Pages: 1 [2]   Go Down

Author Topic: i1Profiler - i1Pro-uvcut vs. i1Pro-normal  (Read 18247 times)

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #20 on: April 27, 2011, 12:40:33 pm »


The image that you see here is in sRGB space, the reported values are from the original file (ProPhotoRGB).
As you can see my impression of a bluish cast in the CM profile is confirmed by the data. I don’t know how the i1Pro UV Cut would behave, but asking several people I trust, I always got suggestion about buying an instrument with no filter, so that the “profiling program” could care of OBAs. Graeme Gill (ArgyllCMS author) recommends to use non-filtered instruments: http://www.argyllcms.com/doc/FWA.html
As I reported in another thread, the exact same thing happened to me when I used Argyll to do a profile using CM of Ilford Gold Fiber Silk.  When I first opened up the profile in PS to look to see the effect I was rather shocked and thought I did something wrong.  I did a second profile this morning using a smaller number of patches and the same thing, blue cast to the profile in PS.  Thing is though, it is a good profile when I print a test image.  However, I'm not sure how to get around the blue cast if I want to use it for soft proofing.  My understanding is that the -f switch in Argyll has no impact on CM profiles.  I guess I can check it easily enough with my smaller patch set.  Also, you can get a much better data set of papers at Ernst Dinkla's site:  http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #21 on: April 27, 2011, 12:57:06 pm »

Right - the blue cast when soft proofing. Add that to the list of reasons to measure UVex for fine art/ pleasing color situations.

What about metallic silver halide papers that don't contain and OBA/FWAs? Why to they profile better from UVex measurements? I keep pairs of devices (UVex and UVin) around for different purposes but I'm finding that real world testing suggests that UVex measurements are the way to go the majority of the time again in fine art/ pleasing color environments.
Logged
Scott Martin
www.on-sight.com

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #22 on: April 27, 2011, 01:00:55 pm »

Right - the blue cast when soft proofing. Add that to the list of reasons to measure UVex for fine art/ pleasing color situations.

What about metallic silver halide papers that don't contain and OBA/FWAs? Why to they profile better from UVex measurements? I keep pairs of devices (UVex and UVin) around for different purposes but I'm finding that real world testing suggests that UVex measurements are the way to go the majority of the time again in fine art/ pleasing color environments.
I guess the ArgyllCMS software cannot get rid of the blue cast; I'll have to post and ask the question.  I do know that the ColorMunki native software for doing profiles does not give the cast.  Ilford Gold Fiber Silk has some OBAs in it but not that much as the spectral curve does not show a big bump that some of the others do 440 nm.  I have not profiled a non-OBA paper with Argyll to see what happens and whether there is a blue cast or not.
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #23 on: April 27, 2011, 08:28:35 pm »

I solved the problem of the blue cast in the profile.  It was generated by creating the profile for viewing under D65 rather than D50 and only shows up when you have Simulate Paper Color checked (which you need to do).  When you prepare a profile under the default of D50, there is no problem.
Logged

PhilipCummins

  • Full Member
  • ***
  • Offline Offline
  • Posts: 133
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #24 on: April 28, 2011, 04:04:37 am »

Not according to this article: http://rmimaging.com/information/Chromaxion_Issue_1.pdf
It could still have a compensation for that blue or that blue may not affect the display calibration but it is not clear.

Note the article refers to the Spectrolino filters that they tested, which they "assumed" would be the same as the i1Pro ones which is incorrect. If they were serious, they could have simply removed the i1Pro filter head for analysis. My own examination (since I have both a UV-Cut and No-Filter i1Pro here) is what I said earlier - the centre aperture is clear, while the outer ring has either an additional purple-blue filter that is present (for UV-Cut) or not present (No-Filter). IMHO X-Rite just makes the same i1Pro, then programs the firmware and adds the filter (or not) as required. I'm due to do some tests to see if switching the No-Filter for the UV-Cut filter simulates UV-Cut appropriately which means you essentially can get both readings from a single i1Pro, however slightly less conveniently vs an Isis.
Logged

PhilipCummins

  • Full Member
  • ***
  • Offline Offline
  • Posts: 133
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #25 on: April 28, 2011, 04:17:15 am »

My only concern would be the effect of the uv filter upon calibration of the i1Pro just before monitor profiling.  I don't know how sharp the uv filter is, i.e. does it affect some of the visible blues, and therefore might impact the accuracy of calculating the blue primary.  I would think it would be a second order effect at worst, but I haven't really tested.

