[Tool] ChromIQ – a macOS and Windows GUI for ArgyllCMS printer profiling (v4.3.0)

[know]

Compatible Profiling Target

The patch layouts used by i1Profiler and ChromIQ differ. Attempts were made to rotate and mirror the target to achieve a layout that can be recognized by both applications, but this was unsuccessful.

The objective is to develop a common profiling target that is compatible with both i1Profiler and ChromIQ, allowing the same physical target to be used for profiling in both applications.

Any recommendations for achieving this?

Currently, ChromIQ supports doing the following Round-trip:

1. In ChromIQ, make the chart as usual. Then use Tools ▸ Convert TI1 →
   i1Profiler (or the chart editor's Save As), which writes -i1profiler.pxf/.txt
   in the patch order of the .ti1, with i1Profiler's own scramble turned off.
   Ticking "Also save a shuffled copy" spreads similar colours apart.
2. Load that patch set into i1Profiler. i1Profiler lays it out and prints it,
   and that print is the one physical chart.
3. Measure it once, in i1Profiler (handheld, i1iO or i1iSis), and build the
   i1Profiler profile there.
4. Export the measurement from i1Profiler (.mxf, .txt or .cxf) and bring
   it into the ChromIQ project. The measurement import
   (workflow/measurement_import.py, the "Add measurements" picker) was written
   for exactly this case ("a person prints a ChromIQ chart, measures it in
   i1Profiler … and wants the readings back"). It pairs each reading with its
   patch by order or by name, and refuses rather than guesses when the two
   disagree.
5. ChromIQ builds its own Argyll profile from the same readings.

The result is one sheet, one measuring pass and two profiles. What this
route does not give: ChromIQ's own measuring screen (overlay, strip checks)
for that sheet, because the sheet was measured in i1Profiler. For a scanner,
ChromIQ can already read an i1Profiler page back from the TIFF i1Profiler saves
(scanner dialog, "Where the chart was laid out: In i1Profiler", #120).

(steps written by claude)

Cheers,
Knut

 

[itsab1989]

@itsab1989 Thank you very much. Do catch up on your sleep and spend time with your family too. You more than deserve it.
The earlier errors with CMYK and .cal files have been resolved in 4.3.3 Beta.7, Also it is great to have the preventive message come up in a Calibration run if one tries to select any chart.
A new problem, perhaps because of this is now the chart will not be random even if random is selected. That setting is ignored completely (Calibration run only).
I am still a bit confused about the .cal part. Since my RIP cannot handle a linearization separately I have to do it in the profile. Is just creating a chart with the -K option (apply and embed) enough to be read and a profile created from it proper? or do i also have to apply the cal file in the "Calibration & Profiling" tab at the time of creating the profile?

Beta 8 made a few tweaks to this. Randomization gets locked for calibration charts because for those each ink ramp is on one strip. The help texts have been updated as well.

From claude:
With Apply & Embed (-K) you also click Apply Calibration after Build Profile, because colprof never puts the calibration into the profile. Our help text said the opposite and is corrected.

https://github.com/itsab1989/ChromIQ/releases/tag/v4.3.3-beta.8

Thanks for testing this and asking those questions by the way. This helps improving things.

 

[jctay]

Currently, ChromIQ supports doing the following Round-trip:

1. In ChromIQ, make the chart as usual. Then use Tools ▸ Convert TI1 →
   i1Profiler (or the chart editor's Save As), which writes -i1profiler.pxf/.txt
   in the patch order of the .ti1, with i1Profiler's own scramble turned off.
   Ticking "Also save a shuffled copy" spreads similar colours apart.
2. Load that patch set into i1Profiler. i1Profiler lays it out and prints it,
   and that print is the one physical chart.
3. Measure it once, in i1Profiler (handheld, i1iO or i1iSis), and build the
   i1Profiler profile there.
4. Export the measurement from i1Profiler (.mxf, .txt or .cxf) and bring
   it into the ChromIQ project. The measurement import
   (workflow/measurement_import.py, the "Add measurements" picker) was written
   for exactly this case ("a person prints a ChromIQ chart, measures it in
   i1Profiler … and wants the readings back"). It pairs each reading with its
   patch by order or by name, and refuses rather than guesses when the two
   disagree.
5. ChromIQ builds its own Argyll profile from the same readings.

