Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: rasworth on September 03, 2011, 01:36:41 pm
-
One of the first things I did with i1Profiler is create a target on a resin coated paper with moderate OBA content, scanned it with both 1Pro-normal and i1Pro-uvcut instruments, created identical parameter profiles, and made test prints. I could not tell any difference in the final print, and therefore at the time concluded there was no reason to use the uvcut instrument.
However over time I noticed an occasional light yellowish cast in high mid-tones using the i1normal created profile, that I did not detect during soft proof. So I went back, re-created the profiles using the latest version of i1Profiler, and this time did a critical comparison of the profiles for soft-proofing in CS5, perceptual mode with "Simulate Paper Color" and "Simulate Black Ink" both checked. There is a definite difference, the attached image is a screen print of a crop with proofing turned on, using the profile created with the i1Pro-uvcut instrument used in the top half and the one created with a i1Pro-uvnormal in the bottom.
I find that for my setup, for which I've tuned the monitor profile white point and brightness for print matching, that the uvcut version provides a better match. My conclusion is there is little or no difference in the profile perceptual forward path, but a definite difference in the back path.
Richard Southworth
-
Obviously my post didn't strike much of a chord with this group. However if one utilizes softproofing (View > Proof Setup >...) in Photoshop then it is a relevant issue. Xrite has stated they have enough "magic" in i1Profiler so that one can obtain good print profile results using an i1Pro-normal instrument on papers with OBAs. And as far as the perceptual/relative forward paths to the printer, i.e. Lab to RGB, it appears to be true.
But scanning an OBA paper (essentially all glossy/semi-gloss resin coated) with an i1Pro-uvcut instrument will result in a significantly different white point recorded in the profile than using an i1Pro-normal instrument. The different white point will generate a different absolute profile sub-section, and it is the absolute back path, i.e. RGB to LAB, that is used in Photoshop to render the printer colors back to the monitor. And at least with my eyes and my setup the softproof using a profile generated with an i1Pro-uvcut instrument is more accurate.
Richard Southworth
-
Hi Richard, I don't think it's that your very useful results do not strike a chord, I just think it's quiet because we're all over at Target and Walmart filling out job applications.
Tyler
-
Xrite has stated they have enough "magic" in i1Profiler so that one can obtain good print profile results using an i1Pro-normal instrument on papers with OBAs.
Where’s that? Seems they have reasons to be building products to measure both ways and have a software package for iSis users to account/adjust for OBAs in their profiles (maybe that’s what you are referring to).
The ‘best’ solution is the most costly: Have one device that measures with and without a UV cut, build a profile both ways. An iSis. One paper’s OBA content may vastly differ from another, so its possible that you could use cut or no cut on one, be just fine, not with the other paper. Of course, how you view those prints play a role.
-
The ‘best’ solution is the most costly: Have one device that measures with and without a UV cut, build a profile both ways. An iSis. One paper’s OBA content may vastly differ from another, so its possible that you could use cut or no cut on one, be just fine, not with the other paper. Of course, how you view those prints play a role.
I'm not sure it's fair to say that any approach is 'best.' While it's cumbersome to use, you can build profiles using ArgyllCMS and address the OBA issue. You only need a no-cut i1 Pro. Software is of course free but the learning curve is substantial.
-
From a recent FAQ in the XRite download section:
------------------------------------------------------------------------------------
PM5 Optical Brightener Correction -> How to achieve in i1Profiler?
Question: ProfileMaker 5 did include a software-based option for Optical Brightener Correction (OBC) at least for the Perceptual table in a profile. If the software detected optical brightener in the measurement data, a checkbox for OBC was enabled and could be switched on or off. This helped me to avoid color casts in a profile. How can I achieve this with i1Profiler?
Answer: The new i1Profiler software includes a completely new profiling engine, which provides already by default a most neutral gray axis and this effects on all Rendering Intent tables. This has also been confirmed for papers containing optical brighteners.
Additionally, i1Profiler provides a special option to enlarge the range of how far the neutralisation of grays goes into the more chromatic color areas: Profile Settings -> Perceptual -> Neutralize Gray. This allows to increase the neutrality of the gray axis further.
------------------------------------------------------------------------------------
And I agree with their statement as far as it goes, i.e. a non-uvcut profile seems to render an image to the printer identically as compared to a uvcut based profile. However, I typically invoke soft proofing on my images and then perform final edits for printing, therefore it's important to me that the back path rendering be as accurate as possible. And I don't believe they included Absolute in their group of rendering intents.
