Why OBAs in this paper?

[jrsforums]

I'm working on an Aardenburg light fade database II which will integrate both light fade and dark storage combined results. It's definitely a work in progress, but once the dark storage yellowing issues are integrated into the overall test result, the current Epson Ultra Premium luster and glossy papers are going to rank among the worst RC inkjet photo papers on the market. Canon RC photo papers (Luster and Semigloss pro, but not Platinum) will have about half of the light-induced post exposure staining as the Epson papers.

Are the manufacturers working on improvement? Hell no! The customer can't see the problem when looking at initial image quality, so the customers aren't complaining...yet! But once the yellowing problems begin to be talked about in forums like LULA's printers and printing forum, I'm optimistic that media longevity improvements will be made. Epson's Proofing Paper White Semimatte is a good example of a low OBA resin coated (RC)  paper exhibiting far less yellowing/discoloration over time. That said, EPPWSM does not have a typical "photo" surface texture that will appeal to all RC photo paper enthusiasts, but it would be easy enough for Epson to specify an "Epson Premium Luster Natural" RC photo paper with all of the good longevity characteristics of Epson Proofing Paper White Semimatte. The printmaking community just has to insist that the manufacturers do better!

cheers,
Mark
http://www.aardenburg-imaging.com

Other than the Canon’s, are there any others that have better characteristics which you could share with us?

And what time frame are we looking at for the degradation?

[MHMG]

Other than the Canon’s, are there any others that have better characteristics which you could share with us?

And what time frame are we looking at for the degradation?

20 Megalux hours of light exposure is more than enough exposure to reveal OBA light fading and post-exposure dark storage issues, albeit higher exposure doses increase the magnitude of the problem. 20 Mlux hours equals twenty years on display at an average display illumination level of 228 lux for 12 hours per day, or just 10 years on display if using the Wilhelm extrapolation assumption for light levels at 450 lux for 12 hours per day.  Maybe that's good enough for some folks, but why use a high quality pigmented inkset that can be fade resistant for 100 Mlux hours on display or more if the paper is going to let you down in 20 Mluxhrs? That's the crux of it, pure and simple, i.e., matching a quality paper with a quality ink set. Essentially none of the RC photo media on the market today have 100 MLux hour robustness, or even 50 Mlux hours for that matter, but some show twice as much discoloration or even more at a given exposure dose than others, e.g., various Epson versus Canon OEM brands.  What more can I say?

[rasworth]

Educating me about OBAs is tough enough, the rest of the world might be more of a challenge.  Red River's one pager on OBAs (first attachment) is a masterpiece of issue minimization.  They do provide alternatives if one is really committed (second attachment).  They have two rc papers represented as having low OBA content, I was able to get a sheet of the Palo Duro Satin and measure.  It squeaks by per Mark A.'s definition of low OBA content:
  M0 l*a*b* = 95.8, -0.2, -3.0
  M2 l*a*b* = 95.2, -0.7, -0.7 (delta b* = 2.3)

My black light shows Palo Duro Satin to be slightly more florescent than Ilford Gold Fibre Silk.  Given its parameters I assume it's a reasonable candidate for a low OBA rc (read inexpensive) photo paper.  I'll crank out a profile, see what I think.

Thanks again to all who have contributed, it's certainly been a fun/useful thread for me.

Richard Southworth

[Doug Gray]

The basic problem with OBA laden papers is that it will look different depending on the uV content of whatever illuminant you choose to display or show the print. Change it and the print can appear different. There are other issues as well such as metameric shift which is especially a problem for cheap LED and fluorescent lighting but one thing at a time.

Whether profiles using M2 are "true" or not depends on whether the illuminant the paper is viewed under has uV or not. If not, it IS true. The problem is determining whether the light you are viewing the print has uV, and if it only has some, which a lot of regular incandescents do, whether M2, M1, or M0 is the closest match to the level of uV from it.

There is a fairly simple say to determine this with OBA papers.

