Luminous Landscape Forum

Raw & Post Processing, Printing => Colour Management => Topic started by: rxchaos on June 05, 2020, 02:17:33 pm

Title: XRite iSis2XL - patch size and reference bar
Post by: rxchaos on June 05, 2020, 02:17:33 pm
i1Profiler sets the patch size for the iSis2XL at 6mm x 6mm.  I assume increasing the width of the patch size would give one a better averaging. but, would increasing the height make any difference?

And, if one were to measure targets more than once for better results would there be any benefit of reading the targets in reverse order and backwards.  If so, is there a way to print the reference bar on the bottom of the test chart as well. I suppose the patch set order can be appropriately reordered using PatchTool.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: JRSmit on June 05, 2020, 02:30:54 pm
There is a DIn/ISO standard that advises the aperture diameter plus 2mm around. In the case of a i1 pro2 or pro3std this is 4.5mm plus 2+2mm is 8.5mm square patch.
Since i use this size i have not experienced measure errors, also measure variations between multiple measurements was max 0.3dE, average much less.


i1Profiler sets the patch size for the iSis2XL at 6mm x 6mm.  I assume increasing the width of the patch size would give one a better averaging. but, would increasing the height make any difference?

And, if one were to measure targets more than once for better results would there be any benefit of reading the targets in reverse order and backwards.  If so, is there a way to print the reference bar on the bottom of the test chart as well. I suppose the patch set order can be appropriately reordered using PatchTool.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: HarveyM43 on June 10, 2020, 06:57:27 am
Doug Gray has done a lot of work describing the iSis' behaviour on the forum.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on June 13, 2020, 05:31:25 pm
i1Profiler sets the patch size for the iSis2XL at 6mm x 6mm.  I assume increasing the width of the patch size would give one a better averaging. but, would increasing the height make any difference?

And, if one were to measure targets more than once for better results would there be any benefit of reading the targets in reverse order and backwards.  If so, is there a way to print the reference bar on the bottom of the test chart as well. I suppose the patch set order can be appropriately reordered using PatchTool.

The only time increasing height makes a difference is if you have some sort of vertical slippage when feeding in the charts. Very rare. Since the vertical sensor width is fixed, there is no improvement is color accuracy increasing the height. Increasing the width, OTOH, does read a larger area and will improve accuracy. I've not found the improvement significant and found more variation based on where the patch is printed than the width of the patches. This is because there is typically a small variation in print head spacing as it goes across the paper.

As for reading the patches with top/bottom reversed. This is something I normally do. The order of the colors in saved CGATs files is simply reversed so is easy to manipulate. The reason I normally do this is that on occasion there will be a bit of dust on the paper. When reading dark patches a tiny piece of dust can increase the readings quite a bit. Like from .007 to .009 or more in a dark region. Reading the paper flipped results in slightly higher dEs than just re-reading the chart normally. This is because the printed area read is slightly different reversed.

You can cut and paste the registration bar from the top to create such a chart. Need to be very careful to align and space it exactly the same as the top bar.

My process automates chart generation and collating forward/reverse scans  with Matlab scripts. It spits out stats and alerts me if the forward/reverse scans differ materially. It also averages the data and saves a new CGATs file.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: rxchaos on June 15, 2020, 08:48:02 am
Doug,  thank you for the response.  I was cutting/pasting the registration bar to create a bar at the bottom to use when scanning the chart in reverse. I thought there may have been a way to do it in i1Profiler; I don't think there is one. I use PatchTool to reverse the order of the patches

Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on June 16, 2020, 10:24:40 am
Doug,  thank you for the response.  I was cutting/pasting the registration bar to create a bar at the bottom to use when scanning the chart in reverse. I thought there may have been a way to do it in i1Profiler; I don't think there is one. I use PatchTool to reverse the order of the patches

It's quite important to leave the same margin of blank space on the top and bottom after you place the black bar on the bottom. The i1iSis is quite picky about having enough blank paper on each end. The standard 957 (33x29 patches) may need to be shifted to equalize the top/bottom blank areas.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: rxchaos on June 16, 2020, 11:02:35 pm
Quote
It's quite important to leave the same margin of blank space on the top and bottom after you place the black bar on the bottom. The i1iSis is quite picky about having enough blank paper on each end. The standard 957 (33x29 patches) may need to be shifted to equalize the top/bottom blank areas.

