Pages: [1]   Go Down

Author Topic: Printer/Paper profiles: Quality of Instrument vs Number of Patches  (Read 1143 times)

aderickson

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 75

There is a thread in the Printing subforum concerning the quality of the profiling instrument when creating profiles. http://forum.luminous-landscape.com/index.php?topic=124934.msg1045713#msg1045713

I've rented the X-rite Colormunki and now own the IStudio (basically the same instrument) and used the Argyll software to create my profiles. I'm pretty satisfied with my results using 784 patches but I do have a bit of banding in some (not all) of Bill's Balls.

For my next paper profile I'm wondering whether increasing the number of patches will give me better shading, or am I just up against the limitations of my instrument.

Allan
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 11526
    • http://www.markdsegal.com
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #1 on: May 23, 2018, 04:57:45 PM »

The best to find out is to try.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #2 on: May 23, 2018, 05:24:49 PM »

There is a thread in the Printing subforum concerning the quality of the profiling instrument when creating profiles. http://forum.luminous-landscape.com/index.php?topic=124934.msg1045713#msg1045713

I've rented the X-rite Colormunki and now own the IStudio (basically the same instrument) and used the Argyll software to create my profiles. I'm pretty satisfied with my results using 784 patches but I do have a bit of banding in some (not all) of Bill's Balls.

For my next paper profile I'm wondering whether increasing the number of patches will give me better shading, or am I just up against the limitations of my instrument.

Allan

The banding in Bill's balls, because a large portion of them aren't in-gamut, tells little about the in gamut part of the printer profile but tells a lot about how the profile transitions from in-gamut to out-of-gamut. Changing the number of patches won't affect this very much. But it is significantly affected by the slider settings that apply to the Perceptual mode such as Colorfulness and Saturation. There is also a "smoothness" slider that affects the entire profile but it's probably best left alone unless experiments show a reason to change it. A high level of smoothness my improve banding but it may also slightly reduce colorimetric accuracy. To some degree banding reduction of OOG colors can come as a consequence of smoother, rather than abrupt, transitions at the gamut boundary.

Getting back to the Perceptual Intent sliders, these specifically change the way OOG colors are mapped to printable ones. Changing these sliders will have the most impact on how a profile renders Bill's Balls but they also change in-gamut colors a little because they affect in gamut colors somewhat as well.

As for numbers of patches, I get quite good results using the "optimize profile" option which creates additional patches that fill in areas I1Profiler thinks it needs additional info. I get as good results with 400 additional patches on top of say, 957, as with a much larger single patch set of 2033.  YMMV.
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Online Online
  • Posts: 3044
    • A Goldhammer Photography
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #3 on: May 25, 2018, 09:59:08 AM »

I am in Spain right now so it is hard to post links or give you my approach using Argyll.  You can make excellent profiles using the ColorMunki and Argyll.  The only drawback is the patches are big so it uses more paper  than if you have an i1 Pro.  There is a very good b tutorial written by Torger who had posted a lot here about camera profiling.   You should be able  to Google it easy enough.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #4 on: May 25, 2018, 10:23:32 AM »

I am in Spain right now so it is hard to post links or give you my approach using Argyll.  You can make excellent profiles using the ColorMunki and Argyll.  The only drawback is the patches are big so it uses more paper  than if you have an i1 Pro.  There is a very good b tutorial written by Torger who had posted a lot here about camera profiling.   You should be able  to Google it easy enough.
Good points.

The other extreme is using an iSis spectrophotometer where you can put 957 (or AFAIK a bit more with Colorport). on a US letter size sheet. The problem with the newer, non I1Pro spectros is that they are all uV cut types so you can't get, or simulate, M1 (full D50) to characterize OBA laden papers. However, that largely impacts Absolute Colorimetric and mostly impacts things like getting good soft proofs on high OBA paper. For those requirements the I1Pro series is the best choice. The iSis has a mode where it does a two pass approach with only uV. From there it calculates M1 and M0 from the additional uV. However, the two pass approach slightly degrades the accuracy of scans because the back hitches cause small changes in the physical registration. These are not normally material but can be problematic scanning larger sheets like super B's.
« Last Edit: May 25, 2018, 10:30:14 AM by Doug Gray »
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • Color Management Services
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #5 on: June 05, 2018, 03:01:24 AM »

The goal is to have enough patches for a good ICC profile. Also proper size of patches is important.
Putting to many patches (as many as fits) is nonsense. ISO 5-4:2009 states that patches needs to be +2mm larger then aperture.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #6 on: June 05, 2018, 02:47:14 PM »

The goal is to have enough patches for a good ICC profile. Also proper size of patches is important.
Putting to many patches (as many as fits) is nonsense. ISO 5-4:2009 states that patches needs to be +2mm larger then aperture.

