ChromIQ-- Delta E errors with i1 Pro 3

[hackensacker]

Hello to all here. I was able to pick up an i1 Pro 3 for a hard-to-resist price and was excited to try it out with itsab1989's nifty ArgyllCMS GUI. But for the life of me I can't get a profile without huge Delta E errors. Rows are verified in the scan (with several retries, because it's tricky to get the i1 to land cleanly on the paper white). I've turned off bidirectional scanning, turned it back on, and increased the patch consistency tolerance, all with the same results-- most if not all of the rows are way off, in the double digits. I had a similar issue with my Colormunki, which turned out to be because I didn't realize that ChromIQ defaulted to one-way strip reading with that device, but that doesn't seem to be the case here. Targets used were the i1ProTC9.18+Spyderprint Grays and i1Pro TC9.18+Extended Grays presets, both in letter size.

I ran i1Diagnostics on the i1 and everything passed.

ChromIQ log attached for anyone who might be willing to throw a life jacket to a drowning man.

 

[W. Fisher]

Just saw this thread as I have an i1 Pro 3 also. Today I had to sort out the TC9.18 as my default save folder was not the x-rite one, so I got that sorted and printed two pages of the TC9.18.

Soon as I saw the output I knew something was way off using Epson Ultra Premium Luster from an ET-8550. I tried the TC9.18 as I read where it had more B&W patches so it should produce a better B&W image, but sadly it didn't. As you say, the Delta E was in the double digits.

I had made a normal profile prior from the default or stock setup with the i1 Profiler making 3 pages of patches. At times, I also do the second iteration run that makes three more pages to further "fine tune" the profile, but sometimes it doesn't work out well either verses just the first-made run only. My play today with the TC9.18 was much as yours with the Delta E way too high. Why they differ so much is a mystery.

Color profiling is pot luck, imho. Change the spectrophotometer (I also have the older i1 Pro 2, and a ColorMunki Photo.) and they all produce something very different at times. I even get different profiles from Windows 10 and Windows 11 as well. I also have a couple of the flatbed scanner ones that operate as a cheaper colorimeter verses the better (or so I hope!) spectrophotometers: Print Prism and the SilverFast Studio 9 AI thing. I also got the printer makers and the paper sellers profiles, and even bought a set or profiles off a web maker to compare. ALL OF THEM produce something very different from the others if you examine the tints of a gray patch scale using same paper and printer. Some fail in blues becoming purples, or reds remaining red and not orange as they should be (e.g. ColorChecker Digital SG card in the M4 and M5 patch colors.), etc. Most all fail to some degree in the highlights to neutralize the paper's OBA base with a cyan, or sickly yellow/green (puke) tint, and then may often go off in the darker shadow side as well with some green, magenta, or blue tint in the blacks. Turn on or off BPC and you may see a strange color (Green for me.) show up near the last few dark steps to the pitch black one too. Only one that seems to correct for that mess is SilverFast since it does some ring-around thing with the highlights, mids, and shadows to neutralize the colors there, but it is tricky to do as a slight change of one of the three may affect the others if they were correct to begin with, imhe. The x-rite i1 Profiler can adjust for the OBA in the papers with the OBC mode in it that makes four neutral patches to compensate for the paper's base or the viewing light, but it's also tedious, imhe.

Quite maddening really, but I've learned to pick out the best one out of the lot and toss out the others for the given situation. Change paper or ink, and maybe the viewing light, and it all starts up again with an unknown outcome while hunting for the profile winning combination.

Good luck in your profile search!

W.F.

 

[pharmacist]

Interesting that it is both a i1Pro3. I have done the target which I have designed several times with both my i1Pro1 (the older GretagMacbeth Eye One Pro version 1) and the i1Pro2 and both generates good profiles that softproof well and prints without any problems with neutral greys on my ET-18100 (no grey ink, so difficult to reproduce good neutral greys without a good profile). Did you guys try to use the command style ArgyllCMS chartread and targen command to reproduce the profiles again so we can exclude some glitches by ChromIQ interface and maybe try AdobeRGB1998.icc as target profile instead of the built-ine ClayRGB1998.icc as reference target.

 

[hackensacker]

When I posted this I had tried at least five times to create a decent profile without success, so I felt pretty certain something was off somewhere. Afterwards I decided to give it another go and got a profile with the highest Delta E around 1 and the average a third of that (I did have to rescan a few strips). I've made a few other profiles since then, and they've all turned out well, so I assume the problem was me. There is a learning curve with the i1 Pro and I think I just developed a feel for where to start and stop the strip read. (As an aside, I find the ergonomics of X-rite products, from the Colormunki to the i1 Pro, pretty abysmal.)

I agree that it's hard to find solid ground with color profiling, as different devices give different results, and even the same device gives different results at different times.