Measure the Lab value of the white patch on your colorchecker card.

Create a patch in Photoshop and it has to be in Photoshop or some other program that allows Abs. Col. printing. Fill the patch with Lab=(Colorchecker white patch). Then print the patch using each of the three M0, M1, and M2 profiles made for the paper using Absolute Colorimetric. It's easy to print these on the same sheet by positioning them differently across the paper's top and making 3 prints on the same sheet.

Now, cut the patches off so you don't see any surround and have three patches. Good idea to mark the patches with M numbers to keep track.

View these patches against the Colorchecker under the illuminant of interest (home lighting, display lighting, outdoor lighting, etc).

Then just use the Mn Profile that most closely matches.

I did this test with a glossy paper that shows a b* of -12 with M1 (D50) and checked the 3 patches printed with M0, M1, and M2 against a variety of lights including 5000K 95 CRI LEDs, Solux 4700K, GMB (Judge II Fluor 5000K (F8 I think), and industrial CWF), various household incandescents, a CFL 92 CRI 5000K I use for scene photog) and assorted 2800K room LEDs with poor CRIs (80 ish). I also checked in outside in daylight and through a North facing window.

In all non-natural light cases except the Solux, the M2 profile was the closest match. All the LED lights easily had the best match with the M2 profile. The incandescents (except the Solux) and fluorescents were between the M2 and M0 but closer to M2.

The North facing window gave the best match to M0. And daylight outside gave the best match to M1 as expected.

Another way to look at it is that the OBAs added to paper don't do a hell of a lot indoors and they do squat for LEDs which are becoming ubiquitous. The better ones exceed good fluorescents spectrally. OTOH, if you are outside or even near a window OBAs can shift your paper color well to the blue/violet and you should use M1 or M0 profiles for prints displayed where window light dominates or outside.

So M2 is my default unless there is some compelling reason to choose M1 or M0.  Better yet, I avoid OBA paper as much as possible. Non OBA papers produce the same profiles under M0, M1, or M2.

[jrsforums]

I did this test with a glossy paper that shows a b* of -12 with M1 (D50) and checked the 3 patches printed with M0, M1, and M2 against a variety of lights including 5000K 95 CRI LEDs, Solux 4700K, GMB (Judge II Fluor 5000K (F8 I think), and industrial CWF), various household incandescents, a CFL 92 CRI 5000K I use for scene photog) and assorted 2800K room LEDs with poor CRIs (80 ish). I also checked in outside in daylight and through a North facing window.

In all non-natural light cases except the Solux, the M2 profile was the closest match. All the LED lights easily had the best match with the M2 profile. The incandescents (except the Solux) and fluorescents were between the M2 and M0 but closer to M2.

The North facing window gave the best match to M0. And daylight outside gave the best match to M1 as expected.

Another way to look at it is that the OBAs added to paper don't do a hell of a lot indoors and they do squat for LEDs which are becoming ubiquitous. The better ones exceed good fluorescents spectrally. OTOH, if you are outside or even near a window OBAs can shift your paper color well to the blue/violet and you should use M1 or M0 profiles for prints displayed where window light dominates or outside.

So M2 is my default unless there is some compelling reason to choose M1 or M0.  Better yet, I avoid OBA paper as much as possible. Non OBA papers produce the same profiles under M0, M1, or M2.

I believe, if I remember correctly, you said, in another thread, that other intents would not show the same differences as AbsCol.  Is that correct?

[Doug Gray]

I believe, if I remember correctly, you said, in another thread, that other intents would not show the same differences as AbsCol.  Is that correct?

That is correct. Abs. Col. will print the specific color requested as long as it is within the printable gamut. So if you specify Lab=50,0,0, you will get a neutral gray that is almost identical to the old 18% neutral card. It will match the D4 patch on a colorchecker.

If you print it using Abs. Col. and an M1 profile it will match outside. If you print it using M2 it will match under LED illumination.