It doesn't since the filter only affects the internal i1Pro light source. Since you use a dark measurement to calibrate the i1Pro in Argyll the entire UV-vs-No-UV issue is avoided as the lamp isn't used to calibrate via the white tile. Other people have mentioned that calibration of monitors with and without the filter has made no difference so I'm inclined to believe them from what I know (owning both & having checked the filters out). I have to admit it would be nice if X-Rite just packaged both filter heads with each i1Pro, can't cost that much to include it...
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #26 on: April 28, 2011, 05:26:04 am »


I have to admit it would be nice if X-Rite just packaged both filter heads with each i1Pro, can't cost that much to include it...


I did wonder about that too. 5 Euro at most in production costs. Make it 50 Euro and it would still sell. If it is just the cone + glass disc inside there.

If an Eye 1 goes back to X-Rite for recalibration it would surprise me if they cleaned that cone and glass disc. More likely the part is replaced.

met vriendelijke groeten, Ernst

New: Spectral plots of +250 inkjet papers:

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
Logged

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #27 on: April 28, 2011, 05:33:17 am »

Note the article refers to the Spectrolino filters that they tested, which they "assumed" would be the same as the i1Pro ones which is incorrect.

The cure however works accordingly. It will be hard to measure the Eye 1 filter without taking cone and glass apart. Taking the cone off is easy.


met vriendelijke groeten, Ernst

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
Logged

Ethan_Hansen

  • Full Member
  • ***
  • Offline Offline
  • Posts: 114
    • Dry Creek Photo
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #28 on: April 29, 2011, 07:20:39 pm »

We finished measuring a range of monitors with over 60 different sensors. A synopsis is here on LuLa while the full article is on our site. As part of this exercise, we looked at 10 i1-Pro units. Four were UV-cut, six were not. There was no statistically significant difference between them in terms of emissive measurements. That is what you would expect, since the UV-cut model adds a yellow filter in front of the measurement lamp to cut the UV component of the light.

On the question of which version to buy for printer profiling, it depends on the software you use. Neither Monaco Profiler nor i1Profiler have software-based optical brightener algorithms for general measurements. i1Profiler has the OBC module, but this only works with the i1iSis. Profiles made from many papers with high levels of optical brighteners using either i1Profiler or Monaco can exhibit yellow casts to highlight areas where the software attempts to compensate for the blue paper tint read by the spectro. For either of these products, your best bet is to use the UV-cut i1.

Now, if you use either iMatch, ProfileMaker, BasICColor Print, Argyll, or any other profiling software that has fluorescent brightener compensation algorithms, I would opt for the normal, no-cut version of i1. All these products usually do a good job of detecting and compensating for optical brighteners. I do not understand why X-Rite dropped these algorithms from i1Profiler. Perhaps they felt the iSis provided the correct way of handling UV - and it does - and did not want to settle for an approximation that only worked 95+% of the time.

Another problem that crops up occasionally with the UV-cut i1-Pro is that X-Rite's software reports measurement values for wavelengths below 400nm. Who knows where these readings come from. The lamp filter effectively cuts off all light emission below 400nm. Nevertheless, a CGATS measurement file contains readings at 380 and 390nm. Most of the time these values are a fairly flat line from 400nm and down. Occasionally, however, you see huge spikes in the 380 or 390nm readings. This behavior appears with most legacy GMB instruments - files from a UV-cut icColor show the same thing. If the short-wavelength readings are sufficiently high, the profiling software happily introduces a color cast to highlight areas.

Making matters worse, this behavior is unpredictable. The instrument is measuring sensor noise at these wavelengths and the data files contain the results of X-Rite's extrapolation algorithms. Why X-Rite does not cut all data below 400nm for UV-cut instruments I can't say. I also do not know if this problem continues to i1Profiler. Measurement data files still show bogus short-wavelength readings but I do not know if the software ignores these data.

rasworth

  • Guest
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #29 on: April 29, 2011, 10:12:04 pm »

Ethan,

As stated earlier, I have an older (2nd generation) i1Pro basic and a brand new i1Pro uvcut.  Using Measure Tool I sampled a bare spot on Museo Silver Rag with both instruments, the spectrum pictures are attached.  Two questions:

1. Apparently I obtained a more well-behaved reading with the i1 uvcut, 380-400mm didn't spike up.  Will the remaining inaccuracies affect my profile generation, within i1Profile?  I'm asking because I did use the uvcut for the measurement scan, and all appears well.  Not sure why I chose it other than it was the newer toy, after reading your analysis I should have used the basic unit on such a well behaved paper.

