Luminous Landscape Forum
Raw & Post Processing, Printing => Printing: Printers, Papers and Inks => Topic started by: rasworth on June 01, 2017, 08:42:35 pm
-
This thread is an offshoot of a previous thread discussing matte papers in general.
I emailed RR and requested a sample of their new Palo Duro Etching paper, received two sheets a couple days later. I decided with two sheets the best quantitative evaluation I could do was create profiles on my Epson 3880. I had previously downloaded the RR profile, was quite surprised after checking it in ColorThink to find it listed a black point of L*a*b* = 2.34,.37,.22, and a total gamut of 774,930. These would be very good numbers for any pk based paper, and fantastic for mk papers.
My 3880 was in Photo Black mode, so I first did a 405 patch (one sheet) profile, using i1Profiler and an i1Pro (non-uv), thinking if the paper were that good with mk black it should come out ok with pk. Wrong, the 0-0-0 patch measured L*a*b*=23.30,.47,2.03. Otherwise the profile appeared proper, smooth gamut volume and fairly congruent neutral rendering curves. The gamut volume read out as 483,902, much smaller than RR's profile.
Back to the printer, switch to mk ink, print out another target. I used Velvet Fine Art as the selected media, per the RR profile instructions, and Quality = 5. The results were better, L*a*b* = 18.24,.90,1.72, and gamut volume 497,603. However, my results were a long way from the RR profile numbers. Again, my profile checked out wrt to other parameters, and looks good doing soft-proof in Photoshop CC, although the high black level detracts.
I'm out of the paper, can't do anything else at this point. I plan to contact Red River support tomorrow and relay my findings. It's possible they sent me the wrong paper, they don't have pre-packaged sample packs yet, so it's too early to criticize. The RR profile copyright indicated Chromix, ICC version 2.2.0. I also created version 2 profiles. I have cranked out several profiles recently, all for pk papers for myself and others, with good results, so I don't think my process is flawed.
I hope there are others doing similar testing, and that they will contribute shortly.
Richard Southworth
-
Richard, I'm not sure why you printed using PK inks as RR says to use MK with Velvet Fine Art setting. No surprise that your results were worse than the recommended settings.
I ordered a box of letter size sheets and it's in transit. According to the tracking, it should be here today. I'm going to prepare a standard profile using my normal patch set with ArgylCMS. I'll compare that to the William Turner profile that I've been using for several years and also report back on the black point measurement. I will also print out the Outback test photo image and see what that looks like.
I should have a report early next week.
Alan
-
Alan,
My printer was initially in photo black mode, so I started with pk ink, using Premium Semigloss as the media. As I reported, the results were bad, and I switched my 3880 to mk ink, chose Velvet Fine Art (per the RR instructions) and printed a second target. The mk profile, which should be directly comparable to the RR profile, indicated a black level of L* = 18.
So yes, I am fully aware that mk ink is the proper choice. I only created a pk ink profile because it was convenient to start in that mode, and I had some reason to believe it might result in a decent black point, given the RR profile parameters.
Again, as reported in my first post, I ended up creating a mk ink based profile, to be apples and apples with the RR profile.
Richard Southworth
-
Richard,
I just checked my data for William Turner and L* is 18 so it would not be surprising if the RR paper were the same.
-
Alan,
It would not be surprising, except that RR has made a big deal out the paper's dense blacks, describing a barrier layer between the ink receptive coating and the substrate that is supposed to keep more of the black ink near the surface. And of course there is the matter of their canned profile and its L*=2.34.
Richard Southworth
-
Alan,
It would not be surprising, except that RR has made a big deal out the paper's dense blacks, describing a barrier layer between the ink receptive coating and the substrate that is supposed to keep more of the black ink near the surface. And of course there is the matter of their canned profile and its L*=2.34.
Richard Southworth
We'll see what my measurement shows. I can tell you that both Ilford Gold Fibre Silk and Museo Silver Rag, the two PK papers that I regularly print on have L* = 5. This data as the William Turner data comes from the 0,0,0 patch in the Argyll patch set.
-
My data is also from the 0-0-0 patch, and my custom profile for Ilford Gold Silk shows L*=4.4, i.e. same as yours. Look forward to seeing your measurements on the RR Palo Duro Etching paper. I have an email into the RR support group asking for clarification. Their profile does not include the data set, so using the black point tag for comparison.
Richard Southworth
-
Alan,
It would not be surprising, except that RR has made a big deal out the paper's dense blacks, describing a barrier layer between the ink receptive coating and the substrate that is supposed to keep more of the black ink near the surface. And of course there is the matter of their canned profile and its L*=2.34.
Richard Southworth
No fine art matte paper is going to get anywhere close to L*= 2.34. It would defy the laws of physics for matte surface scattering. A semimatte RC photo paper like Epson Proofing paper white semimatte can reach L*=5 or or 6 with well matched inkset, but RC media does so because the PE layer is a good smoothing sublayer and is an ultimate barrier layer to further ink penetration.
L*=18 is very typical for matte papers and good ink/media combinations. L*=16 occurs sometimes using the right papers (e.g. Canson rag photographique) in Epson ABW B&W printing mode and on Canon and HP printers as well with the right choices of paper. L*=14 is best L*min I have ever measured from an inkjet printer printing with a highly tuned monochrome ink set using Roy Harrington's Quadtone RIP as the driver in combination with a specific matte fine art paper.
A profile editing package can, however, be used to edit the LUTS in the profile, thus creating a softproof that looks better, but the LAB numbers will no longer match between the forward and inverse transforms. The inverse transform has been "cooked" to give a better softproofing experience, and photoshop's info tool will display the edited numbers. That's what may be going on with the aforesaid L*min=2.34 canned profile for this new paper.
Lastly, Epson made significant claims for improved Dmax with its newest ink HD ink set, and worked with it's media suppliers on its newer Legacy Paper line to ensure that result. This is where some of that "new" enhanced coating technology came about, IMHO, and RR's new etching paper may indeed have tapped into this new refined coating technology from the same source where Epson gets it, if Epson doesn't have exclusivity to the latest tech. It does give improved performance with some media/ink combinations, but not all. It's very ink/media dependent. Use other high quality third party papers with the new HD ink set, for example, and I've actually encountered results that were slightly worse compared to the older K3 ink set. HP and Canon OEM pigmented inks also have some "up and down" performance which is media dependent, but on average, HP's combined used of both PK and MK inks in it's Z series printer driver when printing on matte fine art media still yields best-in-class results on a broader range of matte media, IMHO, beaten only by specialty B&W monochrome ink sets that have been carefully tuned to the media.
cheers,
Mark
http://www.aardenburg-imaging.com
Edit: Addenda to comments above. I just visited the RR website, and RR has published LAB values for a few different OEM inks on this media, all of which have very low L*min values, and all of which are claimed to be measured directly with a spectrophotometer. I'd wager good money the spectrophotometer that was used is out of calibration. An aged bulb in some spectrophotometers will generate increasingly erroneous readings for L* value as L* min values are approached yet the instrument will appear to be working properly. The subsequent illumination deficiency errors get spread evenly over the whole L* range of the measurment target such that an ICC profile made from these systematic errors will still appear to work satisfactorily in the forward conversion to print. Nonetheless, the measured values are wrong!
-
I think they profiled the wrong paper, per my inspection of the profile (see attached screen print). I have a support ticket with Red River, we'll see how they respond.
Richard Southworth
-
Epson Hot Press Natural paper, Epson 3880 printer, QTR, Piezography K7 with Jon Cone's new Ultra HD matte black ink, Dmax= 1.8, L*= 13.
-
My box of paper arrived yesterday but then I found I was out of Vivid Light Magenta and that was the one color that I didn't have a replacement for! >:( I immediately placed an order with Atlex but the cartridge won't be arriving until mid-week so my tests will have to wait. I like Mark's thought about an spectro that has gone bad. This makes the most sense to me.
-
At least Red River appears to be consistent. I downloaded three more matte profiles, two for my Epson 3880 and one for a Canon - all three indicate a black point of L*=3. I used one of their profiles in Photoshop to soft proof the Outback test image, colors look ok but of course the shadows are much blacker than using one of my RR custom profiles.
I presume one could print with their profile, obtain reasonable results, and never know that the soft proof was a hoax, at least wrt black levels. Maybe Mark's earlier theory about deliberately editing the profiles to improve the soft proof is correct. It's been a few years, but I remember using Profilemaker Editor to mess around with such.
Richard Southworth
-
More speculation - manipulating a profile in this manner would provide more shadow differentiation in relative rendering mode. Black point compensation should do the same, but perhaps RR has received complaints from less sophisticated users.
I know, getting way out in left field.
Richard Southworth
-
My goodness! It’s a great big crazy world of color out there isn’t it? This is exactly what’s good about forums like this. You get to bring up a topic and have it viewed from many different perspectives. Hopefully this kind of dialog can be very illuminating.
I’ve worked for CHROMiX over 10 years now, and that whole time I’ve watched Steve Upton pounding the drum for polarized measurements at every chance he gets. He does not keep it a secret. We have hammered on X-Rite to put polarizing filters in more of their instruments. We had Konica Minolta personnel from Japan in our offices just last month, and he was selling the virtues of polarized measurements to the actual instrument makers. We’ve brought this topic up in trade shows, conferences - we’ve mentioned it in our blog, in our newsletter. So from the profile producer’s point of view, it’s rather frustrating that knowledgeable people in the industry still don’t seem to know about or consider polarized measurements.
http://blog.chromix.com/2010/04/
Certain media types benefit from polarized measurements because the polarization filter helps the light from the surface enter the spectrophotometer at a more direct angle. Canvas and other fabrics benefit from this treatment a lot, but also matte papers and even flat papers from toner-based printers. These kinds of media are notorious for having blacks and shadows that get “blocked up” in the shadows. This kind of measurement treatment will produce profiles that give more shadow detail than you’d otherwise see. And this additional shadow detail is built right into the profile, so you receive the benefit of it whether or not you have an app that does black point compensation, and regardless of rendering intent. A by-product of taking polarized measurements is that the darker patches actually measure darker, and the profile reflects those darker measurements. This has no effect on the overall darkness of the printed result.