However, when you print a neutral color using Rel. or Perc. intents it scales to the media white point. So a LAB=100,0,0 will look the same indoors or outdoors using profiles made with M0, M1, or M2. It's unprinted paper.

However, once you start printing colors the M's start to matter but they don't matter nearly as much as printing Abs. Col. For instance if you print neutral colors using Rel. Col. (which I'll refer to since its behavior is specified) then the neutral colors should, while in gamut (above the black point) maintain the same xy coordinate in the CIEXY gamut. This happens with OBA papers but only if illuminated with the uV amount used in M0 or M1 and if those profiles are used to print the neutral colors.

When there is a mismatch between the Illuminant's uV and the Mn used to create the profile the tracking diverges in Rel. Col and Perc. But not a lot. In my limited testing the divergence on neutrals is no more than 2 dE with a paper that shows a 12 dE difference at the white point from M1 to M2. This difference is due to the ink/substrate absorbing uV at a fraction somewhat different than it absorbs visible light. If it was identical there would be no difference at all between M1 and M2 profiles when printing Rel. Col.

The differences are greater with some non-neutral colors. In the 3 papers I looked at the differences were only over 4 dE on a small fraction of colors. Still, the effect of M1-M2 is, of course, much reduced in comparison to Abs. Col.

[rasworth]

Doug,
Thank you for eloquently stating what I have been nibbling at with my posts, i.e. M2 based profiles on rc papers with OBAs are "valid" for producing prints to be viewed in a low uv environment.  We all would probably go for no OBA media if always practical - I needed a reasonable cost 17"x25" paper, and the best choice was the Red River UltraPro Luster 300.

Richard Southworth

[smthopr]

Thanks everyone for this discussion.  I've just begun making my own printer profiles and did not really understand the whole "M" thing.  But now, I do!!!!

Now I just need to find the brightest, whitest, glossyish photo paper that is has very low or no OBAs...   The paper I have now, Innova Ultra Smooth Gloss, seems to have quite a bit of OBAs and it looks quite different in each lighting situation.  It really "pops" in daylight, where virtually no body will every view the prints...  And, I'm quite sure that they will never be viewed in the standard D50 illuminant anywhere...

[digitaldog]

The absolute colorimetric rendering intent reproduces the exact color that existed in the source—absolutely. If the source was light color on the dingy yellow-white of newsprint, the resulting color on your brilliant coated ink jet paper will be dingy yellow. This rendering intent is really designed for making one device simulate the appearance of another device for use in proofing. So otherwise, skip using it!


The perceptual rendering intent transforms the colors so that the image in the destination space is perceived in the same way as the original. The conversions are weighted to deal with luminance over saturation and hue because our eyes will notice differences in luminance far more than differences in saturation or hue. Luminance information provides shape and detail; this is the most important factor to perception.

[jrsforums]

Andrew, appreciate the statements on Absolute and perceptual.  Can you provide similar on RelCol?

[digitaldog]

Andrew, appreciate the statements on Absolute and perceptual.  Can you provide similar on RelCol?

RelCol is identical to Absolute with the exception of mapping of white! They share the same table.
Also, there are no rules in how a profile maker produces a perceptual rendering unlike Colorimetric (but this isn't a perfect world so you may indeed see differences). Think of Perceptual like transparency film; Agfa, Fuji, Kodak all produce different renderings based on what they believe their customers will prefer.

[Doug Gray]

Thought it might be visually useful to show the effects of M1 (full D50 w uV) and M2 (D50, uV cut) profiles printing a neutral tone curve from L=5 to L=100 with a high OBA paper.

This is Costco Glossy paper which has a large amount of OBAs. The paper white under M2 is LAB (95,-1,-2) while under M1 it is LAB(95,1,-12) for a b* shift of -10. For comparison Epson Prem. Glossy only has a b* shift of -2 which comes from the fluorescence of the substrate, not OBAs in the coating.

