Will no decimals in i1Profiler XML files create anomalies

[Shane Webster]

I've been parsing through i1Profiler's XML files looking at their structure and contents.  In a quest to use PMP's MT to average readings together to then use in i1Profiler, I've noticed an oddity.  When i1Profiler creates a patch set, it creates its on XML file with a pxf extension, and a txt file that can be opened by MT and used by MT to generate a color chart.  Only thing is the pxf contains no decimal values but the txt file does.  Thus, the RGB value for my first patch is 109.29, 221 and 17 in the txt file and 109, 221 and 17 in the pxf file.  The pxf is truncating the number.  For instance, my patch 72 in the txt file is 236.79, 102, 255 and in the same patch in the pxf file is 236, 102, 255.  The truncation carries through all i1Profiler's XML files, including the appending of the RGB values to the measurement files and to its test chart files when looking at the txf and the txt files of the same test chart.  I have no idea if there is really any practical difference between a truncated value and a non-truncated RGB value, but they are different and I would at least expect values for the same patch set, test chart, etc. to be the same when i1Profiler creates both files.  Will/could the truncation create anomalies with i1Profiler profiles since presumable the patch set printed was based on RGB values with decimal points but the reference XML to which it looks (assuming it looks to the XML and not the TXT file) contains truncated RGB values?

[Scott Martin]

I agree with your concern and wish both i1P and ColorPort had support for 16 bit targets. That said, the current results without are fantastic.

What RGB process are you working with that you want to average? Looking forward to hearing back from you!

[Shane Webster]

When I build targets for my printers, I like to print two targets at a time, measure each, compare the two for any anomalies in reading the patches, average the two readings together and build my profile from the average. While I do like the profiles i1Profiler builds, I still must incorporate MT if I want to try and maintain my profile methodology.

[Scott Martin]

Inkjet printers then? With what spectro? Do you measure the same targets a few times to average instrument variance?

Do you see any visual or statistical improvements with this method?

Might be interesting to analyze and see how i1Profiler's smoothing plays into all of this.

[Shane Webster]

Yes, inkjet printers--Epsons. I used to measure with an i1Pro but wanted something more precise when I looked at the delta E of the same chart read twice (I don't remember what it was, I just remember thinking there must be something that would read more consistently). I bought an i1iSis, performed the same test and they were virtually identical (again, I don't recall the delta E, it was a while ago). As such, I am comfortable with any instrument variance because if there's any at all, it's very, very small. I'm more interested in print variances. I print and measure 2 targets and take the average to take into consideration print variances (it should also alert me to a potential read error if two patches are widely different from one another).

Statistically, I think taking an average of two readings yields a better (how much I don't know, but the additional time and effort it takes using the iSis is small) result because it smooths out fluctuations with the printer. Visually is it better...one of the reasons I first started looking at my i1Pro was because I didn't think I was getting accurate soft-proof results and was getting odd results when attempting to roundtrip a profile to check the accuracy of it (as discussed in Bruce Fraser's color management book--which I still have to pause a lot while reading so I can try to ingest it all).  I've been very pleased, visually, with my process using the iSis, but that's just me.

[Scott Martin]

So with your iSis have you tested your averaged two target system relative to a traditional single target method using i1Profiler? Where there visual differences? Did you use any tools to measure the final output from both profiles and spit out some delta E variation values?