The typical i1Pro can do a decent job of producing good profiles, but people come to us for profiles when they want to get that last bit of accuracy and quality that make their prints the best they can be. These matte-surfaced profiles for example are not made to satisfy those who want them to look correct in ColorThink, but are made to produce the best prints they possibly can. Soft-proofing with these profiles will naturally show a slightly more saturated preview, but it is a trade-off that most people using this troublesome media gladly prefer, and that’s why we go in that direction.
I like to geek out as well as the next guy, but at thirteen posts we might all want to stop counting numbers, take a step back, take a deep breath. I’d like to hear from someone who has actually printed with one of these profiles?
-
A by-product of taking polarized measurements is that the darker patches actually measure darker, and the profile reflects those darker measurements. This has no effect on the overall darkness of the printed result.
Is there published data on this? Is there a comparison between matte and gloss paper as to the effect of the polarizing filter? I would have to go back to my physics books to try to figure out whether there should be a difference or not (something that I was better prepared to do fifty years ago).
The typical i1Pro can do a decent job of producing good profiles, but people come to us for profiles when they want to get that last bit of accuracy and quality that make their prints the best they can be. These matte-surfaced profiles for example are not made to satisfy those who want them to look correct in ColorThink, but are made to produce the best prints they possibly can. Soft-proofing with these profiles will naturally show a slightly more saturated preview, but it is a trade-off that most people using this troublesome media gladly prefer, and that’s why we go in that direction.
I like to geek out as well as the next guy, but at thirteen posts we might all want to stop counting numbers, take a step back, take a deep breath. I’d like to hear from someone who has actually printed with one of these profiles?
I plan to do just that with a profile I make and the one RR has on their website. By way of you final comments, did CHROMiX prepare the profiles for RR?
-
Certain media types benefit from polarized measurements because the polarization filter helps the light from the surface enter the spectrophotometer at a more direct angle. Canvas and other fabrics benefit from this treatment a lot, but also matte papers and even flat papers from toner-based printers. These kinds of media are notorious for having blacks and shadows that get “blocked up” in the shadows. This kind of measurement treatment will produce profiles that give more shadow detail than you’d otherwise see. And this additional shadow detail is built right into the profile, so you receive the benefit of it whether or not you have an app that does black point compensation, and regardless of rendering intent. A by-product of taking polarized measurements is that the darker patches actually measure darker, and the profile reflects those darker measurements. This has no effect on the overall darkness of the printed result.
Pheroid, thanks. I like to learn something new every day. I just tested out your argument with my vernerable old Spectrolino that does have a Polarizing filter attachment. I always knew is was a useful filter for glossy and very "toothy" canvas stocks that show all sorts of specular reflections coming off the surface every which way, but I never thought it would offer a significantly different measured value when using traditional matte finish media like Hahnemuhle Photo Rag, etc., because there is no specular component to be seen visually. And yet, I just measured a the L*min value on a sheet of Moab Entrada Natural (quite similar in smooth surface matte texture to Hahnemuhle Photo Rag) with my Spectrolino's polarizing filter in place, and then again without. The sample was printed on an HP Z3200. Yikes.. the value without is what I'm used to measuring, i.e. L*min = 17. The value with filter in place dropped to 5.6! I stand before you a humbled man, but I also know that it is now time for me to further study which filter set will truly create the better profile. You say it's the polarized data set. Xrite engineers don't know that? Wow. I would personally like to reserve judgement until I've had a chance to study the matter further. Anyway, thanks to you and Steve for the heads up on this issue. Who Knew! Bummer if it means, my new i!Pro2 just got obsoleted by my classic Spectrolino/spectroscan that's no longer made.
kind regards,
Mark
http://www.aardenburg-imaging.com
-
Sadly, I think you mean "no longer made"?
-
Sadly, I think you mean "no longer made"?
Typo fixed. Thanks
-
And now the discussion turns to "which is the better profile?" I prep images very differently when printing on matte papers versus glossy, in part due to the nature of the high L*min recorded by spectros that aren't using polarizing filters when measuring matte media. Will my editing techniques change significantly if I use a profile built from a data set using polarizing filter? Will it be easier or harder to achieve good shadow detail in the final print? Inquiring minds want to know :)
best,
Mark
http://www.aardenburg-imaging.com
-
Wow also. My VLM cartridge will arrive tomorrow and I can print a profile target. I have also downloaded the RR profile and will compare prints using the Outback test print. I also have some B/W images with tricky shadow detail that might also be good to print.
This still leaves a big question for those of us that create profiles. My i1 is a standard spectrometer, how do we get the necessary instrument if a polarizer is important?
-
Pat has raised a very intriguing prospect about the use of a polarizer filter. With hindsight, it's kind of surprising we hadn't heard more about that in the past from X-Rite or from our other usual sources of Color Management expertise. The first question I would have is whether reading matte profiling targets with such a filter buys one a profile that would actually cause an ink laydown that returns a minimum value of L*5 (Maximum Black) or so on a matte paper.
-
Pat has raised a very intriguing prospect about the use of a polarizer filter. With hindsight, it's kind of surprising we hadn't heard more about that in the past from X-Rite or from our other usual sources of Color Management expertise. The first question I would have is whether reading matte profiling targets with such a filter buys one a profile that would actually cause an ink laydown that returns a minimum value of L*5 (Maximum Black) or so on a matte paper.
It would indeed return the minimum value measured if you use the same spectro/filter combination to measure the printed output. The questions more to the point are: 1) What do observers perceive, something closer to the darker measured value or to the lighter measured values for deep shadows and maximum printed blacks in the printed image? Neither one is likely to perfectly match human perception in a real world viewing environment, but which one is on average the closer choice? 2). Which profile, i.e., the one with the polarized and thus darker measured shadow values or the non polarized one with the lighter measured values is likely to assist the user in producing optimal printed output? The one with darker measured shadow values will indeed give a softproofed image with better visual contrast, but that in turn would suggest less demanding edits are then called for to open up shadow details while in softproof mode, but that in turn might lead to a print with even less shadow separation for any printmaker who goes to the trouble of editing images carefully when using the non polarized version of an ICC profile for matte media. Again, it all depends on what the viewer actually sees when looking at the printed output. I don't know the answer. Perhaps Patrick and Steve at Chromix.com have done those studies and concluded polarized instruments make for better profiles. Any profile returning lower L*min values in softproof mode will exhibit more visual contrast in softproof mode, but does that make for better printed output? I haven't done those studies, so I honestly don't know the answer.
best,
Mark
http://www.aardenburg-imaging.com
-
Good point on the difference between the observer and a spectrometer. I think one has to do some printing to see if there is a real difference. Theoretically, if the B/W range is extended with a deeper black point it should lead to more separation.
-
Good point on the difference between the observer and a spectrometer. I think one has to do some printing to see if there is a real difference. Theoretically, if the B/W range is extended with a deeper black point it should lead to more separation.
Sure, and a difference between L*5 and L*15 would be definitely visible.
But I would still like to see a demonstration somehow of the round-trip minimum L* value between a profiling target read with polarizer-equipped spectro and the black patch value of a print using that profile. Mark's observation that it should work sounds plausible of course, but this is so novel I'm in "show me" mode.
-
Something still seems strange to me. I pulled out my old optics text to see if there might be an explanation but that just showed how much math I have forgotten.
If you think about it the printer prints a maximum black based on the driver and paper settings. This is fixed but we make two measurements using different instruments. What does this mean? I need to sleep on this.
-
Something still seems strange to me. I pulled out my old optics text to see if there might be an explanation but that just showed how much math I have forgotten.
If you think about it the printer prints a maximum black based on the driver and paper settings. This is fixed but we make two measurements using different instruments. What does this mean? I need to sleep on this.
Me, too. But one thing for sure. The black patch I just measured with two distinctly different values in polarized versus non polarized instrument mode didn't change its appearance in terms of how I perceive it under any given lighting condition. It only got a lower value assigned to it with polarized data set versus the non polarized data set. Whether that makes for better, worse, or no difference in the ICC profile values sent to the printer is unclear. As for softproofing, I have no doubt the polarized data set will create an inverse transform LUT in the profile that will appear to have more visual contrast and color saturation in the softproof on screen. You can just turn a matte media softproof mode off if you want to see how increased image contrast appears on screen. Yet, the question remains "how does it print?".
best,
Mark
http://www.aardenburg-imaging.com
-
Who Knew!
If you can get your hands on a “Batamag 12X wide field magnifier with dual LED lights”, the experience could be a real eye-opener. I suspect the view, with all its problems, is very similar to what the common spectrophotometer sees. Lots of glare from texture on certain media.
Link to the Betamag loupe source (https://betascreen.net/products/betamag-12x-w-light)
So, “Who Knew?”
Some folks who design fine art gallery lighting and who use polarized light sources for copy work or regularly check printing tests and rosettes using a common loupe in the industry might have an idea.