I'm not about to use 3 pages to print targets that fit on one page, or 8 pages instead of 3 pages of 6mm square patches.

But initially, I shared your caution re the small, default iSis patch sizes. So I made a test to measure the optical iSis margins relative to adjacent patches and also made a test to measure adjacent patch light leakage. I was frankly surprised at how good it was, averaging around  0.01% which is like the difference between 100 Lux and .01 Lux or a DMax of 4.

I do have other concerns. The Isis leaves tracks from it's rollers on glossy and semigloss type papers. The iSis  uses thin, rubberized wheels for feeding paper. Big problem initially which, interesting, has become almost non-existent after about the first 3 months of use. Significant dE variation on reading pre and post track marks.

But they have pretty much gone away, I'm suspicious that the cause was plasticizer used in the wheel's traction rubber. It has gradually been either rubbed off or hardened into the substrate or simply isn't causing dE errors for some other indeterminate reason.

The next issue of significance is that the iSis illuminant is a white LED, not illuminant A per ISO,  and a bit of its shorter wavelengths (430-440nm) near the LED blue peak are absorbed and emitted at a slightly higher wavelength. So the reflected spectrum of high OBA paper differs slightly between I1Pros (Ill. A) and iSis/i1Studio/ColorMunki (Whilte LED). Also, the white LED has virtually no output below 430nm so apparently the instrument just extrapolates for M2 only scans. However, these are curiosities. The iSis has proven to provide highly repeatable measurements. The iSis light source is incredibly stable and there is much less variation reading charts with multiple passes. At least there is now since the tread mark problem has become insignificant.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • Color Management Services
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #7 on: June 08, 2018, 11:41:44 AM »

Quote
I'm not about to use 3 pages to print targets that fit on one page, or 8 pages instead of 3 pages of 6mm square patches.

Well that bugs me too, increasing the patch size even slightly makes page count double etc. because of the stupidity of the generation software. But you wan to be ISO compliant then increased page size is not avoidable imho.

The i1Profiler generates 821x1189 vs. 822x1190 measuretool.
The i1Profiler generates RGB patches by taking whole numbers while measuretool averages them like it should be.
The i1Profiler generates patches 1px x 1px smaller then measuretool (25x25px vs. 26x26px)

So whoever made the i1Profiler could not even copy same generation parameters.

Quote
But initially, I shared your caution re the small, default iSis patch sizes. So I made a test to measure the optical iSis margins relative to adjacent patches and also made a test to measure adjacent patch light leakage. I was frankly surprised at how good it was, averaging around  0.01% which is like the difference between 100 Lux and .01 Lux or a DMax of 4.

What patches was this tested with? Nobody provided me with clear statement what physical aperture size i1isis has so I can't say what patches are ISO 5-4:2009 compliant, in regard to i1Pro that has 4.5mm aperture that is 8.5x.8.5mm.

Quote
I do have other concerns. The Isis leaves tracks from it's rollers on glossy and semigloss type papers. The iSis  uses thin, rubberized wheels for feeding paper. Big problem initially which, interesting, has become almost non-existent after about the first 3 months of use. Significant dE variation on reading pre and post track marks.

Well I noticed that my isis has 2 sets of metal shafts with rubber o-ring wheels that drive the paper. Front has smaller diameter the the main shaft is larger. The rubber is red and clean when new. Now it's picket up ink or whatever and turned black. It is cleanable with isopropyl alcohol 99%. But it gets dirty quite fast. Why is this I don't know.
I wanted to replace it with silicone o-rings but did not find any hard silicone o-rings and soft ones will not work.

Quote
But they have pretty much gone away, I'm suspicious that the cause was plasticizer used in the wheel's traction rubber. It has gradually been either rubbed off or hardened into the substrate or simply isn't causing dE errors for some other indeterminate reason.

As you know o-rings are made from a mold that has 2 parts and in the middle is a visible mold separation line. That like looks visible on new o-rings. Perhaps this is why new isis leaves marks.

