Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: Jeff-Grant on March 06, 2015, 07:58:47 pm
-
Last year I had my i1io converted to an i1io2, and also bought an i1Pro2. That all worked well until recently. I have discovered that the 1i1o2's measurements are all over the place. Using the i1Pro2 by hand, repeatability is good. My suspicion is that something is awry in the i1io2, Unsurprisingly, it passes the i1diagnostics. I'll attach a screen grab from CT Pro to show what is happening.
I'm using i1Profiler 1.5.6 with XRD 2.3.5.3 on a 10.9.5 iMac.
If anyone has any idea what's going on, I'd love to know.
-
If you are creating your chart with i1 Profiler, then the standard tile size is not big enough and will render a lot of reading errors. Even x-rite has admitted this.
-
I also find the i1iO2 finicky. I'm using i1Profiler v1.5.6, XRD 2.3.5.2 on PC, i1iO2 firmware version 1.07.
I'm now getting reliable results using chart created in i1Profiler with the default patch layout only when the following conditions are met:
- The paper surface texture allows the head to glide smoothly. If the paper is even just a little grabby, the head will jump slightly and the results will be unreliable. I found it impossible to measure certain smooth glossy papers.
- The head make good and consistent contact with the paper. If the arm rotation axis is not perfectly perpendicular with the media table, then the head angle relative to the paper will vary during reading and results will be skewed.
- The height adjustment should be set so that the angle between the head and the calibration patch is the same as the angle with the head and the paper.
I used my smartphone level app to measure angles and make adjustment as it was not calibrated properly out of factory. I also ran the head over fine sand papers (finished with 6000 grit) to adjust the white plastic piece so that it would be perfectly parallel with the paper and make a perfect seal. Out of factory it was a little skewed.
In scan mode the first four rows will read about 1% lower than the rest due to temperature stabilization issues. In spot mode the first row ramp up about 1% and then progressively drop 1% over the remaining rows. YMMV.
Does your head glide perfectly across the chart?
-
Jeff, your screen grab from CT Pro shows the i1io2 measurements to be very consistent, while the table of 21 measurements on top is the one with inconsistent measurement - which looks to be the handheld set. Can you confirm?
-
If you are creating your chart with i1 Profiler, then the standard tile size is not big enough and will render a lot of reading errors. Even x-rite has admitted this.
The chart that I'm using is a 21 patch grey scale, probably from the QTR package. I'm assuming that it's big enough as it works on the i1Pro 2. I have not noticed any issues with using standard patch sizes. I'll go back and take another look. Do you have a reference for the XRite statement? I'm talking to them in Australia so it would help if I can point them to something.
-
I also find the i1iO2 finicky. I'm using i1Profiler v1.5.6, XRD 2.3.5.2 on PC, i1iO2 firmware version 1.07.
I'm now getting reliable results using chart created in i1Profiler with the default patch layout only when the following conditions are met:
- The paper surface texture allows the head to glide smoothly. If the paper is even just a little grabby, the head will jump slightly and the results will be unreliable. I found it impossible to measure certain smooth glossy papers.
- The head make good and consistent contact with the paper. If the arm rotation axis is not perfectly perpendicular with the media table, then the head angle relative to the paper will vary during reading and results will be skewed.
- The height adjustment should be set so that the angle between the head and the calibration patch is the same as the angle with the head and the paper.
I used my smartphone level app to measure angles and make adjustment as it was not calibrated properly out of factory. I also ran the head over fine sand papers (finished with 6000 grit) to adjust the white plastic piece so that it would be perfectly parallel with the paper and make a perfect seal. Out of factory it was a little skewed.
In scan mode the first four rows will read about 1% lower than the rest due to temperature stabilization issues. In spot mode the first row ramp up about 1% and then progressively drop 1% over the remaining rows. YMMV.
Does your head glide perfectly across the chart?
Luc, thanks. That's a lot to think about and try. The paper that I am scanning is IGFS with the Inkjetmall Gloss Overcoat. I'll try some matte first. Perhaps it is the paper. I'll start there.
-
Jeff, your screen grab from CT Pro shows the i1io2 measurements to be very consistent, while the table of 21 measurements on top is the one with inconsistent measurement - which looks to be the handheld set. Can you confirm?
Apologies, the top set is with the i1io2 while the bottom set is hand scanned with the i1Pro2.
-
Luc, thanks. That's a lot to think about and try. The paper that I am scanning is IGFS with the Inkjetmall Gloss Overcoat. I'll try some matte first. Perhaps it is the paper. I'll start there.