As an illustration to portray what an unpolarized light source reflects as a worst-case…
Attached is a photo of the right side of the Bill’s Balls test image printed on a popular, premium glossy canvas from an Epson GS6000. (The metal object along the right edge is the back side of a 13 inch Kodak mercury thermometer acting as a weight to hold the print flat.) Although the common spectrophotometers only measure a very small swatch, the sightline is as troublesome. In this particular case of profiling, the typical readings produced hard-edged banding in some of the gradients and was solved by using the profile from the Satin finish canvas instead. Unfortunately, the alternative to use a polarized spectro was not in the cards at the time.
-
Yet, the question remains "how does it print?".
Quoting a snip from the Chromix blog from April 30, 2010 about the Barbieri Spectro LFP Series 3 with polarizing filter:
<if you make printer profiles for media that has flat or bumpy surfaces like canvas or other fabrics - this will get you the shadow detail in your profiles that you won't get any other way.>
This makes sense to me because shadow detail is a common problem with canvas and, with a loupe, you can see white specular highlights against the black(ish) field. Therefore, it seems the profile software could interpret darks as too light and erroneously make them darker.
-
Reading the specs of modern spectrometers, polarized measurements right now are M3 measurements that do not include UV light. That may not be a real problem as I guess a lot of profiles used by forum members are made with UV cut spectrometers anyway. OBA effect added to the measurements are then often extrapolated from readings in the visible spectrum. I wonder whether the SpectraScan could use both UV + polarized filtering in one measurement. Another question is whether measurements with a polarization filter have an influence on the rest of the tone range including paper whites, it must be a diminishing effect.
Edit2:
Given the above, it should be good to measure in both conditions OBA free and OBA containing papers and see what happens then with Dmax L numbers.
Matte and gloss texturised papers and canvasses both exist and anything in between including horrible glossy embossed textures, I would expect the glossy ones to be even more prone to deviations like mentioned. A perfect Ulbricht integrating sphere should cope with this, the 45 degr light cone of most spectrometers in use does not cope it seems.
Edit1:
Bringing this to the color management forum with the right subject line may give us better information from people like Graeme Gill, no offense meant to the forum members here.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
-
Me, too. But one thing for sure. The black patch I just measured with two distinctly different values in polarized versus non polarized instrument mode didn't change its appearance in terms of how I perceive it under any given lighting condition. It only got a lower value assigned to it with polarized data set versus the non polarized data set. Whether that makes for better, worse, or no difference in the ICC profile values sent to the printer is unclear. As for softproofing, I have no doubt the polarized data set will create an inverse transform LUT in the profile that will appear to have more visual contrast and color saturation in the softproof on screen. You can just turn a matte media softproof mode off if you want to see how increased image contrast appears on screen. Yet, the question remains "how does it print?".
best,
Mark
http://www.aardenburg-imaging.com
I recall experiences where matte prints black areas looked different in density but did not differ in L value or showed reversed L numbers. My friend Bernard and me and early Canon iPF printer versus Epson models on Hahnemühle texturised papers. Not at all scientifically theorized or researched afterwards. Just uttered: instruments and humans.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
-
I recall experiences where matte prints black areas looked different in density but did not differ in L value or showed reversed L numbers. ............
Agree. I've reported on such experiences in my review of the Canon Pro-2000 printer and some papers that Canon had brought to market - cases where differences of apparent Black density are larger than differences of L* values would suggest they may be.
-
Reading the specs of modern spectrometers, polarized measurements right now are M3 measurements that do not include UV light. That may not be a real problem as I guess a lot of profiles used by forum members are made with UV cut spectrometers anyway. OBA effect added to the measurements are then often extrapolated from readings in the visible spectrum. I wonder whether the SpectraScan could use both UV + polarized filtering in one measurement. Another question is whether measurements with a polarization filter have an influence on the rest of the tone range including paper whites, it must be a diminishing effect.
Edit: bringing this to the color management forum with the right subject line may give us better information from people like Graeme Gill, no offense meant to the forum members here.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
Hello Ernst, thanks for supplying the M3 term for the polarized measurement mode. I had forgotten about it, and probably with good reason. BTW, the Gretag Spectrolino/Spectroscan models measure M0, M2, and M3. M1(with more specified UV in the instrument's light source) is the missing one because it was defined only recently. Only one filter condition can be used at a time so, M3 +M2 combined can't be measured, but fortunately I don't think it's all that necessary for the reasons you also stated, and then some.
After some further M0 and M3 measurements on various media (including a Macbeth Color checker chart which has matte finish patches and a Kodak greyscale that has glossy photographic "F" surface finish, I don't think M3 is all that useful a mode for making ICC profiles with most photographic and fine art inkjet media. It will indeed open up shadows but at a significant mismatched expense to softproofing and to midtone lightness and chroma reproduction. I base this opinion on the fact that as RGB values go down and the printed ink density rises, the M3 condition measures progressively darker and more saturated LAB values compared to M0 (or M1). The CMM's evaluation of the forward transform of the profile will then be attempting to both lighten the whole printed density scale and desaturate as well. This could be an interesting "rendering" condition for some images on some matte media, but I think the color errors are just too high to expect that further image edits won't be necessary to make a good print, and furthermore, those edits will have to be done without benefit of accurate softproofing. The softproof will appear richer and more contrasty while the actual print will appear lighter, with possibly more separation in the shadows at the expense of midtones plus lower in overall color saturation. Here's a few measured L* values:
For Macbeth ColorChecker black patch: M0 L*= 20. 6, M3 L*=6.7 (note: the published value for this black patch agrees with MO measurement)
For Macbeth ColorChecker midtone grey patch M0 L* =50.7, M3 = 44.1 (note: the published value for this black patch agrees with MO measurement)
For the Macbeth ColorChecker green patch M0 Lab = 54.2, -39.4, 34.1 whereas M3 Lab = 50.1,-47.5, 41.7 (i.e. darker and more saturated), so to print a match to the digitally encoded green patch value using a digital version of the chart, a profile made with M3 condition will attempt to send a higher and less vivid RGB triplet to the printer compared to a profile made with M0, M1, or M2 conditions. That can't be good with respect to the final printed color accuracy.
I looked at some Moab Entrada color patches as well with M0 and M3 measurements. Similar story. Also interesting was comparing the Entrada max printed black value made with my Z3200 to the Entrada max printed value made with my Epson P600. The M0 measurement was 17.0 for the Z3200 on this paper and the P600 measured 17.7. For M3 measurement, HP dropped to 5.6 and P600 value dropped to 1.0!!!! Now we have a good reason to believe why RR's profiles are showing such low L*min values. The were likely built with M3 measurements. However, to my eye, both the Z3200 print and the P600 print both had very consistent visual appearance in the black printed patches, and to my eye a good black but definitely not worthy of L=1.0 when looking at them in various lighting conditions in my studio. M0 measurements are thus closer to what I think makes visual sense to the human observer when viewing most photographic and/or fine art media, and logically, it now makes total sense to me why M3 is rarely used to make ICC profiles of photographic and inkjet media.
That said, I will go ahead today and build two profiles for the Entrada Rag Natural paper (no OBAs), one using M3 data and the other using the conventional M0 data. It will be very interesting to see if the M3 built profile prints the way I anticipate it will, and whether the rendered outcome has any useful relevance in the profile maker's toolbox :)
cheers,
Mark
http://www.aardenburg-imaging.com
-
Hello Ernst, thanks for supplying the M3 term for the polarized measurement mode. I had forgotten about it, and probably with good reason. BTW, the Gretag Spectrolino/Spectroscan models measure M0, M2, and M3. M1(with more specified UV in the instrument's light source) is the missing one because it was defined only recently. Only one filter condition can be used at a time so, M3 +M2 combined can't be measured, but fortunately I don't think it's all that necessary for the reasons you also stated, and then some.
After some further M0 and M3 measurements on various media (including a Macbeth Color checker chart which has matte finish patches and a Kodak greyscale that has glossy photographic "F" surface finish, I don't think M3 is all that useful a mode for making ICC profiles with most photographic and fine art inkjet media. It will indeed open up shadows but at a significant mismatched expense to softproofing and to midtone lightness and chroma reproduction. I base this opinion on the fact that as RGB values go down and the printed ink density rises, the M3 condition measures progressively darker and more saturated LAB values compared to M0 (or M1). The CMM's evaluation of the forward transform of the profile will then be attempting to both lighten the whole printed density scale and desaturate as well. This could be an interesting "rendering" condition for some images on some matte media, but I think the color errors are just too high to expect that further image edits won't be necessary to make a good print, and furthermore, those edits will have to be done without benefit of accurate softproofing. The softproof will appear richer and more contrasty while the actual print will appear lighter, with possibly more separation in the shadows at the expense of midtones plus lower in overall color saturation. Here's a few measured L* values:
For Macbeth ColorChecker black patch: M0 L*= 20. 6, M3 L*=6.7 (note: the published value for this black patch agrees with MO measurement)
For Macbeth ColorChecker midtone grey patch M0 L* =50.7, M3 = 44.1 (note: the published value for this black patch agrees with MO measurement)
For the Macbeth ColorChecker green patch M0 Lab = 54.2, -39.4, 34.1 whereas M3 Lab = 50.1,-47.5, 41.7 (i.e. darker and more saturated), so to print a match to the digitally encoded green patch value using a digital version of the chart, a profile made with M3 condition will attempt to send a higher and less vivid RGB triplet to the printer compared to a profile made with M0, M1, or M2 conditions. That can't be good with respect to the final printed color accuracy.