Red River Ultra Pro Satin measures about -3 on the b axis, Epson Premium Glossy about -4. Both of these values go close to zero when the paper is measured with an uvcut instrument, resulting in a very different white point, and therefore a different Absolute rendering (for those who may not be aware the Absolute rendering intent is calculated from the Relative intent, using the white point to appropriately bend the curves). I certainly agree the dual measurement and subsequent calculation using an Isis is the best solution, however there are still a lot of us slogging along with i1Pros, and I believe the uvcut version provides the best softproofing on papers with significant OBAs.
Richard Southworth
Added by edit - Alan, my statements were only in regard to i1Profiler, I agree that OBA compensation, such as I have with ProfileMaker, can do a reasonable job with an i1Pro-normal.
-
I'm not sure it's fair to say that any approach is 'best.' While it's cumbersome to use, you can build profiles using ArgyllCMS and address the OBA issue. You only need a no-cut i1 Pro. Software is of course free but the learning curve is substantial.
Yes but how does it deal with the data feed to it when you need uncut data and only have cut data? IOW, how and why isn’t the best approach to gather both kinds of data? OBA compensation is one thing, but its only as useful as the data collected.
-
From a recent FAQ in the XRite download section:
------------------------------------------------------------------------------------
PM5 Optical Brightener Correction -> How to achieve in i1Profiler?
Question: ProfileMaker 5 did include a software-based option for Optical Brightener Correction (OBC) at least for the Perceptual table in a profile. If the software detected optical brightener in the measurement data, a checkbox for OBC was enabled and could be switched on or off. This helped me to avoid color casts in a profile. How can I achieve this with i1Profiler?
Answer: The new i1Profiler software includes a completely new profiling engine, which provides already by default a most neutral gray axis and this effects on all Rendering Intent tables. This has also been confirmed for papers containing optical brighteners.
Additionally, i1Profiler provides a special option to enlarge the range of how far the neutralisation of grays goes into the more chromatic color areas: Profile Settings -> Perceptual -> Neutralize Gray. This allows to increase the neutrality of the gray axis further.
------------------------------------------------------------------------------------
All good marketing speak but it doesn’t address anything really. Its fairly obvious there is a totally new profile engine. In ProfileMaker Pro, there was a check box we could see when presumably that profile engine detected OBA’s (and whether cut or non cut spectral data was used should affect this) and the compensation was only available in the Perceptual table. The creation of that table is undoubtably different in both products. The only salient point I can gather from the above text is, the two products are different. In no way does this suggest that either product is such that having a dual mode instrument not necessary nor ideal. Again, my take is, the best solution is to measure both ways and feed that data to the profiler of your choice and examine the results. Then use the profile you prefer. In fact, a better solution is to measure this way and use the OBA module that thus far, can only be used with an iSis due to its mode of capture.
-
Andrew,
I'm not disagreeing with you, but my points have to do with those of us who don't have an ISIS, and therefore must do the best they can (again only referring to i1Profiler) with one or the other i1Pros. I believe XRite overlooked a salient point (softproofing) in their claim that the i1Profiler results tend to be adequate for OBA papers using only an i1Pro-normal.
Richard Southworth
-
In fact, a better solution is to measure this way and use the OBA module that thus far, can only be used with an iSis due to its mode of capture.
And I'll just be running by with my rant about X-Rite not enabling this for i1Pro owners. >:(
-
And I'll just be running by with my rant about X-Rite not enabling this for i1Pro owners. >:(
They can’t unless you have two units (one cut and one no-cut). They need both sets of data.
-
They can’t unless you have two units (one cut and one no-cut). They need both sets of data.
Andrew, please elaborate. How could something like PMP do it with a non UV-cut alone but now a newer (better?) package can't?
-
How could something like PMP do it with a non UV-cut alone but now a newer (better?) package can't?
OBA compensation (with some assumptions about OBA’s based on spectral data) in ProfileMaker Pro and the newer, far more robust OBA module (that has you print out and adjust the results visually) are not comparable. IOW, ProfileMaker Pro ‘didn’t do it’ the same way or produce the same results. The newer package can do it, it needs the specific data that dual measurements provide.
-
The specifics (I’ll let Marc from X-rite’s post on the ColorSync list speak for itself):
Well hello list,
I saw this little thread about the new optical brightener stuff from X-Rite and just had to jump in. Even though I no longer carry the ol' X-Rite business card, I'm pretty close to the technology and should be able to esplein.
First of all, this is not simple averaging of 2 data sets (shame on you Terry.....and you call yourself a scientist!). Also, this is a far cry from the Optical Brightener Correction that is in PM5. Having been the product manager for that product, I can say that I am 99.9% sure that PM5 applies OBC in the perceptual intent - not the others. Considering that the #1 thing that OBC impacts the most is the reproduction of paper white (absolute colorimetric), it seems as if the technology was never able to offer the full value of its potential.