Thank you.  That's probably why I was having trouble.  I'll retry paying closer attention to the top and bottom margins. 

Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on July 26, 2021, 05:18:43 pm
i1Profiler sets the patch size for the iSis2XL at 6mm x 6mm.  I assume increasing the width of the patch size would give one a better averaging. but, would increasing the height make any difference?

And, if one were to measure targets more than once for better results would there be any benefit of reading the targets in reverse order and backwards.  If so, is there a way to print the reference bar on the bottom of the test chart as well. I suppose the patch set order can be appropriately reordered using PatchTool.

While this is an old post, it aligns with something I'm currently finishing work on. And that includes printing i1isis compatible charts with registration bars on the top and bottom so the charts can be read in reverse.

However, the project has larger goals.

1. printing sets of reference colors along the edges of each sheet which are used to detect significant changes in printer inking during multiple page chart printing. I've done this because some printers, particularly the Canon 9500 II, and the Pro1000, have significant shifts between the first page and subsequent pages.

2. Combining different patch targets along with a random RGB set so that the random RGB values can be used to test relative accuracy and performance of profiles. To eliminate these variables and provide repeatable evaluation of profile targets with different printers, the the combination is scrambled across all pages. So any anomalies in printing/inking are distributed. This allows comparison of different profile patch set's performance without the variation that naturally occurs in inking from one page or run, to the next.

3. The program creates a combined CGATs RGB file and associated tif target images. A reverse CGATs RGB file is also created which can be used to read in the charts in reverse order and upside down.

4. After printing using ACPU or other means of printing w/o color management, the prints are scanned with an i1isis and the results saved in CGATs format. Optionally, reverse scans can be done and saved. These, if they exist, in the next step will be automatically compared with the forward measurement files. The forward and reverse patches will also be averaged.

5. After measuring the charts with an i1isis, the program will descramble and separate each RGB patch set and save them as individual CGATs measurement files.

6. Profiles for each of the profile measurement sets can then be made with i1Profiler or Argyll. This requires a license for i1Profiler. The other steps do not require an i1Profiler license as it can be used to measure patches w/o generating profiles.

7. Now program will evaluate the profiles against the random RGB measurement set. Since the random patch set is not correlated with the patches used to create the profiles, the result is a high quality evaluation of how well the different  profile patch sets work with the specific printer, settings, and paper. This makes it possible to objectively determine the sufficient, if not the best patch set for one's future needs.

8. The program can also be used simply to make a new profile by including only the desired profile patch set and random RGB patches for profile performance verification.

I have got it all working on my printers and am cleaning up the documentation and example files. Will be posting it to github within the next week or so.

Limitation:
Command line program running on 64 bit Windows 10.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: aaron125 on July 29, 2021, 02:09:39 am
The only time increasing height makes a difference is if you have some sort of vertical slippage when feeding in the charts. Very rare. Since the vertical sensor width is fixed, there is no improvement is color accuracy increasing the height. Increasing the width, OTOH, does read a larger area and will improve accuracy. I've not found the improvement significant and found more variation based on where the patch is printed than the width of the patches. This is because there is typically a small variation in print head spacing as it goes across the paper.
I’m not speaking from personal experience but in the short review of the Isis on Chromix’s site, they state that increasing the size of the patches well above the minimum size does indeed increase the accuracy of Isis/Isis2 devices. The following is a quote from their review:

“At CHROMiX, accuracy is of critical importance in our Maxwell service. The difference between a company's press being okay to run or being out of tolerance can be a fraction of 1 dE. While this instrument allows a minimum patch size as small as 6 mm, we have found that using larger patch sizes can improve accuracy and consistency in the measurements. Giving the instrument a larger patch to read gives it more information to work with, gives it more samples to use. For typical profiling purposes, the average patch sizes will do nicely, but when an extra level of accuracy is desired, such as in daily mission-critical color verification of high-volume presses, larger patches are a good practice. We often recommend going as high as 12 mm (almost 1/2 inch) in order to get a good balance between high accuracy and efficiency.”