I looked at some Moab Entrada color patches as well with M0 and M3 measurements. Similar story. Also interesting was comparing the Entrada max printed black value made with my Z3200 to the Entrada max printed value made with my Epson P600. The M0 measurement was 17.0 for the Z3200 on this paper and the P600 measured 17.7. For M3 measurement, HP dropped to 5.6 and P600 value dropped to 1.0!!!! Now we have a good reason to believe why RR's profiles are showing such low L*min values. The were likely built with M3 measurements. However, to my eye, both the Z3200 print and the P600 print both had very consistent visual appearance in the black printed patches, and to my eye a good black but definitely not worthy of L=1.0 when looking at them in various lighting conditions in my studio. M0 measurements are thus closer to what I think makes visual sense to the human observer when viewing most photographic and/or fine art media, and logically, it now makes total sense to me why M3 is rarely used to make ICC profiles of photographic and inkjet media.
That said, I will go ahead today and build two profiles for the Entrada Rag Natural paper (no OBAs), one using M3 data and the other using the conventional M0 data. It will be very interesting to see if the M3 built profile prints the way I anticipate it will, and whether the rendered outcome has any useful relevance in the profile maker's toolbox :)
cheers,
Mark
http://www.aardenburg-imaging.com
I added some edits to my message. A spectrometer with a better integrating sphere would be the right answer to the problem in my opinion. that is beyond handheld ones.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
-
Something still seems strange to me. I pulled out my old optics text to see if there might be an explanation but that just showed how much math I have forgotten.
Possibly what is happening is that matte paper is scattering the polarisation: if you light it with light that is strongly polarised in a given orientation (ie E-field vertical) and it is reflected after being scattered into random angle, in addition to a proportion absorbed by the pigment. A viewer with no polarisation preference would see the full amount of light reflected, whereas one viewing through a narrow angle polariser would see only the vertical component: ie the average over theta of cos²(theta), which would reduce the measured return by a factor of 1/2.
It would be interesting to measure the light reflected in each of the two orientations, ie with the receiver polariser parallel or orthogonal to the illumination polariser...
-
I built two profiles today, one with M0 measurements and one with M3 measurements of a TC918 color target printed by my Epson P600 on Moab Entrada Natural. I used BasicColor Drop RGB as the profiling software. Because Entrada Natural is OBA-free, my M0 measurements will match M1 and M2 conditions as well. M3 is the odd one out, measuring the colors significantly darker and more saturated as I discussed earlier in this thread. The result was as I predicted. M3 data does indeed open up deep shadows, but it comes at a price in overall color and tone fidelity at least when using matte finish media.
The M0 data set made a profile which delivered a predictable and color accurate print of the Macbeth ColorChecker chart and also this Aardenburg test target (attached) which exercises the full sRGB color and tonal gamut in 12 Lab hue planes plus skintone quadrants. Softproof in PS with "simulate paper color" turned on was colorimetrically accurate on screen and a good match to the printed output. The prints made with the profile using the M3 data set did indeed exhibit very open shadows, but it came at the expense of midtone lightness and contrast relationships as well as reduced color saturation in the finished print. So pick your poison as to what image edits you want to make to improve your color reproduction on matte media. The M3 data set makes it more challenging not less, IMHO. Also Softproofing is no longer accurate, definitely not with "simulate paper color" invoked when using the M3 data set. Some folks call the simulate-color-paper feature the "make it look ugly button" and don't use this softproof feature, but I use it all the time, because it forces one to make very specific edits to restore visual contrast to the image reproduction on matte papers. Thus, the M0,M1,or M2 data sets are required to simulate the printed colors accurately. Profiles made with M3 data give up that feature, and even if you don't invoke it in PS but still turn on softproofing, the softproof is still not quite right. Hence, M3 built profiles will require more iterative printing to achieve edits to correct color saturation, midtone lightness and contrast, etc. I don't think that is a very good tradeoff for the sake of opening up shadows for which there are many other useful tools to achieve the same desired effect in PS, LR, and other image editing apps.
Interestingly enough the printed output from the M3 profile wasn't so far off the beam that many people would reject the profile out of hand. For amateur printmakers like many of Red River's customers, they might be perfectly satisfied with this profile output quality. However, it's definitely not a profiling methodology I would recommend to discerning printmakers for photographic and fine art media. Perhaps the exception is seriously glossy and textured canvas media that show lots of specular highlights off the surface and where the polarizing filter will help to reduce those color robbing reflections. I never print on that type of canvas, so I can't say for certain.
cheers,
Mark
http://www.aardenburg-imaging.com
-
Interestingly enough the printed output from the M3 profile wasn't so far off the beam that many people would reject the profile out of hand. For amateur printmakers like many of Red River's customers, they might be perfectly satisfied with this profile output quality. However, it's definitely not a profiling methodology I would recommend to discerning printmakers for photographic and fine art media. Perhaps the exception is seriously glossy and textured canvas media that show lots of specular highlights off the surface and where the polarizing filter will help to reduce those color robbing reflections. I never print on that type of canvas, so I can't say for certain.
cheers,
Mark
http://www.aardenburg-imaging.com
Is the target you posted the "Hue Slice" image from the website? If so, can I print it directly from LR after converting it in Photoshop?
I understand the point about really glossy papers and how a polarizing filter might help out in that case. I'm still baffled by the effect seen with matte papers where one doesn't get as much reflection. One would be concerned if there was color shifting taking place using the RR profile approach. I'll see when I do some test prints.
Unblocking highlights is easy to understand as the black point gets shifted to that the range from paper white to pure black is extended with a spectro read of L* = 5 compared to L* = 18 (the normal case I observe when profiling matte papers using Argyll). The darker shadow tones would see more separation BUT is this only something that is seen under soft proofing? As I noted in an earlier post, the true black on the print is not changing, only the apparent reading is. It's all pretty strange and I'll wait to see what the prints end up looking like.
-
Is the target you posted the "Hue Slice" image from the website? If so, can I print it directly from LR after converting it in Photoshop?
The hue slice target you are referring too is different. The one attached in this thread is not currently available on the Aardenburg website. I should probably put it there, but if you save out the one attached in this thread, it's a jpeg in sRGB colorspace, so it should import and print just fine in PS or LR onto a letter-size sheet of paper. The lab hues were designed and encoded originally working in LAB colorspace, but were intended to port in gamut to sRGB colorspace, so this jpeg version will serve the same useful color reproduction evaluation purpose as the master file from which it is derived.
kind regards,
Mark
-
The hue slice target you are referring too is different. The one attached in this thread is not currently available on the Aardenburg website. I should probably put it there, but if you save out the one attached in this thread, it's a jpeg in sRGB colorspace, so it should import and print just fine in PS or LR onto a letter-size sheet of paper. The lab hues were designed and encoded originally working in LAB colorspace, but were intended to port in gamut to sRGB colorspace, so this jpeg version will serve the same useful color reproduction evaluation purpose as the master file from which it is derived.
kind regards,
Mark
Thanks, I've downloaded it.
Alan
-
Only on LuLu would you get this extraordinarily technical discussion of the intricacies of profile a matte paper. I love it!
On a more prosaic note, if any of you does develop a great profile of the Palo Duro Etching paper and you aren't restricted by copyright/software limitations I hope you will share it.
And just today Red River has made sample packs of the Palo Duro Etching paper available. 5 sheets for $4.99 including shipping. I don't normally use textured papers but I've ordered a sample pack to see if I like it. Link is: http://enews.redriverpaper.com/q/Gs77EJbP8cue0XLSVwq99be2wW38cPFHW0sZcOJcGF1bC5yaWNrZXJ0QGdtYWlsLmNvbcOIEyzZmIjGfKLNYZUAQPmtnehcl2g
-
Hi all,
It's fascinating how much discussion this has brought up. As Pat mentioned, we've been using the polarizing filter, selectively, at CHROMiX for at least 17-18 years.
A few responses to questions / comments in the thread so far
I wonder whether the SpectraScan could use both UV + polarized filtering in one measurement
/
Only one filter condition can be used at a time so, M3 +M2 combined can't be measured
First, both M3 standard and the Pol filter for the 'Lino include UV filtering. So, by definition, M3 is really M3(+M2)
The first question I would have is whether reading matte profiling targets with such a filter buys one a profile that would actually cause an ink lay-down that returns a minimum value of L*5 (Maximum Black) or so on a matte paper
/
Theoretically, if the B/W range is extended with a deeper black point it should lead to more separation
As postulated above, using a Pol filter does not make the printer print to a higher density. For the measurement numbers it decreases the L* value as well as increases saturation, especially in the darker colors.
The physics are fairly simple. "First surface reflection" is the key here. In the case of a glossy surface there's very little light scattering and typically a spectral highlight reflection that we avoid viewing directly. The more bumpy the surface (and it increases as inks dry into the media) the more scattering that occurs from this first boundary. This scattering occurs before any ink colorants have a chance to do their filtering and often before the color of the media itself has an influence. So it's white light and it dilutes our perception of the color.
Polarization works in a least two ways. Light is naturally polarized when it reflects off a surface. This is why polarized sunglasses effectively cut the glare from car windows and the surface of water. Additionally, if the light source is polarized prior to reflecting off the sample, then there's a greater chance of the cross-polarized filter cutting down the first surface reflection. Just like sunglasses allow seeing fish in a river, polarized readings allow discrimination of shadow detail in print.
...I don't think M3 is all that useful a mode for making ICC profiles with most photographic...
There are a lot of mental calisthenics that can go into this. For us, as I expect for most of you, the proof is in the pudding. Over the years we've found that Pol measurements greatly improve some profiles while not improving others. Surprises like significant improvements in some toner profiles compell us to try it in many different situations.
The reality is that neither instrument is likely "correct". Unfiltered i1's may measure 23 L* while a Pol filtered 'Lino or Barbieri spectro LFP may measure down to 2-5 L*. The reality is somewhere in between.