I would also add that - even in its "perceptual form", there was no real input for this control. You either got it (or at least had the opportunity to get it) if the software thought you needed it, or didn't If the software didn't think you needed it. All in all, it worked (and works) "sufficiently", but is/was not something built to specifically compensate for the difference between a measured value an the "visual" value of a color, printed on a media, in a specific viewing condition.
So what the heck is different about this new OBC thing. For one, it uses a reference target. Like a color checker. Called a UV checker. It's kinda like a gray balance card. The key is that it doesn't fluoresce. What that means is that colors (grays) appear consistent, relative to the temperature of the viewing illuminant, no matter what type of UV content your viewing illuminant has. A constant, if you will.
The second interesting point is that the correction is specific. The workflow entails printing a target, like an ICC target, and measuring it with the iSis (UV incl+excl). Without giving too much away..... the software crunches the numbers and comes up with a new target that is then printed and compared to the checker, visually (because there's really to way to fully characterize the UV parts of this scenario with instrumentation that either costs less than $30,000 USD or weighs less than 30,000 lbs, or both).
So, as you can imagine, the printed patches look somewhat like the "checker" patches, but vary slightly in color ranging from "this is how it would look with UVin" to "this is how it would look with UVout". The user picks the visual match (or matches) and sets the appropriate params in the software.
At that point, the software will spool up a new set of data for your profiling pleasure. Then, when using this profile to transform your image data, you should see a greater similarity between the colors that the image is supposed to be (what's in the PCS), and what you visually experience.
Rogers's comment is right on the money about the press sheet - it is a 2 step process. Most high quality sheets have limited brighteners, so matching them should not be a big deal. For sheets that do have optical brighteners, you would probably want to run the 2 step process on the press side. Many times, press setups use 2 runs (1st for plate curves, 2nd for press char data), and I suppose that - with some trickery - it could be made to work with no extra "cost".
In any event, it's interesting stuff. If you're working with optically brightened media....it should be especially so.
Also, for those of you who do not know, I have found a home under the umbrella of the Color Management Group. Thanks again to all those who supported me in my short time away.
Best regards,
Marc
--
Marc Levine
Color Management Guy
marc@colormanagement.com
-
All very interesting, but not directly relevant to my original post.
I'm making a specific claim, for my setup I achieve a better visual match between print and Photoshop softproofing if I create an i1Profiler printer (in my case an Epson 3880) profile for a paper with OBA content using an i1Pro-uvcut instrument, as compared to using an i1Pro-normal.
My criticism of Xrite has to do with their hand-waving FAQ that states i1Profiler software somehow does an ok job on OBA papers regardless of not having an OBA compensation mode. Fairly obviously the dual scan process using an Isis looks to be technically a good solution, but there are still many of us using only i1Pros and therefore I believe this information may be useful to that community.
Richard Southworth
-
All very interesting, but not directly relevant to my original post.
Its relevant in explaining to Tony why its necessary to have dual measurements. The point is, you need both and you need to build the profile both ways and pick the winner. You yourself found this to be true:
One of the first things I did with i1Profiler is create a target on a resin coated paper with moderate OBA content, scanned it with both 1Pro-normal and i1Pro-uvcut instruments, created identical parameter profiles, and made test prints. I could not tell any difference in the final print, and therefore at the time concluded there was no reason to use the uvcut instrument.
You later found a preference for one mode over the other. You have two devices, you can build dual measurement profiles (although its more ideal to use the same Spectrophotometer ala iSis). A different paper with differing OBAs may produce opposite results. Bottom line, there’s a reason why some of us measure dual mode and examine the results profile by profile. And then we can tweak farther with the OBC module.
Lastly, it appears you are saying there is no difference in the print but rather the soft proof? Because if so, the area to examine is the preview tables in the profiles.
My conclusion is there is little or no difference in the profile perceptual forward path, but a definite difference in the back path.
That is something you should ask X-rite about. It doesn’t sound kosher, the two should match in an ideal world. I don’t know that we can attribute this to the instrument data provided (could be a bug in how the tables are being written).
-
The difference in the softproofing makes sense to me, assuming I understand what Photoshop is doing. Creating profiles with the uvcut instrument and the normal version results in profiles with different white points, primarily in the b field. The Relative gray balance curves appear virtually identical, using either Profile Editor or ColorThink Pro to examine. However the absolute curves are quite different, since they are derived from the Relative profile sub-section but with a white point adjustment.