The 4 images are as seen under D50 without uV. The top-left image(1) is Rel. Col. using an M2 Profile. The bottom-left image(2) is also Rel. Col. but using an M1 Profile made with full D50 uV. Notice how close they are to each other in spite of the large difference in M1/2 white points from OBAs. But if looked at closely, or sampled with the eyedropper, the bottom neutral curve is slightly bluer by 1-2 dE.

Basically, using Rel. Col, (or Perceptual) there is not much difference between using a M1 or M0 profile and an M2 profile to print. This is because these intents are scaled off the media's white point so everything is adjusted to LAB(100,0,0) being mapped to that white point.

However.Abs. Col. intent is another matter entirely. If you print LAB(50,0,0) then you should measure LAB(50,0,0) within system tolerances. If the color can't be printed, as is the case when you ask to print LAB(100,0,0), it will print the closest color and it will do so based on being illuminated with D50 (uv cut - M2) or (full uV D50 - M1). Since the uV increases the spectral response at the bluer wavelengths, attempting to print LAB(100,0,0) will lay down significant yellow ink to bring the b* towards 0 and counteract the uV induced blues.

The top-right image(3) is how the image would print using the M2 profile. The printed tone curve is neutral until the gamut edge is reached (L*=95).  This is because the profile was made with M2 (uV cut D50).  It's slightly warmer than the images on the left because it is a completely neutral gray while the Rel. Col. images scale from media white which is very slightly bluish for the Costco glossy under uV cut D50.

The bottom-right image, made with a M1 profile,  requires much more yellow ink to bring b* from -12 to 0 because this printed image is designed to be viewed in a fully compliant D50 booth or outdoors, under daylight, with high levels of uV. The result is that it looks pretty awful with no uV. But it would look fine outside. It also measures neutral using an M1 spectrophotometer.

[jrsforums]

Doug, you have provide a series of great info, along with support bits from Andrew.

You might want to pull it all together and see if Kevin would publish.  I am sure it would be valuable to many.

[digitaldog]

However.Abs. Col. intent is another matter entirely. If you print LAB(50,0,0) ...

Hold on. Why would anyone use Absolute Colorimetric for a print? I know why I'd use it; for cross rendering. Making one print match another printed process. Make an Epson match a press sheet or contract proof. That's why the table option was created. So you'd be hard pressed to use two very dissimilar papers with or without OBAs. If you examine Proofing papers (like Epson's), very low if any OBAs, not very white and designed for matching another print process. So I don't see why anyone short of cross rendering would use Absolute rather than Relative Colorimetric. Now if your goal is an output specific paper profile based on the viewing conditions, you would (should) measure the illuminant, save a CxF file and load that into the products that correctly allow for this to be used instead of D50 which is an assumption. We've had this capability since ProfileMaker Pro days and it works quite well.
Assuming the above is correct, isn't bringing Absolute Colorimetric into the discussion of OBAs a bit like talking of how many ICC Profiles can dance on a pin? When or why would a Costco customer consider using this RI?

[digitaldog]

http://dba.med.sc.edu/price/irf/Adobe_tg/manage/renderintent.html

Absolute Colorimetric
Colors match exactly with no adjustment made for white point or black point that would alter the image's brightness. Absolute colorimetric is valuable for rendering "signature colors", those colors that are highly identified with a commercial product such as the yellow used by the Eastman Kodak Company, or the red used by the Coca-Cola Company.


Interesting that Lightroom has no such option for any output.

[Doug Gray]

http://dba.med.sc.edu/price/irf/Adobe_tg/manage/renderintent.html

Absolute Colorimetric
Colors match exactly with no adjustment made for white point or black point that would alter the image's brightness. Absolute colorimetric is valuable for rendering "signature colors", those colors that are highly identified with a commercial product such as the yellow used by the Eastman Kodak Company, or the red used by the Coca-Cola Company.