Quote
The next issue of significance is that the iSis illuminant is a white LED, not illuminant A per ISO,  and a bit of its shorter wavelengths (430-440nm) near the LED blue peak are absorbed and emitted at a slightly higher wavelength. So the reflected spectrum of high OBA paper differs slightly between I1Pros (Ill. A) and iSis/i1Studio/ColorMunki (Whilte LED). Also, the white LED has virtually no output below 430nm so apparently the instrument just extrapolates for M2 only scans. However, these are curiosities. The iSis has proven to provide highly repeatable measurements. The iSis light source is incredibly stable and there is much less variation reading charts with multiple passes. At least there is now since the tread mark problem has become insignificant.

I myself while agree that I1Pros (Ill. A) is better and ISO compliant way the isis LED is remarcable achievement in longevity and stability of the light source. Only the Konica Minolta FD-9 (isis on steroids in quality and price) is better imho in every regard.
The Barbieri LFP is another option if you do transmissible measurements only.

Now if CHROMIX or Babelcolor would make an actual chart generator that would be best move since Babelcolor made patchtool to edit the targets. Having to hack targets with photoshop and numerous actions is so 2007.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • Color Management Services
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #8 on: June 08, 2018, 11:56:44 AM »

There was a post perhaps on this forum by Dry Creek Photo that isis makes allot of errors while measuring with i1profiler early versions. I can't find it but was it ever fixed?
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 14138
    • http://www.digitaldog.net/
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #9 on: June 08, 2018, 11:57:35 AM »

Nobody provided me with clear statement what physical aperture size i1isis has so I can't say what patches are ISO 5-4:2009 compliant, in regard to i1Pro that has 4.5mm aperture that is 8.5x.8.5mm.
It's really not difficult to find such info!

https://www.xrite.com/categories/scanning-instruments/i1isis-2

Aperture:   Effective measurement aperture during scanning is depending on the patch size.

Patch size IS important! I'd strongly advise NOT attempting the minimum patch size creation in i1P with an iSis for multiple reasons!
Logged
Andrew Rodney
Author “Color Management for Photographers"

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 14138
    • http://www.digitaldog.net/
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #10 on: June 08, 2018, 11:59:26 AM »

There was a post perhaps on this forum by Dry Creek Photo that isis makes allot of errors while measuring with i1profiler early versions. I can't find it but was it ever fixed?
Having owned and worked with an iSis and i1P long before release, that's all news to me. There's plenty X-rite needs to fix, this isn't one of them IMHO and based on a lot of data....
Now the i1 i0 and errors as the arm moves away from home position? Yes, that's a reality.
Logged
Andrew Rodney
Author “Color Management for Photographers"

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #11 on: June 08, 2018, 02:55:09 PM »

Having owned and worked with an iSis and i1P long before release, that's all news to me. There's plenty X-rite needs to fix, this isn't one of them IMHO and based on a lot of data....
Now the i1 i0 and errors as the arm moves away from home position? Yes, that's a reality.

I quite agree. I've only had an iSis 7 months but have extensively tested it. I've measured effective aperture and posted the results here. I've also measured cross light contamination with a black/white checkerboard patter so each black patch is surrounded by white patches and the cross contamination of light is very low, less than 1/5000 for 6mm patches and 1/10000 for 7mm patches. It doesn't have one aperture except vertically, where it is about 3mm. Physical aperture is irrelevant because the iSis' measurement head does not contact the paper. Light, were it not focused near center, would go beyond the physical aperture.

I tossed my IO. It would often produce bad scans due to registration errors and making the patches large enough to overcome the positional uncertainties required a lot of paper or running lots of scans and tossing the bad readings. The I1Pro 2 is quite good. Easy to position over patches and fairly repeatable results. Most of the variation the I1Pro 2 has is from the tungsten lamp. It also benefits from using a position quadrature which tracks to 0.5mm so smoothness of manual scanning is much less critical than the older I1Pro.