If one checks "Simulate Paper Color" in the View menu, then I believe the preview choice from printer space to Photoshop is Absolute, and therefore the normal instrument based profile tends toward blue as compared to the uvcut based profile. Since I usually perform final edits on my images with softproofing turned on I was often seeing a yellow cast in the final print if I used the normal profile, because I was trying to tune out a bluish cast that didn't really appear in my prints when viewed under my (not very UV rich) illumination.
Richard Southworth
-
The difference in the softproofing makes sense to me, assuming I understand what Photoshop is doing. Creating profiles with the uvcut instrument and the normal version results in profiles with different white points, primarily in the b field.
Why no difference in the output? The tables should match.
If one checks "Simulate Paper Color" in the View menu, then I believe the preview choice from printer space to Photoshop is Absolute, and therefore the normal instrument based profile tends toward blue as compared to the uvcut based profile.
Yes, but again, the two should match.
-
No, the absolute curves shouldn't match, and they don't, per my attachment of the neutral rendering curves from both profiles.
As you well know, the normal use for forward absolute conversions is to match the white of one paper onto another for cross proofing, and since the i1Pro-normal instrument is "fooled" by the uv fluorescence the absolute curves show a dropped blue curve (in effect adding yellow ink) to match "real" white. However Photoshop has seen fit to use the absolute curves for softproofing, in effect trying to match the white of the paper onto the display. Of course it doesn't achieve a real match, since everything gets rendered up thru the monitor profile, but on a relative basis the difference shows up, per my original post.
The relative/perceptual/saturation curves all converge at the white point by definition, and therefore if one takes a constant image and prints it thru both profiles the result is essentially identical. It's only when softproofing is employed that the white point difference affects the displayed image.
Richard Southworth
Richard Southworth
-
No, the absolute curves shouldn't match
The color appearance should. The soft proof and output should match. You are saying they don’t. What’s going on with the AtoB and BtoA tables?
-
Forget print matching, all I'm saying is the softproofs from two profiles will have a slightly different hue, one derived from a i1Pro-normal and one from an i1Pro-uvcut, using the same printer target and both profiles generated from i1Profiler.
My understanding is the B to A path is via perceptual intent (or relative or sautration), and the A to B "return trip" is via absolute intent, if Display Paper White is checked. Since we have two differing sets of absolute neutral rendering curves, the softproof has to be different.
Richard Southworth
-
Forget print matching
I have from the beginning because if I understand you, you report they both match.
all I'm saying is the softproofs from two profiles will have a slightly different hue, one derived from a i1Pro-normal and one from an i1Pro-uvcut, using the same printer target and both profiles generated from i1Profiler.
Exactly. So where’s the disconnect? In the proofing tables. They should match the output. One apparently doesn’t. You feel, and you may be spot on, that the differences is due to how the data was collected. But I’m asking why there is this disconnect in only the one table.
My understanding is the B to A path is via perceptual intent (or relative or sautration), and the A to B "return trip" is via absolute intent, if Display Paper White is checked. Since we have two differing sets of absolute neutral rendering curves, the softproof has to be different.
You need the Absolute rendering for the display yes, but that doesn’t change the aim which is a match. Would you agree that the soft proof with the paper simulation should match the print? You seem to be saying they don’t. The question is why.
We use the simulate check box to see a more correct soft proof. With it off, I would not expect a good print to screen match. But with it on, there’s a mismatch but only from one profile and the only differences is the filter. I’d think both tables would be affected assuming the software is building both tables correctly.
-
Ok, I believe I understand your point, and I believe I "know" the answer.
Test prints made with both profiles resulted in essentially identical prints, as viewed under my fairly bright somewhere around 4000K environment. So my conclusion is the actual rgb values fed to the printer in perceptual intent (I didn't try much with relative or sautration) at least in the high mid-tones were also close to identical. The only way to do a truly "honest" print comparison would be to have a strong uv presence for the print done via the i1Pro-normal and a weak uv source (my actual viewing environment) on the print done via the i1Pro-uvcut. Assuming the illumination was otherwise identical I would have different print appearances, and I might well have good softproof results with both profiles.
But that's truly the issue, we don't generally view prints with illuminants containing strong uv. Almost every profile I've examined from paper manufacturers (at least for those papers with OBAs) was obviously created with a uvcut instrument, i.e. the white point was much less "blue" than anything I could detect using an i1Pro-normal. So given that we don't had the capability to do some sort of sophisticated mix of targets measured with and without uv detection, a uvcut instrument may usually provide the "best" solution, at least in terms of matching the softproof of papers with OBAs.