B&W on the ET-8550 is always going to be tricky because of the black dye ink's spectral response. Even if you use the VFA setting for the pigment black, you still get cyan/magenta shifts in the middle grays depending on the light source (a 2700K LED bulb in my case). Print Prism actually does a nice job with this, and is easy to recommend as the entry point for anyone who wants to start making their own profiles. For a lot more money and time, the i1 Pro (using ChromIQ, which is great) is better-- not hugely, but noticeably. I made a profile with Pharmacist's TC9.18 + extended grays, and a preconditioned one with 1160 patches, using a Print Prism profile as the source. The prints from each look nearly identical to me, with improvements in those middle grays, and cleaner near-whites.

 

[W. Fisher]

Interesting that it is both a i1Pro3. I have done the target which I have designed several times with both my i1Pro1 (the older GretagMacbeth Eye One Pro version 1) and the i1Pro2 and both generates good profiles that softproof well and prints without any problems with neutral greys on my ET-18100 (no grey ink, so difficult to reproduce good neutral greys without a good profile). Did you guys try to use the command style ArgyllCMS chartread and targen command to reproduce the profiles again so we can exclude some glitches by ChromIQ interface and maybe try AdobeRGB1998.icc as target profile instead of the built-ine ClayRGB1998.icc as reference target.

The OP used Argyll with ChromIQ in hid attached text file which may have been what tripped him up. I cannot use Argyll as I have a USB port that feeds a multi-hub that runs around 5-6 other USB devices plugged into it and Argyll fouls that up to the point even the wireless USB mouse will skip with its USB transmitter plugged into it, or it loses the card reader. So I avoid the Argyll USB matter by not using it, but still even staying within x-rite and Windows 11 (ugh!) I get bad profiles at times with either the flatbeds or spectros plugged in. Often thought of using it on some cheaper laptop and not dealing with the USB switching matter and Argyll. Aside, I have ArgyllPro on my Android phone, but it differs in its readings with the same spectrophotometer plugged into it, versus the reading it gets using the x-rite software on the Windows computer and x-rite software. Go figger!

Granted, I may be super picky in that I look for any errant a* and b* hue in some L* known gray to be <0.9 if possible, and I have hit a perfect zero at times (Very rare, but perseverance pays off - sometimes.). Only one that allows me to get there is SilverFast over reading three set gray tones with the spectrophotometer, but it takes me a lot of time to sort out their highlights, mids, and shadow hues using their editor. I use that Red River printer test image that has the L=25, L=50, and L=75 gray patches to try and null them out as best I can. Others over-look trying to null out the highlights and shadows, so the colors may differ in the highs and lows using them, imhe.

W.F.

 

[Alan G]

I've been one of the beta testers ever since the app came out. I don't use any of the pre-made target sets and prefer to make a precondition profile and use that for the final profile using Argyll tools. I've used this method for over ten years and the app certainly streamlines things. If you are trying it out for the first time, don't waste a lot of paper. Just use a one page target and read that. If everything is fine you should get a decent profile with very low error rates. I have never seen a maximum error over 1.3 or so and the average errors are always low. the high errors are usually on dark black patches on matte paper which is to be expected.

If you see high error rates, it is usually a sign that you have not set up your printer color management correctly. You also need to check the settings for every page as the windows print utility that the app uses resets unlike the Dry Creek Photo Utility that I use when manually printing targets. I have an i1 Pro rev D and can use dense patch scanning that lets me put more patches on the page.

 

[itsab1989]

Just saw this thread right now. Have you tried your i1Pro with Argyll from the command line before? Was it better there? I few months ago I bought an i1Pro from eBay. It was supposed to be mint condition but it clearly was not so I returned it because it was not what I had agreed on. But I made a profile with it before returning (with Knuts Argyll printer Profiler script) and it turned out pretty bad as well. Especially dark colors looked horrible. I thought that I had done something wrong or it was because I had used the VFA media setting on my ET8550 but now I am not sure anymore.

 

[itsab1989]

From Claude:

Re: i1Pro3 profiles coming out bad (big ΔE numbers) — what's going on and how to fix it

Hi everyone,

Thanks for flagging this. I spent some time digging into how ArgyllCMS actually talks to the i1Pro3 (the regular one, not the bigger "i1Pro3 Plus"), and I think I can explain why some of you are getting profiles that look badly off — the kind where a quick quality check shows colour errors in the double digits instead of being nice and small.

I'll keep this jargon-free and walk through it step by step. There are basically two different things that can cause this, and they have different fixes, so it's worth checking both.

---
A quick bit of background (so the rest makes sense)

When you profile a printer, the workflow is:

1. Print a sheet full of little colour squares ("patches").
2. Scan that sheet with your measuring device (your i1Pro3).
3. The software turns those measurements into a profile.
4. You sanity-check the profile — if the numbers are good, errors are tiny; if something went wrong upstream, the errors are huge.

