear full tonal range in the image to the shortened non linear range of something like pt/pd or salt printing chemistry.  They particularly take care of the clumped up shadows in salt prints due to 'self-masking" effect. 

One can use the soft-proofing function to guess a good curve that will open-up those shadows using an appropriate image like a step wedge in the area of interest. 


:Niranjan.

Niranjan,  I sort of overlooked your comment after reading the first line suggesting applying a profiling curve, which, of course I do. (You have to as otherwise its totally plugged up) But it was unfair of me not to read through more thoroughly, and your suggestion of creating a step wedge and print that through various adjusted curves and see where it settles. Good methodology, thanks, I'll give it a try.

My fear is that the curve is so extreme to get shadow definition, that the deepest black (dmax) is lifted to the point of the image loosing its impact.  Yes it would be lovely to be able to see a clear difference between 0,0,0 and 5,5,5, and 10,10,10, etc. but it seems shadow differentiation start at around 12 to 15 on my Epson, even with very good custom profiles on the best PK papers.

Title: Re: Building profiles especially for shadow separation?
Post by: arobinson7547 on June 01, 2018, 10:09:53 am
Niranjan,  I sort of overlooked your comment after reading the first line suggesting applying a profiling curve, which, of course I do. (You have to as otherwise its totally plugged up) But it was unfair of me not to read through more thoroughly, and your suggestion of creating a step wedge and print that through various adjusted curves and see where it settles. Good methodology, thanks, I'll give it a try.

My fear is that the curve is so extreme to get shadow definition, that the deepest black (dmax) is lifted to the point of the image loosing its impact.  Yes it would be lovely to be able to see a clear difference between 0,0,0 and 5,5,5, and 10,10,10, etc. but it seems shadow differentiation start at around 12 to 15 on my Epson, even with very good custom profiles on the best PK papers.

"That" is the trade off; Density vs Detail. In the end I choose Detail. The dark black blog vs lighter area with something to see and draw interest from the observer.
Title: Re: Building profiles especially for shadow separation?
Post by: nirpat89 on June 03, 2018, 06:39:06 pm
Niranjan,  I sort of overlooked your comment after reading the first line suggesting applying a profiling curve, which, of course I do. (You have to as otherwise its totally plugged up) But it was unfair of me not to read through more thoroughly, and your suggestion of creating a step wedge and print that through various adjusted curves and see where it settles. Good methodology, thanks, I'll give it a try.

My fear is that the curve is so extreme to get shadow definition, that the deepest black (dmax) is lifted to the point of the image loosing its impact.  Yes it would be lovely to be able to see a clear difference between 0,0,0 and 5,5,5, and 10,10,10, etc. but it seems shadow differentiation start at around 12 to 15 on my Epson, even with very good custom profiles on the best PK papers.

I guess we are even...I somehow also missed that you did say you had already used a shadow correction curve in your first post.  In any case, here is a test pattern that I like to use to probe the capability of my digital negatives to print shadows and highlights, if it is useful.

:Niranjan
Title: Re: Building profiles especially for shadow separation?
Post by: Tim Lookingbill on June 03, 2018, 10:59:17 pm
I guess we are even...I somehow also missed that you did say you had already used a shadow correction curve in your first post.  In any case, here is a test pattern that I like to use to probe the capability of my digital negatives to print shadows and highlights, if it is useful.

:Niranjan

Just FYI but that target you posted doesn't have an embedded profile. To see how important it is to know from which space it was created drag/dropped to your desktop from the browser and open in Photoshop and assign sRGB and then AdobeRGB and note the difference between 13RGB vs 000RGB black. 

And on the subject of knowing for sure if your printer is laying down the blackest black it can muster print a black patch on clear inkjet film (Epson used sell such film) and view it backlit by a bright light and see if there is show through. If you can't see the shape of the light through the black patch, then that black is as dense as it is going to get and thus is not the problem of your printer.

More than likely it's the amount of absorption the paper is subjecting to that black ink that the profile creation is adjusting upward to maintain linearity and smooth ramping of shadow detail out of black no matter how dense it looks on paper.