Looking at your results, the differences are huge. So maybe none of my observations are relevant to your issue. The head jumping around on the paper issue can best be recognized by listening - it makes a distinct sound.
-
Thanks Luc. There's no noise. What is funny is that the i1io2 moves up and down the strip and then reads each patch separately. I would expect it to read in one scan pass.
-
Thanks Luc. There's no noise. What is funny is that the i1io2 moves up and down the strip and then reads each patch separately. I would expect it to read in one scan pass.
I've seen mine do that occasionally. I think when a row scan fails, the unit reverts to spot reading for that row.
-
Ok, going on the paper theory, I just tried out some HPR with a 21x4 patch set that I had lying around. It worked perfectly. Max dE is 0.4 with an average of 0.12. I've never seen this happen before. I guess that I will hand scan anything with gloss overcoat in the future.
Thanks for all your responses.
-
I never tried the QTR ramp, but when I make charts in i1profiler I get much larger DeltaE average than if I did my usual in EFIs software. And they use a larger patch size. It was my instructor who said xRite had admitted the problem with small patch sizes. Will ask if he can point to where.
Perhaps for fun, try a larger size. I think EFIs are around 8.5x9.3mm.
-
Will do, thanks.
-
Just out of curiosity, why profile the paper with a gloss varnish or coat over the ink? Is it to softproof? I can't see why you'd want to profile the paper with the coat on since it would degrade the image even further with a profile that's not the best the paper can achieve?
-
Got it in one. I'm only doing this for soft proofing using the QTR ICC profile build tool.
-
Last night I read my test chart containing 6 pages, saved each finished measurement and imported it in PM5's compare tool. I made 6 measurements, and comparing these, all had about average 0.8-1.4 I deltaE2000. I tried moving the paper on the table each round to see if there were something to do. As you said, the head angle isn't flat against the paper, and as it swivels the angle changes. Tried adding 6 papers underneath my chart to raise it, holding it stuck with tape, and it Acutally worked better.
Strange to think the angle from the head against the paper is wrong from the manufacturer?
-
By 'working better' do you mean that you got a better deltaE? I've often wondered about the io. With hand scanning, conventional wisdom is that you have a number of pages under the one you are measuring but that doesn't seem to be the case with the io. My understanding is that, to have multiple pages would negate the paper hold down and that you would need to raise the arm
-
Yes, better deltaE.
I taped the chart against the table with sheets underneath, and lowered the arm, so it was or less at a straight line instead of 30-40 degree angle.
-
That's interesting. I'll look at mine in the morning, too tired now.
-
Last night I read my test chart containing 6 pages, saved each finished measurement and imported it in PM5's compare tool. I made 6 measurements, and comparing these, all had about average 0.8-1.4 I deltaE2000. I tried moving the paper on the table each round to see if there were something to do. As you said, the head angle isn't flat against the paper, and as it swivels the angle changes. Tried adding 6 papers underneath my chart to raise it, holding it stuck with tape, and it Acutally worked better.
Strange to think the angle from the head against the paper is wrong from the manufacturer?
This has been an i0 issue from day one. It's a good device, it's no iSis in terms of what you'd have seen looking over 6 measurements made in a row. Anything over a dE of 1 is kind of high but what can you do? Measure the same target several times after rotating it, then average? Just of with it? FWIW, when measuring targets from presses, due to issues over the page/blanket, it's common, recommended to measure a target that's been printed in all four orientation and average. For inkjet and the like, not necessary. For a Spectrophotometer who produces errors as the head moves from it's base, that's kind of an issue and perhaps averaging would help.
You could also try much larger patches.
-
Andrew, I agree. The iO is, although convenient, yet it does degrade your measurements. I have increased my patches to 9,5x9,5mm.
I did an average from 6 readings, didn't trust just one measuring, but next time I think I will create a chart for the regular i1Pro2 och measure it by hand instead. Think it will have less deltaE than the iO.
-
One thing that I've noticed with the i0 after conversion and with the Pro 2 is that repeatability is good. I just went back and looked at a few old results. They all look good. On HPR, only one patch was 1.14 while all the others were well under 1. I'll attach the results from IGFS. This is with one sheet of paper on the io.
When I got the 3880, I did a lot of reprofiling. Initially, I was doing multiple passes and averaging with Profiler but after looking at the results in CT Pro, it didn't seem to need it. Maybe repeatability is another little detail that XRite neglected to mention. As always,I am happy to be corrected.