Here's an interesting test: . .
- Print a graduated gray ramp on glossy paper and on the uncoated paper in question
- Measure both and Write the measured L* value beside each patch
- Place them under controlled lighting (not just color but angle of illumination as well)
- Visually determine the match between the darkest patch on the uncoated paper and visual equivalent patch on the glossy paper
The L* value of the matching glossy patch will be lower than the uncoated patch and what the instrument probably should have measured, if it weren't for all the surface scattering. You can numerically stretch the uncoated paper's measurements down to that lower value and it will likely produce a profile that's better at proofing it's expected output. We've just added a feature to our Curve4 software that does this very scaling. It helps create a match between an uncoated press sheet (measuring too light) and a proof on proofing paper (Curve4 is for G7 calibration of digital and conventional presses, not really appropriate for this crowd)
But will it produce a profile that actually renders output better? Probably not.
Why?
When measurements are used to generate a profile and it's used for rendering output, it's mostly about relative mapping. The lightest point will map to paper white (even if you're using the wrong profile) and the blackest point will likely map to the blackest print point (due to the behavior of the perceptual intent or black point compensation with rel col intent).
So, the effect of Pol measurements is felt in two separate ways:
1. As mentioned, the lower L* and saturation creates a profile with a larger gamut volume and this will create soft proofs with a higher dynamic range. They are probably eggagerated somewhat, though we have scaled measurements to correct for this in the past.
2. The greater dynamic range of the measurement allows for the detection and encoding of more color gradation. This often translates into a profile that can address such detail and create prints that have fewer problems with washed out shadows. We've built thousands of profiles over the years and I can attest to it working, and working well, in some printing situations.
The situations in which it makes the difference is something we've arrived at through experimentation.
As for the instruments? Gretag MacBeth's SpectroLino/SpectroScan and the Barieri Spectro LFP are the two automatic instruments that we've used over the years - and the 'Lino's long out of production. Some hand-helds like the eXact, KMS FD-7 and SpectroDens will take M3 measurements and can measure strips in a manner similar to an i1Pro or be mounted in the ColorScout table. None of these solutions are inexpensive. - the polarization filter alone for the LFP is around $800 (!)
As Pat mentioned, we hound manufacturers continuously to add M3 to their instruments like the i1Pro, iSis and FD-9 but they haven't so far. M3 can cut down the light significantly so it slows the measurement process down, but for us it's worth it. Our SpectroScan tables with the Pol filter can take an hour and a half to measure a single target - which is why we have a bank of them. But again, when it's worth it, it's worth it.
I hope this helps.
Steve
-
As Pat mentioned, we hound manufacturers continuously to add M3 to their instruments like the i1Pro, iSis and FD-9 but they haven't so far.
Steve
Of course not - let's remember who one is dealing with at least with i1Pro and ISis. First they obsolete perfectly good equipment, and then they won't listen when knowledgeable professionals make serious suggestions to them It's typical X-Rite through and through.
Anyhow, on the broader question, thanks very much for all the experience you've brought to this discussion. Very helpful.
-
As Pat mentioned, we hound manufacturers continuously to add M3 to their instruments like the i1Pro, iSis and FD-9 but they haven't so far. M3 can cut down the light significantly so it slows the measurement process down, but for us it's worth it. Our SpectroScan tables with the Pol filter can take an hour and a half to measure a single target - which is why we have a bank of them. But again, when it's worth it, it's worth it.
I hope this helps.
Steve
I'm still curious whether a better integrating sphere would not do a better job than filtering out specular light reflection that is partly caused by the simpler 45 degr cone lighting found in the usual spectrometers. Both dynamic range and color information could be improved compared to both methods discussed in this thread.
http://blog.xrite.com/what-is-a-spectrophotometer/
https://www.avantes.com/products/accessories/item/269-integrating-spheres
I am aware of the standards in use for the graphic industry so I do not expect (affordable) instruments like that to appear for profile creation in this industry.
For B&W aficionados and their customized workflow/software/drivers to quadprinters the use of polarization filtered measurements could be a godsend. The more as they tend to edit their B&W curves or profiles to their taste too.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
-
With all due respect to the experts at Chromix, I'm just not feeling the love for the M3 polarized measurement condition for routine use with smooth fine art matte media. Admittedly, I built just one profile, but Moab Entrada Rag Natural is highly representative of this broad class of matte finish media on the market today. Yes, M3 expanded the L* range in the data and the chroma range as well (not actually on the print) and it did open up deep shadows (definitely on the print) when compared to the profile built with the M0 data set. But the result is abnormally overcorrected in my opinion, and introduces other unwanted color and tone relationships, both in the print and unequally in the softproof thus rendering softproofing far less useful. An L* minimum of 1.0 for Epson HP ink on Matte paper? Seriously, that is just one JND unit away from blackest human observable black, but the observed tonal range of the print under any normal viewing conditions one is likely to encounter in the real world definitely doesn't look like that.
Thus, I'm not personally seeing M3 data collection as a pathway forward for superior custom made profiles. That said, I do recognize that the M0, M1, and M2 data sets along with the 0/45 degree instrument geometry can somewhat understate the perceived black and deep shadows in matte media, so some type of error correction may very well be of some benefit. I am indeed surprised that the M3 data set doesn't alter the profile output in a much more heavy handed way, so I defer to Steve and Patrick's considerable judgement that they like the "error correction" provided by the M3 measurement condition. Again, the fact that this rescaled data set doesn't totally screw up the profile and provides some custom goodness that others apparently like, suggests to me a more controlled correction could be routinely achieved simply by using Excel to manipulate the data. One could easily import the text file, rescale the L* data to taste with or without messing with the a* and b* data, for example, then paste back into the CGATS text file to then be used to build the profile. I have never tried it before because, like the M3 data set, it will apply the correction to all the LUTS in the profile. Hence, a true profile editing app that lets you customize inverse transform separate from forward transform always seemed called for, but now I think Excel could do the trick and in less time than it takes to collect the M3 data.
Another day of learning something new :)
kind regards,
Mark
http://www.aardenburg-imaging.com
-
\ Again, the that fact that this rescaled data set doesn't totally screw up the profile and provides some custom goodness that others apparently like, suggests to me a more controlled correction could be routinely achieved simply by using Excel to manipulate the data. One could easily import the text file, rescale the L* data to taste with or without messing with the a* and b* data, for example, then paste back into the CGATS text file to then be used to build the profile. I have never tried it before because, like the M3 data set, it will apply the correction to all the LUTS in the profile. Hence, a true profile editing app that lets you customize inverse transform separate from forward transform always seemed called for, but now I think Excel could do the trick and in less time than it takes to collect the M3 data.
I pretty sure that ArgyllCMS will allow one to do something like this (though it would be nice to have Graeme Gill weigh on to make sure).
-
Hence, a true profile editing app that lets you customize inverse transform separate from forward transform always seemed called for, but now I think Excel could do the trick and in less time than it takes to collect the M3 data.
Another day of learning something new :)
kind regards,
Mark
http://www.aardenburg-imaging.com
Which Kodak Colorflow Custom Colors could do I think.
Met vriendelijke groet, Ernst
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots
-
I believe there is an issue with the Red River/Chromix profiling strategy that hasn't been directly addressed. I teach people how to use Photoshop, including a heavy dose of soft proofing. As has been pointed out the soft proof with these type of profiles is essentially useless for print prediction, it is definitely NOT a subtle effect, I've built a custom Etching profile and compared its soft proof to the RR profile.
Up to now I've advised my students to download the paper manufacturer's profiles and use them for soft proof and printing. If the results are satisfactory, which they usually are, then no custom profile is needed. However given the RR approach I feel I now have to examine each profile before anybody I advise uses it.
Red River/Chromix should include in their profile usage instructions a description of the polarizer effect upon soft proofing, and perhaps consider offering an alternative profile that is built without polarization. Anything less is a disservice to their customers.
Richard Southworth
-
An L* minimum of 1.0 for Epson HP ink on Matte paper? Seriously, that is just one JND unit away from blackest human observable black, but the observed tonal range of the print under any normal viewing conditions one is likely to encounter in the real world definitely doesn't look like that.
I don't disagree with you there
...suggests to me a more controlled correction could be routinely achieved simply by using Excel to manipulate the data. One could easily import the text file, rescale the L* data to taste with or without messing with the a* and b* data, for example, then paste back into the CGATS text file to then be used to build the profile.
If this were sufficient then I assure you, we would have been doing it years ago.
You can do a lot in math, overall, but you can't cheat the physics.
It's analogous to getting a scan from a consumer-level scanner and not being happy with the blacks. Just setting a new black point in Photoshop isn't going to open up the shadows while it darkens the image. If the data wasn't there in the first place... well, you know.
Also, as I mentioned in my earlier post, the calculation and use of the profile effectively scales the output dynamic range automatically.
....As has been pointed out the soft proof with these type of profiles is essentially useless for print prediction, it is definitely NOT a subtle effect, I've built a custom Etching profile and compared its soft proof to the RR profile.
Up to now I've advised my students to download the paper manufacturer's profiles and use them for soft proof and printing. If the results are satisfactory, which they usually are, then no custom profile is needed. However given the RR approach I feel I now have to examine each profile before anybody I advise uses it.
Overall I agree but I also caution against considering the measurements of a non-M3 device to be correct. I'm sure they're what people are accustomed to but the reality is really found somewhere between the M3 measurements and non-M3.
Red River/Chromix should include in their profile usage instructions a description of the polarizer effect upon soft proofing, and perhaps consider offering an alternative profile that is built without polarization.