The only problems I've seen with patch size and the iSis occurred when scanning 13"x19" sheets in full M mode where it has to backhitch. The back and forth motion easily introduces registration errors if the paper isn't on a flat surface over the full travel. For instance, if one end starts to droop over the end of a table just a few inches large errors can occur. The iSis also requires larger patches using any media that has any stretch to it like canvas and most canvas is too thick for it anyway. Also, good idea to make the patches larger if printing using Adobe's ACPU on Windows since it shrinks the print about 3%.  At minimum patch size it's right on the edge of getting kicked out by the iSis.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • Color Management Services
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #12 on: June 09, 2018, 07:51:31 AM »

The isis problems were reported by Dry creek Photo was it http://forum.luminous-landscape.com/index.php?action=profile;u=810 or somebody else on newsgroup I do not remember. They said that the problem was spotted by accident because they used their own targets with error detection patches built in for statistical purposes. This is how the error was found. They said they reverted to using measuretool for reading the targets.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 14138
    • http://www.digitaldog.net/
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #13 on: June 09, 2018, 09:26:10 AM »

The isis problems were reported by Dry creek Photo was it http://forum.luminous-landscape.com/index.php?action=profile;u=810 or somebody else on newsgroup I do not remember. They said that the problem was spotted by accident because they used their own targets with error detection patches built in for statistical purposes. This is how the error was found. They said they reverted to using measuretool for reading the targets.
I don’t care what they reported. I work with dozens upon dozens of iSis units on site for a client ALL tracking process control of individual press. ALL measurement data was correlated to ensure each unit had an acceptable matching tolerance. And they did. I attribute your report as FUD! Ignore it and please stop suggesting this is a common issue, it’s not IF all the iSis units are of the same rev!

« Last Edit: June 09, 2018, 10:00:57 AM by andrewrodney »
Logged
Andrew Rodney
Author “Color Management for Photographers"

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #14 on: June 09, 2018, 11:04:29 AM »

There was a post perhaps on this forum by Dry Creek Photo that isis makes allot of errors while measuring with i1profiler early versions. I can't find it but was it ever fixed?
I have not heard of this. But even assuming there was an iSis issue with early i1Profiler s/w it obviously has been fixed. The device is widely used and a problem like that, unfixed, wouldn't show up only on some post you vaguely remember but can't link to. It would be widely reported.

Also, many of the patch set scans I do have numerous, duplicate, patches to gather statistics on printer settings such as vacuum, head spacing, single/dual pass, resolution, ink load. A side benefit is being able to detect unusual scan readings.

Also, I make my own scan charts with a registration bar on both the top and bottom which lets me read the chart with the left/right and top/bottom flipped.

These redundancies allow frequent tests on iSis consistency and I have never seen any issue.
Logged

smilem

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 272
    • Color Management Services
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #15 on: June 14, 2018, 04:52:56 PM »

Quote
Also, I make my own scan charts with a registration bar on both the top and bottom which lets me read the chart with the left/right and top/bottom flipped.

Interesting idea, I guess that is one use for patchtool to make flipped source files for software to read flipped charts.

Quote
duplicate, patches to gather statistics on printer settings such as vacuum, head spacing, single/dual pass, resolution, ink load

I wonder what kind of patches would detect any of these things you list?
To detect the necessity of same scrambled patches in different locations (2 versions of the chart) I think simple sheet wide patches or series of patches that seem as a long one for each color a printer has plus RGB would show variance in ink on paper.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Online Online
  • Posts: 1350
Re: Printer/Paper profiles: Quality of Instrument vs Number of Patches
« Reply #16 on: June 14, 2018, 06:28:18 PM »

Interesting idea, I guess that is one use for patchtool to make flipped source files for software to read flipped charts.
I don't use patch tool for that. I have an automated process for reading forward and reversed cgats files, collating the results, and flagging any readings that differ significantly.
Quote
I wonder what kind of patches would detect any of these things you list?
To detect the necessity of same scrambled patches in different locations (2 versions of the chart) I think simple sheet wide patches or series of patches that seem as a long one for each color a printer has plus RGB would show variance in ink on paper.
I'm referring to duplication of patches within the same chart. Typically, 10 to 20 duplicates of a set of 45 to 90 different colors. The max and distribution of the standard deviation of the same colors is sufficient to identify optimum printer settings.

I'm not sure why setting the vacuum level makes so much difference in the variance of the same color patches but I'd guess it has something to do with variations in the head to paper distance as it feeds. Whatever it is it significantly improved the repeatability of color values once I optimize if for the paper.
« Last Edit: June 14, 2018, 10:12:22 PM by Doug Gray »
Logged
Pages: [1]   Go Up