2. Per my first post, I created two profiles on another paper with ob's, one with the i1basic and one with the i1uvcut, which yielded to my eyes identical results on test prints.  How can you be sure i1Profiler is not applying some degree of ob compensation?

Thanks for your detailed work,

Richard Southworth
Logged

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #30 on: April 29, 2011, 10:15:31 pm »

Now, if you use either iMatch, ProfileMaker, BasICColor Print, Argyll, or any other profiling software that has fluorescent brightener compensation algorithms, I would opt for the normal, no-cut version of i1. All these products usually do a good job of detecting and compensating for optical brighteners.

That compensation is only for the Perceptual intent and problems can occur when using the other intent. I'd argue that there's merit in using UVex data even for these app as well.

I do not understand why X-Rite dropped these algorithms from i1Profiler. Perhaps they felt the iSis provided the correct way of handling UV - and it does - and did not want to settle for an approximation that only worked 95+% of the time.

Maybe because it was an imperfect band-aid and they're rather work towards a more universal solution? The iSis approach is totally different and the variable nature of it is pretty cool. I suspect there's more to come on this front.
Logged
Scott Martin
www.on-sight.com

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #31 on: April 29, 2011, 10:18:25 pm »

I created two profiles on another paper with ob's, one with the i1basic and one with the i1uvcut, which yielded to my eyes identical results on test prints.  How can you be sure i1Profiler is not applying some degree of ob compensation?

Direct discussion with the developers confirms there is no OBA compensation. The difference can be visually slight but becomes more obvious when you a) compare papers with huge amounts of OBAs and b) when these prints are viewed under daylight with lots of UV and c) when you soft proof these files (which uses the other side of the profile). There is a difference to be seen but it won't blow your socks off.
« Last Edit: April 29, 2011, 11:28:12 pm by Onsight »
Logged
Scott Martin
www.on-sight.com

Ethan_Hansen

  • Full Member
  • ***
  • Offline Offline
  • Posts: 114
    • Dry Creek Photo
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #32 on: April 29, 2011, 11:15:51 pm »

Richard - Your graphs show exactly what happens with X-Rite UV-cut instruments. The i1Basic/no-cut plot on top shows the reflectance from the paper dropping off at wavelengths below - swagging it from the picture - 460nm or so. That's a typical paper spectrum for OBA-free stock. The UV-cut i1 plot shows a slight increase in measured intensity at 380 and 390nm. That's X-Rite's screwy algorithm at work rather than actual data. We have seen cases where the reported intensity exceeds 100% at 380nm.

[Quick digression: We build profiles for photo labs, re-profiling at regular intervals. Part of the service we provide is monitoring printer performance. We have tracked some printers for nearly a decade, so we have a good baseline to compare to. We also track the paper measurements. This catches the occasional bad emulsion batch and has been fed back to the paper manufacturers for performance analysis.

We tried using a UV-cut iCColor and our data integrity checks flagged the occasional paper white as being off. Our targets contain a number of repeat color patches, including white. The majority of the oddball readings occurred on some but not all of the white patches. Our first thought was a gross non-uniformity on the paper itself. Measuring on a Spectroscan disproved that idea and remeasuring on the UV-cut iCColor gave non-repeatable results. Finally we figured out that the short wavelength measurement data were a guess rather than reality. Conversations with GMB engineers (this was pre-X-Rite merger) confirmed our suspicions.]

If you are using a standard target with only one white patch, a white patch that reads as artificially blue will throw your profile off.

Scott: I agree that the iSis solution with the OBA compensation routine is very cool. Were it not such a hack and PITA to implement on a volume basis that would be better still. Nevertheless, there is a place for software-based OBA compensation. The original method GretagMacbeth implemented in ProfileMaker V4.something was good, the refinements in V5 made it better.

Yes, the iSis OBC method works better in most instances (except for some pathologically OBC-laden papers where the OBC target contains no neutrals). That is no reason to leave non-UV cut i1Pro owners with no means of compensating for OBAs or for those working in real-world environments where printing a second target and evaluating it is not in the cards.

Scott Martin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1315
    • Onsight
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #33 on: April 29, 2011, 11:27:13 pm »

Your graphs show exactly what happens with X-Rite UV-cut instruments.

All of them? Or are we just talking about EyeOnePros?

That is no reason to leave non-UV cut i1Pro owners with no means of compensating for OBAs or for those working in real-world environments where printing a second target and evaluating it is not in the cards.