Read the review here: Chromix Isis Review (http://www.colorwiki.com/wiki/Instrument_Spotlight_iSis)
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on July 29, 2021, 11:22:02 am
I’m not speaking from personal experience but in the short review of the Isis on Chromix’s site, they state that increasing the size of the patches well above the minimum size does indeed increase the accuracy of Isis/Isis2 devices. The following is a quote from their review:

“At CHROMiX, accuracy is of critical importance in our Maxwell service. The difference between a company's press being okay to run or being out of tolerance can be a fraction of 1 dE. While this instrument allows a minimum patch size as small as 6 mm, we have found that using larger patch sizes can improve accuracy and consistency in the measurements. Giving the instrument a larger patch to read gives it more information to work with, gives it more samples to use. For typical profiling purposes, the average patch sizes will do nicely, but when an extra level of accuracy is desired, such as in daily mission-critical color verification of high-volume presses, larger patches are a good practice. We often recommend going as high as 12 mm (almost 1/2 inch) in order to get a good balance between high accuracy and efficiency.”

Read the review here: Chromix Isis Review (http://www.colorwiki.com/wiki/Instrument_Spotlight_iSis)

I agree that larger patch sizes increases accuracy. At least increasing width does. Increasing thickness (vertical size) doesn't because the vertical aperture is fixed and stops before the 6mm vertical edges. The same amount of a patch is measured whether the thickness is 6mm or any larger size. I've measured this. However, I've also found that increasing width from 6mm to 12mm, while this does increase accuracy, is not quite as effective as simply duplicating patches and randomizing them. The paper usage is the same but the randomization additionally reduces the effects of location sensitivity which is significant. At least on the three printers I've tested.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: JRSmit on July 29, 2021, 11:55:21 am
On occasion I also use an isis, and found it difficult to get the paper go through it straight (not skewed). In such a case the height size of the patch do matter I think?
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: digitaldog on July 29, 2021, 12:30:29 pm
Nothing really new here; X-rite's idea of a minimum patch size should not be followed, for far more than averaging accuracy. Way, way to finicky. 
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on July 29, 2021, 03:09:23 pm
On occasion I also use an isis, and found it difficult to get the paper go through it straight (not skewed). In such a case the height size of the patch do matter I think?
The i1isis has registration diamonds on the left and right sides of the patch sets. It uses these to correct for feeding the paper in slightly off. It also corrects for slight shifts. You can see and hear this as the device stepper motor adjusts the vertical on eaqch horizontal scan. I've tested up to 1mm difference between the left and right. Height makes no difference. Usually. One difference I've found is on 13"x19" sheets if the i1isis isn't on a flat, smooth surface and especially when doing back hitches which are needed when doing M0/1/2 and not just M2. Never been an issue on 8.5"x11.
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: Doug Gray on July 29, 2021, 03:52:38 pm
Nothing really new here; X-rite's idea of a minimum patch size should not be followed, for far more than averaging accuracy. Way, way to finicky.

It's certainly safer to do larger patches and larger patches will produce more accurate color averages. I used to always do that until I decided to measure the actual effective aperture, both vertical and horizontal, and was rather surprised at how little extra area was captured with larger patch sizes. I also looked at feed through from adjacent patches. This is easily done with a black/white checkerboard pattern. No problem with 6mm patches. Except*.  So then I compared randomized patches where one set of N patches printed at 12mm/6mm was compared to a set of the same but doubled. The 2N randomized patches proved to have higher self consistency.  So now, when I want higher color accuracy, I use 6x6mm patches, double the patch count, and randomize.

* One thing I discovered was that a small amount of dust on the i1isis optical internals can create significant crosstalk from adjacent patches. This led me to putting registration bars on both the top and bottom. By doing a reverse scan and running a program that reverses, averages,  and compares the two, any registration errors become apparent before I use the measurements. I've taken to using a clean microfiber cloth to wipe prints before scanning and the incidence of problems has become quite small.

*device is an i1isis 2 XL
Title: Re: XRite iSis2XL - patch size and reference bar
Post by: JRSmit on July 31, 2021, 04:18:11 am
The i1isis has registration diamonds on the left and right sides of the patch sets. It uses these to correct for feeding the paper in slightly off. It also corrects for slight shifts. You can see and hear this as the device stepper motor adjusts the vertical on eaqch horizontal scan. I've tested up to 1mm difference between the left and right. Height makes no difference. Usually. One difference I've found is on 13"x19" sheets if the i1isis isn't on a flat, smooth surface and especially when doing back hitches which are needed when doing M0/1/2 and not just M2. Never been an issue on 8.5"x11.
Ah, clear. Thusfar only used 13” wide sheets. And the flat surface to put the isis on is indeed a must. The floor of the isis is rather flexible, a but of a surprise given the objective and cost of the device.