Here's the key idea: the profile is only ever as good as the measurements you feed it. If the device hands over wrong numbers, the software will faithfully build a wrong profile. So a "bad profile" is almost always really a "bad measurement" problem in disguise. With the i1Pro3 specifically, there are two ways the measurements can go wrong.

---
Cause #1: An older version of ArgyllCMS (the easy fix)

The i1Pro3 is a newer, unusual device, and support for it in ArgyllCMS is relatively recent. There was a real bug in older versions that did exactly this — it could hand back bad readings while scanning, which then poisons the whole profile.

The good news: it was fixed in ArgyllCMS version 3.4.0 (July 2025). The official fix note even says it was about the i1Pro3 "returning bad values" while reading a strip. And importantly, this bug only ever affected the i1Pro3 — older i1Pro, i1Pro2, and the i1Pro3 Plus were never affected. That lines up perfectly with the fact that it's only i1Pro3 owners noticing this.

So the very first thing to check: which version of ArgyllCMS are you running?

- If you're on anything older than 3.4.0, please update to the latest (3.5.0) and just measure your chart again. There's a very good chance that alone fixes it.
- You do not need to reprint anything or change any settings. The chart layout the software makes for the i1Pro3 is already the correct one — that part isn't the problem.

If you're not sure how to check your version, reply here and we'll help you find it.

---
Cause #2: How the i1Pro3 itself works (the trickier one)

This second part matters even if you're already on the newest version, so don't skip it.

Here's the thing that makes the i1Pro3 special — and a little troublesome. Older devices like the i1Pro2 had an actual little light bulb inside that shone a known, steady white light on the paper. Simple and predictable.

The i1Pro3 does not have that bulb. Instead it uses a cluster of LEDs (including an ultraviolet one), and then the software has to do a lot of clever maths to reconstruct what the colour "should" look like under standard lighting. ArgyllCMS does its best to copy what X-Rite's own software does here — but it's essentially an educated reconstruction, not a direct measurement.

The author of ArgyllCMS is refreshingly honest about this in the code's own notes. Paraphrasing: this reconstruction "would almost certainly not exactly match the results of the manufacturer's driver, especially for papers with optical brighteners."

In everyday terms:

- ArgyllCMS's i1Pro3 readings can be slightly off compared to X-Rite's official software, and
- it's worst on bright-white papers — the ones treated with "optical brightening agents" (OBAs) to make them look extra white. Most modern inkjet and photo papers are like this.
and photo papers are like this.

This isn't a bug someone can simply patch out — it's a side effect of the i1Pro3 having no real lamp and the support being reverse-engineered. On normal papers it usually shouldn't be terrible, but on very bright papers it can add up.

---
How to tell which one is affecting you

A simple test sorts this out:

- Measure the same chart again and check it against the profile you just made. If that shows big errors, it's a reading problem — most likely the old-version bug (Cause #1), or the scanner losing its place while sliding across the patches.
- Compare your i1Pro3 against a different measuring device, or print a familiar photo and judge the colours. If those look wrong but the re-measure check above looked fine, then it's the reconstruction quirk (Cause #2), and bright paper is probably involved.

---
Things to try right now

1. Update ArgyllCMS to 3.5.0. Honestly, do this first — it's the single most likely fix.
2. If you own the i1Pro3's "zebra ruler" (the striped guide strip the device slides along), use it when scanning. It's the i1Pro3's preferred, most reliable way to keep track of which patch it's on, and it avoids the shakier "guess from timing" method.
3. Try a paper without optical brighteners for a test. If your results suddenly get much better, you've found the culprit, and it's the reconstruction quirk on bright paper.
4. Spot-check a few patches one at a time (single measurements, not a sliding scan) and compare. If single measurements look good but full scans look bad, the device is losing its place during the scan rather than mismeasuring the colours.

---
To help us help you

If you're still stuck after the above, please reply with:

1. Your ArgyllCMS version (the big one!).
2. Where the big errors show up — when you re-check against your own chart, or only when comparing to something external?
3. What paper you're using (especially whether it's a bright/OBA photo or inkjet paper).

With those three things we can usually pin it down fast. Hang in there — in most cases this turns out to be the version fix, and you'll be back to clean profiles.

 

[Alan G]

I don't know if Graeme Gill has an i13 Pro to check what 'might' be happening. It could also be something as simple as the ArgyllCMS driver does not behave properly for this instrument. Unfortunately, I only have an i1 Pro and could not double check against the newer spectro.

 

[pharmacist]

For those without a working license for i1Profiler: I have made target that is for both ChromIQ and 1Profiler (pxf-file). The patche layout are both the same. Just scan with i1Profiler with your i1Pro3 and export in cgat txt file and use the ChromIQ utility to generate profile (license needed with i1Profiler but ChromIQ will do the job). This way you can compare both profiles (same patch layout) and exclude any problems with the built-in ArgyllCMS i1Pro 1/2/3 driver and the X-rite proprietary driver.