Good old Abs. Useful when I was matching paint colors on some outside doors that were being replaced. Rarely, arguably never, useful for normal photographic printing. One of the few places it comes in handy, aside from pre-press hard proofing which you noted, is repro work but that is hardly a normal activity.

Interesting that Lightroom has no such option for any output.

And not surprising given what Lightroom is used for. Might as well leave it out. OTOH, Lightroom bakes in BPC when selecting Rel. Col. and that causes an unnecessary lightening of the entire image even if the image doesn't have blacks below the media's black point. They really should have an option for that since a significant fraction of photographic images don't have the deep blacks that benefit from BPC and they shouldn't have to suffer to use Rel. Col.

[Doug Gray]

Hold on. Why would anyone use Absolute Colorimetric for a print? I know why I'd use it; for cross rendering. Making one print match another printed process. Make an Epson match a press sheet or contract proof. That's why the table option was created. So you'd be hard pressed to use two very dissimilar papers with or without OBAs. If you examine Proofing papers (like Epson's), very low if any OBAs, not very white and designed for matching another print process. So I don't see why anyone short of cross rendering would use Absolute rather than Relative Colorimetric.

I pretty much agree Andrew, though there are a few applications beyond cross-rendering, where Abs. Col. is useful. For instance matching colors when doing remodeling More seriously, Repro work where you are trying to make as precise a match as possible to an original print or artwork, including fading and yellowing that may have occurred for older pieces.  But generally, yes, there is rarely a need for Abs. Col. People that have to ask whether they should use Abs. Col. or not probably shouldn't. If you need to use it, and understand what it does, then it's there. At least in some apps like Photoshop.

Now if your goal is an output specific paper profile based on the viewing conditions, you would (should) measure the illuminant, save a CxF file and load that into the products that correctly allow for this to be used instead of D50 which is an assumption. We've had this capability since ProfileMaker Pro days and it works quite well.

Yes, quite a nice feature. One can even make prints that look good under industrial fluorescent light. And useful for all Intents.

Assuming the above is correct, isn't bringing Absolute Colorimetric into the discussion of OBAs a bit like talking of how many ICC Profiles can dance on a pin? When or why would a Costco customer consider using this RI?

I think it's useful in the sense that what I've been pointing out is just how small the difference is when printing images in Rel. Col, Perc., or Sat. intents between M0, M1, and M2 profiles.  You get only the smallest differences and they are very hard to see in prints. Even those with huge amounts of OBAs.

But boy do OBAs show up using Abs. Col. That's a lot of the reason for the recent standardization using full D50 with uV for hard proof viewing booths and combining that with M1 profiles. But that's also the special case where Abs. Col. is used so syncing up M1 with full uV D50 addressed the OBA issue. Not that many still don't bother.

[digitaldog]

But boy do OBAs show up using Abs. Col.

They do, indicating:
If you need such an RI for what it is useful for in rare cases, don't mess with OBAs. 

[Doug Gray]

They do, indicating:
If you need such an RI for what it is useful for in rare cases, don't mess with OBAs. 

I agree that avoiding OBAs is highly desirable but for possibly different reasons.

First, it is not because prints using "normal" intents with different M profiles look different. They rarely do. At most the differences are very subtle and far less than the differences between, say, Perceptual and Relative.

However, I'm concerned that OBAs will yellow quicker than paper without them but I'm no expert in that area and defer to Mark A.

OBAs also screw up view proofing because Abs. Col. is used behind the scenes on all Intents when selecting "show paper color."  Getting correct view proofing when selecting that requires that your viewing setup have the correct amount of uV matching the M used and that's difficult w/o spending serious $.  Avoiding OBAs gets rid of that problem and view proof works fine. A workaround, which I try to avoid, is not selecting show paper color and using a specific setup for each paper so the hard print viewing whites match the soft proof for each paper. That's relatively easy to do with things like ColorNavigator but becomes unnecessary if you aren't dealing with OBAs.