That's a good idea. We'll also review here (@ CHROMiX) some techniques for improving M3 measurements for the soft-proofing part of profile use.
We tend to be very focussed on the rendering capabilities of our profiles - which I think we would all agree is the more important function for photo work - but it sounds like we could improve the proofing side of things as well.
Thanks everyone for your reasoned feedback
Steve
-
I just printed out two test prints with the Outback set of images. I won't have my own profile until tomorrow when the ink is dry and I can do the spectro readings. I also have a test print of Hahnemuhle William Turner. The Red River paper looks quite a bit like the Hahnemuhle paper but with slightly less texture. I don't know the Epson textured papers at all so I cannot say provide a comparison. In terms of the most visible black patch, I can see the patch #6 black patch on the William Turner paper but not on the Red River. #8 patch is visible on the RR paper. In terms of the white patches, I can see patches at #250 but the William Turner is a little more visible. The yellow on the RR print seems a little off. The aspen leaves are not as vibrant and the yellow patch seems a little muted.
I'll know a little more when I prepare the profile. I think in terms of a textured paper, I still prefer the better texture of William Turner. Looking at prices, the William Turner appears to be about $6 a box more than the Red River paper (Letter Size sheets).
-
In the Aardenburg color target I posted earlier in this thread there is a black stripe running along the bottom of the image. In that stripe are sets of numerical values 1, 2, 3, 4, 5 followed by letters, R, G, B, C, M, or Y. the values 1-5 are, you probably can guess, encoded in increasing lightness values going from L* = 1, 2, 3, 4, and 5, and each letter is L*=6. I consider these to be the deep shadows in a digital image that a great print shouldn't throw away. If an output profile made with M0 measurements can print those deep shadow values in a perceptibly smooth ramp away from the stripe's max black (RGB = 0,0,0), then, IMHO, there really is no deep shadow problem to be concerned with on account of measuring one's profiling target data with the M0 condition. My own rolled profiles are usually able to print those values cleanly, so again I see no need to resort to M3 measurements to solve a problem I usually don't have. The profile I did make with the M3 measurements actually overcorrected, and worse, there was a subtle but still disconcerting visual jump between 5-15 in the higher shadow values that was then leading to an oddly flatter tone response for some colors in the 15-30 L* range. There were other subtle tone and color reproduction issues as well which could be seen in other areas of the 12 hue planes in that target. Perhaps the folks at Chromix have their own custom profiling software that works better with the M3 data sets. I only have access to i1Profiler, PM5, and BasicColor software and the M3 data set isn't playing as nicely as I'd like to see with those profiling apps.
When I do run into trouble with the lowest shadow values plugging up it is generally due to over inking by the printer. With limited "media settings" to choose from on typical desktop photo printers and the fact that some manufacturers do a better job at linearizing the printer than others, sometimes the reproduction of those deep values is not quite as good as I'd like. The 1, 2, and 3 L* tones can plug up in some of the colors, even the gray ramp. That said, my Canon Pro-1000 has beautiful ink ramps. It's a joy to build profiles for that printer, and the deep shadow detail is all there. My Epson P600 slightly over inks in the photo gray channels on some media/media setting combinations such that a minus percentage value in the density slider found in the printer driver's advance setting menu will help alleviate. Unfortunately, that density slider is a global one affecting all the ink ramps, so system color gamut can be lost if one tries to take the slider far enough negative to fully open up the grey channel ramp. It ends up being a bit of a compromise for some matte media. Glossy/luster papers rarely have an issue with the deep shadow reproduction values. Perhaps M3 measurements might be a way to rescue an over inking situation, however, paying close attention to what media is being selected to build the profile is usually the place to start.
Lastly, as for soft proofs needing some tweaking because the M0 data isn't a perfect match to what we see in the print, a good portion of that mismatch actually has nothing to do with the M0 data. It has to do with the customary way display profiles remap monitor white and monitor black luminance values to L=100 and L=0 respectively, no matter what the actual luminance range of your monitor really is. To simulate a true zero to 100 L* range the monitor's luminance range has to be over 700:1 (I've forgotten the exact calculation, but it's over 700) Many new monitors claim to have those specs, but in reality a great many fall short. Also, unless you don't mind image editing in a cave, the ambient room lighting adds a little veiling flare to your perception of monitor blacks. The end result is that the posted L* shadow values will be lifted a little lighter in reality and that contributes to the perception that the softproof of a matte paper is just a little bit flatter than it should be. The display's baked in tone curve error is not a big error, but it does add to the lower image contrast problem for softproofing matte media especially. I studied this display profile scaling error extensively in the days of CRT monitors that typically had luminance ranges no better than 400:1. I found that by using a profile editor like the one in the Profilemaker 5 suite or Kodak Colorflow software (as Ernst noted earlier) I could correct any perceived M0 data deficiencies and compensate for the actual display errors quite nicely just by rescaling the tone curve in the inverse transform LUT of the profile such that matte media black values of typically 17 or 18 would then post to the softproof view as if they were L*min of approximately 4L darker, e.g. 13-14. Thats all it took to get a near perfect match, and nowadays with higher contrast displays, I don't go to the trouble of making this profile edit anymore. I've learned to live with the slightly lower contrast in the softproof. It actually works to my advantage because it forces me to pay very close attention to the image edits needed to lay the image down beautifully on the print.
kind regards,
Mark
http://www.aardenburg-imaging.com
-
I just finished profiling the paper using ArgyllCMS with a 51 step B/W patch set (overkill, I know). It was done with an i1 Pro with no UV cut. The profile had a very low standard error. I get a black point of L* = 18.8 which is slightly higher than I got for my William Turner profile and it's also in the same range as other matte papers that I use. I don't see as good separation in the very dark shades as with the available RR profile indicating that the expanded B/W range using a polarizing filter may lead to better separation. Looking at the Outback test image critical points I see little difference between the colors from the RR profile and the one I created. I also printed the Aardenburg hue target that Mark posted yesterday. Using my profile I could see the separation he mentioned.
I don't print B/W images on textured papers, preferring a smooth surface. Blocking up of the dark ranges are of less concern for me. Since I like the better texture of the William Turner paper, that will be my paper of choice when I need a non-smooth surface for an image.
-
Using my profile I could see the separation he mentioned.
Alan,
Were you able to see the L*=1-6 characters for all seven groups in the black strip, per Mark's description?
Richard Southworth
-
Alan,
Were you able to see the L*=1-6 characters for all seven groups in the black strip, per Mark's description?
Richard Southworth
Yes, though there is some blocking up between the first two groups. I only printed that test out with my profile, not the one prepared for Red River.
-
Yes, though there is some blocking up between the first two groups. I only printed that test out with my profile, not the one prepared for Red River.
Perhaps it may be helpful to discuss what L*=1, 2, 3 means visually in the CIELAB color model. All three values are very close to full black. One unit of L* change without any change in a* or b* creates one JND (just noticeable difference) with a probability of 50% that the viewer can perceive that difference under the viewing conditions where this color modeling was originally conducted. So, going from maximum digital black L*=0 in the image file (i.e. RGB = 0,0,0) to L*=1 in the neutral (no chroma) direction should be barely noticeable not easiliy noticeable. For L*=2 increase the probability that it can be noticed jumps to about 95%, ie. most people can see that step under good lighting conditions with high certainty, but it's still a just noticeable change not an easily noticeable one. If those values become easily noticeable in the reproduction, they are actually over corrected. By the time we get to L=5 it should be a very easily noticeable difference from black. If a printer is not separating the deep shadows well then that 5 level in the image can get pushed down to the 1 or 2 in the print and even back to max printable black if the printer's tone curve is clipping the shadows.
All that said, one other factor, lateral adaptation, comes distincly into play as dark areas in the image get smaller and the surrounding image area gets lighter. This is the case with the Aardenburg target. The stripe is pretty thin at the bottom of the overall image and surrounded by much lighter areas. Thus, even on the monitor, those numbers and letters embedded in that black stripe can appear to merge into black unless you get your eyes closer to them (e.g. by zooming in on the image in your display or by examing the print at close distance and with very good light. So, with those caveats in mind, one can learn to visually assess those numbers and letters in that stripe along with other color and tone ramps in the Aardenburg iStar color target in order to evaluate whether shadow details are printing too dark, too light, or just right :)
cheers,
Mark
http://www.aardenburg-imaging.com
-
I finally had some time to generate some more data and compare the profile that I made using ArgyllCMS and the one that is on the Red River site which I believe was done by CHROMiX using a polarizing filter to do the spectral readings. This is where things get pretty interesting. As I noted the other day, the visual prints look pretty much the same. Both are printed using the Velvet Fine Art setting which is what I used to do my profiling and what Red River suggests as the setting. However, the measured gamut volumes are hugely different. I only have the ArgyllCMS command line tools to do analyses. Here are the calculated volumes:
Red River Profile: 774514
Alan's Profile: 482558
Alan's Profile of Ilford Gold Fibre Silk: 814289
Alan's Profile of Hahnemuhle Photorag Ultrasmooth (a matte paper I often print on): 491824
Using a polarizing spectro to do the readings has an impact on the measurements but does this translate to something that is visible. I'll defer to someone else who wants to do some profiling of this paper and do a visual comparison. I did some comparing of some color patches of the Outback test image and didn't much of difference either in the printed versions and looking at the profiles on my screen and using the Affinity Photo color pick tool. So the CHROMiX generated profile does not have quite the gamut of Ilford Gold Fibre Silk but is much larger than the other two matte paper profiles that I prepared.
-
Alan,
Your results look more like one would expect.
What printer?
-
Alan,
Your results look more like one would expect.
What printer?