I think the reasons will be revealed in good time. For the time being one might suggest that the improvements in profile quality outweigh the minor discrepancies seen without OBA correction. Others here have stated they can't even see these differences for example. And one might suggest that pros will always need to ability to generate both UVin and UVex measurements.
Logged
Scott Martin
www.on-sight.com

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #34 on: April 30, 2011, 06:28:38 am »

Direct discussion with the developers confirms there is no OBA compensation. The difference can be visually slight but becomes more obvious when you a) compare papers with huge amounts of OBAs and b) when these prints are viewed under daylight with lots of UV and c) when you soft proof these files (which uses the other side of the profile). There is a difference to be seen but it won't blow your socks off.

Within the condition of the thread subject: i1Profiler as the software an Eye 1 UV-cut model would be more suitable. In a sense so far it has been developed for that version of the Eye 1. You sketched the limitations of that development goal if the Eye 1 is the hardware to be used.

The illumination A lamp of the Eye 1s in general already reduces the FBA effect, a SpectroCam without UV filter delivers more. What exactly the iSis modes have for illumination is something I'm curious about, without UV light does it deliver true measured numbers from 380 to 730 NM, is the UV light going to shorter wavelengths than any other X-rite model?

Ethan sketched the other suitable combinations of profile creators and hardware and in that case the Eye 1 Basic without a UV filter will be more versatile and should be more accurate on the blue to UV range in all cases. Even more compared to a ColorMunki that is measuring a shorter range of wavelengths than the Eye 1 UV-cut but isn't shy to fill the wider gaps with arbitrary data.


met vriendelijke groeten, Ernst

New: Spectral plots of +250 inkjet papers:

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm





Logged

rasworth

  • Guest
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #35 on: April 30, 2011, 09:20:35 am »

Scott,

Yes, I understand the AnTOBn side of profiles, I compared soft-proofs as well as test prints of the differently generated profiles of a paper with OBs.  I can buy that the particular paper I used (RR Ultra Pro Satin) doesn't have huge amounts of OBs and therefore i1Profiler with no OB compensation might still produce similar results between the two profiles, if I have the energy at some point I'll do a dual scan of one that's loaded.

In the mean time I believe I will use the i1Pro basic for "good" papers and the uvcut version for others, seems to be the most logical course with what we know about i1Profiler.  I also have PMP5, but I really do like the profiling controls in i1Profiler, particularly the contrast, enables one to better tailor profiles to taste, particularly with fine art papers.

Richard Southworth


Richard Southworth
Logged

Jalok

  • Newbie
  • *
  • Offline Offline
  • Posts: 43
Re: i1Profiler - i1Pro-uvcut vs. i1Pro-normal
« Reply #36 on: May 03, 2011, 08:59:08 am »

The common conclusion in all the forums seems to be that UV-cut profiles (in what way they are created doesn't matter here) make prints that are generally more pleasing to look at under varied consumer lighting conditions. I guess that does not include going through photo albums on the porch. What surprises me more is that 99% of RC papers has FBA content, 20% deliver b -10 brightness. A lot of other paper qualities rely on it too. We do not get other papers (excluding the ones for archival reasons) and we seem to like what actually warps colour management, a profile made for a "neutral" paper while the paper varies from b -2 to -10. And I know from practice that even indoors FBA effect is seen with RC papers. More subtle profiling features like adapting the grey axis to paper white starts in that case with data that is already compromised. I am actually not questioning the consumer's eye sight or the experts' vision on it but question why this practice isn't discussed as being warped CM. The more I read and learn about CM, the more I get the impression that it isn't a hermetic system in its fundamentals (for example: Delta E is not the same throughout the space) and CM practice saws on the other legs of the stool.

Thanks, Dinkla, for your valuable words. I agree with all of them. But I like to add a variable that is absent to many of us who write in this forum. As I live in a country that has the highest importing/duty taxes around the world (around 80% excluding the shipping costs), it's almost always very difficult and very expensive to import good low-OBA papers. And as I don't sell my printings and I'm not a professional photographer, only a few printing jobs justify the time and money expenses in purchasing these good papers. So most the time I have to live with the best I can find in domestic market. You could imagine how valuable your spectral plots for inkjets papers are to me, which I was aware of only a few weeks ago, as they help to narrow choices on the first round. Thanks again!
« Last Edit: May 03, 2011, 06:43:04 pm by Jalok »
Logged
Pages: 1 [2]   Go Up