If I'm correct then this is as much a problem with PMP as i1Profiler, more an issue with the whiteners used in resin coated papers. And I never found much improvement using the OBA correction in PMP, the few times I tried it both ways I couldn't tell any difference. To be really honest, I've had past issues with softproofing in terms of the yellowish tendency, and it's only now using an uvcut instrument that I seem to have better results. Of course, that could be due to several factors, including my eyeballs.
Richard Southworth
-
So my conclusion is the actual rgb values fed to the printer in perceptual intent (I didn't try much with relative or sautration) at least in the high mid-tones were also close to identical.
Easy to test in Photoshop (better if you have ColorThink Pro). In Photoshop, convert an image both ways (two documents). Go to Image > Apply Image.
Set whichever image isn't listed as the target as the source. Set the Channel as RGB. Set the Blending to Subtract, with an Opacity of 100, a Scale of 1, and an Offset of 128.
If the images were truly 100% identical, every pixel in the image would be a solid level 128 gray. Pixels that aren't level 128 gray are different by the amount they depart from 128 gray. You can use Levels to exaggerate the difference, which makes patterns easier to see. Just move the sliders on either side to the center as far as you can.
In ColorThink you can do this on a low rez image (or better, a target), load and build two Color Lists. Save out Lab. Load both, run a dE report. You’ll see exactly what colors are the same or different.
-
Very interesting experiment.
First I performed for perceptual and relative intents, little difference, very narrow histogram, had to bring the level sliders way in to visually observe much difference, barely discernable near-gray tones at 100 left and 155 right.
I then converted the image (PDI-Target) thru both profiles using absolute intent, significant difference particularly in the higher mid-tones, bringing in the level sliders to 100 and 155 created a very visible difference image.
So I believe my conclusion was correct, i.e. profiles created using both instruments show very little difference in the forward direction (lab to rgb) for perceptual and relative intents, and a very distinct difference for absolute. This is consistent with my visual evaluation of prints made with both profiles, obviously very similar since the 3880 received nearly identical rgb values for every pixel. And it is consistent with the difference I saw in softproofs.
Richard Southworth
-
So I believe my conclusion was correct, i.e. profiles created using both instruments show very little difference in the forward direction (lab to rgb) for perceptual and relative intents, and a very distinct difference for absolute. This is consistent with my visual evaluation of prints made with both profiles, obviously very similar since the 3880 received nearly identical rgb values for every pixel. And it is consistent with the difference I saw in softproofs.
I would expect two different instruments to have slight differences. Even the same instrument measuring the same target twice will (if you compare the lab values in ColorThinnk). The differences should be pretty small (an average of less than 1 but worst could be 1).
But that they are different still doesn’t explain the mismatch in the tables.
-
What mismatch? Given the relative/perceptual tables are close, then the absolute has to be significantly different because the white points are significantly different. And if I were able to irradiate the uvnormal print with sufficient uv I might also be able to achieve a reasonable softproof match for the uvnormal version as well as the uvcut, and the two prints would look different even though the B to A tables were similar.
I don't see an inconsistency.
Richard Southworth
-
Why doesn’t the soft proof match?
-
Because of the white point "error". The profile made from the uvnormal instrument contains a white point that is bluer than the actual white point of the paper under any illuminant that doesn't contain significant uv. I believe one requirement for successful softproofing is a match between the profile white point and the actual paper white point.
I understand that the paper white point is a parameter that is normally independent of any viewing illuminant, but it becomes warped in the presence of uv.
Richard Southworth
-
Its relevant in explaining to Tony why its necessary to have dual measurements. The point is, you need both and you need to build the profile both ways and pick the winner. You yourself found this to be true:
And I do appreciate the explanation from both you and Marc. So to veer a bit off course from where this conversation is heading, where does that leave people like me who have a non UV-cut and are thinking about the i1Profiler upgrade under the current rebate period? Would we wind up with a solution that is not ideal? (assume I use papers with OBAs, which I do.)
-
So to veer a bit off course from where this conversation is heading, where does that leave people like me who have a non UV-cut and are thinking about the i1Profiler upgrade under the current rebate period? Would we wind up with a solution that is not ideal? (assume I use papers with OBAs, which I do.)
Compared to what? The “use or don’t use UV Cut” debate is old and on-going. Some say software compensation is useful. Some may say its not. You have hardware that is what it is. I don’t think its all that pertinent in terms of should you upgrade or not. If the new features, the new color engine or the ability to run in say Lion are useful to you, you’ll probably want to upgrade.
There are disadvantages to papers with OBA’s no matter what Spectrophotometer or software you use.