Epson 3880 and I made sure to download the correct profile from the Red River website. It's tagged Epson 3880. Can you clarify your statement that I have highlighted. I don't intuitively understand why the gamut volume should be so high for a matte paper. If the use of a polarized light source is so important it means that a lot of profiling that one does without such a source may be of questionable value. I only have Argyll so I cannot say whether one using other profiling software would obtain similar results.
-
Epson 3880 and I made sure to download the correct profile from the Red River website. It's tagged Epson 3880. Can you clarify your statement that I have highlighted. I don't intuitively understand why the gamut volume should be so high for a matte paper. If the use of a polarized light source is so important it means that a lot of profiling that one does without such a source may be of questionable value. I only have Argyll so I cannot say whether one using other profiling software would obtain similar results.
Hi Alan,
Yes, what I meant with that statement is that the profile which RR provides for its Palo Duro Etching paper, which is a matte paper, has a gamut volume that far exceeds anything I have ever seen for a matte paper - in fact reaches well into expectations for a PK paper, and I measure a lot of them. The result you obtained in your own profiling with Argyll is much more in line with expectations for a matte paper and therefore on the face of it more credible.
-
I received my sample pack today, so I immediately loaded the RR profile for PDE for my Epson R3000, went into LR and printed a profiled (not ABE) BW image. I have no measurement equipment so this is a purely subject assessment. I am very impressed with how rich and deep the blacks appear. I plan on ordering more PDE while it's still 20% off.
Edit: ABW, not ABE
Sent from my iPhone using Tapatalk
-
I’ve worked for CHROMiX over 10 years now, and that whole time I’ve watched Steve Upton pounding the drum for polarized measurements at every chance he gets. He does not keep it a secret. We have hammered on X-Rite to put polarizing filters in more of their instruments. We had Konica Minolta personnel from Japan in our offices just last month, and he was selling the virtues of polarized measurements to the actual instrument makers. We’ve brought this topic up in trade shows, conferences - we’ve mentioned it in our blog, in our newsletter. So from the profile producer’s point of view, it’s rather frustrating that knowledgeable people in the industry still don’t seem to know about or consider polarized measurements.
http://blog.chromix.com/2010/04/
Why ? Few observers will be wearing polarized sunglasses !
It's pretty simply - for creating profiles for reproduction for human observers, the best measurements are those that closely mimic what a human observer sees. So unless you are assuming all your viewers are wearing polarized glasses, it's hard to understand your enthusiasm for polarized measurements!
To put it another way - how does it help for the profile to show a fake, super-good black point when it looks nowhere near that black in real life ?
Certain media types benefit from polarized measurements because the polarization filter helps the light from the surface enter the spectrophotometer at a more direct angle.
Care to re-phrase that explanation in scientific terms rather than hand wavy terms ?
[ From a technical measurement point of view I can see that polarized measurement will greatly improve the
Signal to Noise (S/N) ratio of the measurement of darker colors, but the resulting numbers should then be
restored to visual correspondence (by adding back in the correct level of diffused specular reflection)
before being used to construct a profile that represents the real world. But the visual significance of the
extra S/N (= accuracy) will be diminished by the very real diffused specular component from the paper. ]
-
I have been doing some test prints on the RR PDE paper. I really am impressed with the blacks, but I (and others) have noted over saturated orange colors. Is this a profile issue, or some setting in the driver? Would playing with the Epson Color Density setting help or should I just look at having a custom profile created for my R3000?
Sent from my iPad using Tapatalk
-
"Why ? Few observers will be wearing polarized sunglasses !
It's pretty simply - for creating profiles for reproduction for human observers, the best measurements are those that closely mimic what a human observer sees. So unless you are assuming all your viewers are wearing polarized glasses, it's hard to understand your enthusiasm for polarized measurements!"
I am not getting into a debate over this - I'm simply going to state that this is nonsense.
The use of a polariser when making profiles with matt paper significantly improves the shadow reads - it does so because the polariser removes light scatter. It has absolutely nothing to do with looking at images with polarised glasses and such a notion is laughable. If I show you two prints, one made with a polarised profile and one without you wouldn't know the difference other than the fact the polarised profile had much better shadow detail!
-
If I show you two prints, one made with a polarised profile and one without you wouldn't know the difference other than the fact the polarised profile had much better shadow detail!
Unless both image files used to make the two prints are carefully edited to bring out the desired amount of shadow detail using either type of profile. In which case, results will be entirely comparable. Been there, done that. The problem with the polarized data set is that it screws up the profile soft proofing accuracy significantly more than a non polarized set of data, so iterative printing is more likely to be required for other aspects of the image that have little to do with shadow retention when using a polarized data set to build the profile. I genuinely believe it's best to think of polarized data sets as special subsets of the general ICC profiling process, useful with some images for sure, just like choosing a particular rendering flavor, but not a universal win for all images in all situations.
-
The use of a polariser when making profiles with matt paper significantly improves the shadow reads - it does so because the polariser removes light scatter. It has absolutely nothing to do with looking at images with polarised glasses and such a notion is laughable. If I show you two prints, one made with a polarised profile and one without you wouldn't know the difference other than the fact the polarised profile had much better shadow detail!
Getting measurements that don't correspond to the visual characteristics is not profiling, and will just lead you astray in the long run.
It's possible you might get "better shadow detail" due to poor workflow :- i.e. if you are not properly linking the source and destination profiles using gamut mapping, but instead using a colorimetric (or poorly implemented perceptual) type intent, then yes the mismatch between the source black point (typically 0 for idealized source spaces like sRGB , AdobeRGB etc.) with the actual print black point will cause loss of shadows. So fudging the measurements to give the print an artificially good black point using polarized measurements will improve the situation, without actually tackling the underlying problem. The alternative is to not fool yourself - take measurements that actually correspond to the visual color, and use a good workflow that maps the luminances ranges of source and destination appropriately.
-
The use of a polariser when making profiles with matt paper significantly improves the shadow reads - it does so because the polariser removes light scatter. It has absolutely nothing to do with looking at images with polarised glasses and such a notion is laughable. If I show you two prints, one made with a polarised profile and one without you wouldn't know the difference other than the fact the polarised profile had much better shadow detail!
Actually, Josh, it does. Reflected light from the rough matte, inked surfaces itself selectively reflects light depending on the direction of polarization of each photon. Unpolarised light, when reflecting from a non-metallic surface with a different index of refraction from air will therefor reflect partially polarized light. M3 pre-polarizes light such that light that reflects off these surfaces towards the spectrophotometer is reduced. The effect is stronger with resin encapsulated pigment inks.
The black reduction effect is similar whether the polarizer is at the light source (M3) or at the viewer (polarized sunglasses).
-
Actually, Josh, it does. Reflected light from the rough matte, inked surfaces itself selectively reflects light depending on the direction of polarization of each photon. Unpolarised light, when reflecting from a non-metallic surface with a different index of refraction from air will therefor reflect partially polarized light. M3 pre-polarizes light such that light that reflects off these surfaces towards the spectrophotometer is reduced. The effect is stronger with resin encapsulated pigment inks.
The black reduction effect is similar whether the polarizer is at the light source (M3) or at the viewer (polarized sunglasses).
Its the notion that you need polarised glasses to look at a print made with an M3 polarised profile that is laughable.
By that logic we should we look at photographs shot with a polariser with polarised glasses on!
-
Its the notion that you need polarised glasses to look at a print made with an M3 polarised profile that is laughable.
By that logic we should we look at photographs shot with a polariser with polarised glasses on!
I totally agree, it's a simplistic view of the problem and the use of M3 in profiling.
To continue the flawed logic, it would follow that any profiles built with UV filtering should only be viewed with UV filtered lighting - not a bad idea but also definitely not necessary.
We agree that the proofing intents of M3 profiles need to be improved and we're working on the process.
But, our belief is that M3 measurements, when used with certain media, creates significantly better rendered output and we have years of experience and happy customers to back that up.
An important question that has to be asked here is "Why do people presume that M0, M1 or M2 measurements are necessarily correct?". Our feeling is that the profiles built from those measurements also contain measurement flaws, especially when the media has an optically bumpy surface that reflects a significant amount of light off the surface. It's likely that the best set of measurements is some blend of M3 and others.
We're working the issue and will respond to the list when we know more.
In the meantime, try profiles built with M3 on different before condemning the practice. Mark S (to his credit) approached the idea with an open mind and came away preferring the M3 profiles.
-
Its the notion that you need polarised glasses to look at a print made with an M3 polarised profile that is laughable.
Obviously. However, M3 can overstate DMax by discarding light that would normally reach a viewer of the print. Read again what I wrote. It is the reflections themselves that preferentially select and de-select polarized light reflected at an angle. The purpose of M3 is to discard that portion of light preferentially reflected to the viewer. This creates a false DMax and should not be done. The spectro should be measuring exactly the same light levels that unpolarized light at 45 degrees produces. Which is what M0-2 accomplish though with varying levels of uV to complicate life.
-
This creates a false DMax and should not be done.
The DMax comes out differently in the M3 mode, but false relative to what? It may well be correct within the framework of its own technology - one can discuss that, but of what practical significance, the objective being to produce visibly better prints. So to say it should not be done kind of surprises me, when a number of us have seen that in certain contexts it produces visually superior results - in particular, Black shade appearance and shadow detail in the deep quartertones, and as Josh Holko demonstrated in his article, superior rendition of highlight tonality in very delicate photographs. The effect varies from substantial to insignificant depending on the photo and the paper. So I for one, based on what I've seen and experimented with, would not come down so categorically on it.
-
Obviously. However, M3 can overstate DMax by discarding light that would normally reach a viewer of the print. Read again what I wrote. It is the reflections themselves that preferentially select and de-select polarized light reflected at an angle. The purpose of M3 is to discard that portion of light preferentially reflected to the viewer. This creates a false DMax and should not be done. The spectro should be measuring exactly the same light levels that unpolarized light at 45 degrees produces. Which is what M0-2 accomplish though with varying levels of uV to complicate life.
So what I hear you saying is that you choose to believe the non-M3 measurements as gospel.
Why?
With all the spectral reflections off the surface of glossy canvas, and the human tendency to slightly move one's head to minimize such reflections, yet the instruments inability to move from the 0/45 geometry, why are you thinking that the instrument is seeing things like you?
Again, I agree that the dMax returned from the profile is too dark, but I also counter that the dMax returned from non-M3 profiles is too light
-
So what I hear you saying is that you choose to believe the non-M3 measurements as gospel.
Not at all. M0-2 have their own, very real limitations. M3 takes those limitations and alters the lighting by using polarized light that creates measurements that overstate both dynamic range and saturation. Unless, of course, the prints that use the profiles are illuminated from 45 degrees with polarizing filters. In that case they would be quite nice. And the dynamic range and colorimetry baked into the profile would be exactly right. The viewer wouldn't need polarizing glasses either. But illuminated "normally" they would exhibit poorer dynamic range and less saturation.
Why?
With all the spectral reflections off the surface of glossy canvas, and the human tendency to slightly move one's head to minimize such reflections, yet the instruments inability to move from the 0/45 geometry, why are you thinking that the instrument is seeing things like you?
Again, I agree that the dMax returned from the profile is too dark, but I also counter that the dMax returned from non-M3 profiles is too light
Yes, and that illustrates the limitations of both profile colorimetry and non-Lambertian prints. No profile can account for how people will view prints with specular effects from any glossy or luster type print. Which is why, when I discuss hard proof comparisons I take pains to point out that the illuminant should be at an angle (45 degrees preferably) and the ambient at much lower lux levels.
And, to your point, if properly made glossy type prints are displayed in a white room, evenly lit with diffuse lighting, they will appear to have much lower dynamic range as well as significantly reduced saturation.
Which is why prints of images that are in gamut on both matte and glossy paper, the glossy one will appear quite poorly in the white room environment while the matte prints made with m0-2 profiles will retain their proper appearance. Well, unless you put that matte prints behind standard glass.
But there is a case to be made for M3. If you display both matte prints and glossy type prints in the same white room and illuminate the prints only by ambient room light then matte prints made with M3 will better match glossy prints made with M2. The glossy prints will be effectively reduced in saturation and dynamic range to closer to that of the matte print.
-
The DMax comes out differently in the M3 mode, but false relative to what?
The Dmax is calculated directly from the spectrophotometer measurement of reflected light from the blackest patch. With M2 it is from reflected, randomly polarized light. When M3 is used it measures the fraction of reflected light that was polarized prior to reflection off the print. That will be significantly different on most matte paper surfaces and doesn't scale from the white reference white. Interestingly, it makes little difference on glossy papers. It is only accurate if the print is illuminated with polarized light in the same orientation. That's possible to do but can't say I've seen it done.
Additional info:
I don't want to overstate opposition to M3. There are situations where M3 can produce better visual results:
I do believe it's likely to make better profiles (in a visual sense) with canvas. Especially canvas with enough texture that significant specular reflections are present. Canvas presents a particularly difficult profile target in shadows because specular noise is unavoidable. One approach could be to create a larger set of duplicate patches. Scan them and reject the higher luminance outliers but this is tedious and time consuming. Normal use of a single, if large, patch set is not sufficient. One needs to create at least 3, and likely more duplicate prints of the patch set. Further, averaging is sub optimal. It's the outliers that need to be rejected. Just one remaining outlier can easily bump a patch luminance enough to create strange gradient bumps in prints using that profile.
Matte prints with M3 profiles are more problematic - see the discussion of how they are illuminated and the diffuse light (white room) ambient. I suspect M3 provides better results in the latter situation. Especially when comparing to glossy prints in the same setting. The colors should be more consistent between the two in that environment.
Also, while cross polarization is used for M3, it's not necessary for viewing prints from them to use polarized glasses since the light that is reflected is already highly polarized unless coming off things like metallic surfaces. Polarizing the illuminant accomplishes most of the specular attenuation.
Creating profiles using a light integrating sphere spectro would significantly decrease the DMax of glossy prints and drop the profile gamut considerably but it would provide a more colorimetrically accurate print when viewed in the diffuse white room I mentioned in earlier. It's also highly impractical, slow, and expensive.
The ICC recommends not using M3 and I believe I understand their reasoning. However, I suspect it is at least a reasonable choice on some substrates such as canvas.
-
But illuminated "normally" they would exhibit poorer dynamic range and less saturation.
But they don't. That's the point
-
But they don't. That's the point
That's a question that can be resolved simply by analyzing the profiles with M2 and M3. Or colorimetric prints made with them viewed side by side. By definition, if the prints are illuminated in the same manner as profiles are made, that is from the side at 45 degrees, the reflectance of D50 from the prints will match the M2 profile uV effects aside. If the M3 profile differs then the illuminated print will not be the same using colorimetric intents (Perceptual and BPC realigns the tone curves but not the saturation reduction). How visible the difference is depends on the magnitude of the difference and the simplest approach is to compare the profiles.
-
I do not print on canvas but from my reading on LuLa, most printers coat canvas prints prior to framing. Does the use of such coatings mean that the profile which is likely done on non-coated canvas targets is not as accurate?
Doug Gray is correct about viewing conditions. I was down at my old office over the summer taking some light measurements of a series of prints that have been on display for 10 years. I wanted to get some Lux measurements to get an idea about print stability. I should have hung a color checker swatch that could be measured for fading but folks would probably have thought it was just some kind of op-art print. As I reported at the time, the prints still look as good as when they were originally hung. It's interesting that the display conditions, traditional office lighting, are illuminated at almost the 45 degree angle that Doug mentions.
-
That's a question that can be resolved simply by analyzing the profiles with M2 and M3. Or colorimetric prints made with them viewed side by side. By definition, if the prints are illuminated in the same manner as profiles are made, that is from the side at 45 degrees, the reflectance of D50 from the prints will match the M2 profile uV effects aside. If the M3 profile differs then the illuminated print will not be the same using colorimetric intents (Perceptual and BPC realigns the tone curves but not the saturation reduction). How visible the difference is depends on the magnitude of the difference and the simplest approach is to compare the profiles.
In my experience comparing numbers and 2D/3D diagrams of many profiles with their corresponding prints, I've come to the realization that numbers and diagrams only take you so far, then you need to look at the prints. The numbers and diagrams are useful so I use them, but prints are the final product - the rest is inputs. The print producer needs to care about the inputs, but not be mesmerized by them. My normal viewing environment is roughly in the range of D50/45degree to the light source, but not exactly, so my comparative vantage point is within a range of near consistency, and I've seen disconnects between what one may infer from data and what one sees on paper - sometimes rather glaring ones - so one needs to have a balanced approach to reliance on inputs and product in deciding what works best for the print of a photograph (not a patch chart).
-
In my experience comparing numbers and 2D/3D diagrams of many profiles with their corresponding prints, I've come to the realization that numbers and diagrams only take you so far, then you need to look at the prints. The numbers and diagrams are useful so I use them, but prints are the final product - the rest is inputs. The print producer needs to care about the inputs, but not be mesmerized by them. My normal viewing environment is roughly in the range of D50/45degree to the light source, but not exactly, so my comparative vantage point is within a range of near consistency, and I've seen disconnects between what one may infer from data and what one sees on paper - sometimes rather glaring ones - so one needs to have a balanced approach to reliance on inputs and product in deciding what works best for the print of a photograph (not a patch chart).
First, a comment about 2D and 3D gamut plots. To me they are eye candy and rarely useful. They tell nothing about Perceptual Intent, and nothing about how colors that are outside the gamut are mapped to printable colors. They are useful largely for just determining whether a color is printable and while of value for repro work it isn't useful for most photographic prints because even with Relative Colorimetric, one has to convert the color values to the scaled L* values actually printed. So, mostly eye candy. But I do use the 2d gamut viewer in PM5's profile editor (or more likely a Matlab routine) when a color I'm trying to print may be marginal. Usually I'm doing repro in those cases.
But as for instruments, I too have had unexpected "surprises." I love them! In large part because I come to printing from an engineering background where the most interesting questions, and often opportunities, show up when a discrepancy occurs between something instruments say should happen and what actually happens. That means the instrument readings are outside operating parameters, a mistake was made, or my understanding is wrong. Since it's usually the latter, especially doing something new, I become obsessive about understanding the discrepancy then verifying it.
Also, my earliest printing work was creating analog charts with superimposed sine waves to measure imager registration plane error to sub micron accuracy where printing controlled linearity was critical.
So whenever I get results that apparently differ from what instrumentation says, I dig into it until I understand why.
As an aside, My experience with canvas is limited. When I tried, years ago, to make profiles - and canvas prints - for some artist friends, the results were not pleasing. Especially in shadows. I was getting erratic results in the darker patches and wound up using Epson's canned canvas profiles. That has been the only media I have wound up having to use canned profiles. And they worked quite well. I suspect if I had M3 it would have produced better visual results even if at the cost of colorimetric accuracy. However, colorimetric accuracy is of value mostly in repro and printing proofs. Using BPC and/or Perceptual intent already discards colorimetric accuracy. Obviously not a big problem in the photographic world since Lightroom forces BPC.