The result is one sheet, one measuring pass and two profiles. What this
route does not give: ChromIQ's own measuring screen (overlay, strip checks)
for that sheet, because the sheet was measured in i1Profiler. For a scanner,
ChromIQ can already read an i1Profiler page back from the TIFF i1Profiler saves
(scanner dialog, "Where the chart was laid out: In i1Profiler", #120).

(steps written by claude)

Cheers,
Knut

I’m confident that your workflow should work.

I have a different approach in mind, briefly:

1. Generate the profiling target using ChromIQ.
2. Generate the TIFF file* from the profiling target.
3. Print the TIFF—the only physical chart—using the intended workflow application.
4. Perform the M0, M1, and M2 measurements using i1Profiler.
5. Export the measurements from i1Profiler and import them into ChromIQ for further processing and analysis.

* The TIFF file will be generated by ChromIQ.

 

[ajaytanna]

Beta9 already! You guys are amazing.
@itsab1989 Thanks for looking into this.
All in CMYK workflows exclusively with me.
Conventionally, linearization charts mostly appear sequential, so that's fine.
Questions? Sure, I have some more since things are getting interesting.
On the create chart tab, a chart created with a .cal applied with (Apply & Embed -K) is different from the same chart without the .cal applied or even with (Embed only -I). The difference is visible clearly. Measurements in Photoshop confirm C-M-Y-K 100-68-0-48 becomes 100-61-0-42. All colors are affected.
Naturally both of these charts are going to print differently and measure differently as well. So .cal is "baked" into the printed chart already. That's about as baked as it can get. It appears as if the application of the .cal file while creating the profile from that chart is simply applying it to an already "linearized" chart of patches. I have test prints that seem to support this intuition. What say you Santa Claude?

 

[itsab1989]

answer from claude:

@ajaytanna Your observation is right, and it is exactly what Apply & Embed (-K) is meant to do. The calibration is not applied twice, though. Here is why.

**What -K does to the chart.** The calibration curves are applied to the colour each patch is *printed* with, so on paper your 100-68-0-48 really is printed as 100-61-0-42. That is the "baked in" part you measured.

**What the measurement file keeps.** The chart's .ti2, and the measurement made from it, still list every patch with its *original* value (100-68-0-48), paired with the colour you measured from the calibrated print. So the profile ArgyllCMS builds describes "the printer **with** calibration": you send it 100-68-0-48, the calibration turns that into 100-61-0-42, and the printer prints the colour you measured.

**What Apply Calibration does.** It does not touch the chart or the measurement. It adds the same calibration curves to the profile's *output* side. When you later print an image through that profile, the profile produces a CMYK value and the curves then calibrate it, the same way the chart was calibrated. So on the way to the printer the calibration is applied exactly once: on the chart by -K, and on your images by the profile after Apply Calibration. Without that step, your images would go to the printer uncalibrated, while the profile assumes they are calibrated.

**When you must NOT use Apply Calibration:** when your RIP or printer driver applies the .cal itself. Then make the chart with Embed only (-I), and do not apply the calibration to the profile, or it would really be applied twice.

So for a RIP that cannot linearise: calibration chart → Create Calibration File → profiling chart with Apply & Embed (-K) → print → measure → Build Profile → Apply Calibration → use that profile in the RIP, with the RIP's own linearisation off.

If your test prints look as if the calibration was applied twice, check two things: that the RIP's own calibration/linearisation is off when you print through the calibrated profile, and that the profile you print with is the one after Apply Calibration (not the one from Build Profile alone). If it still looks wrong, I would be glad to look at the chart, the .cal and the profile.

 

[ajaytanna]

I just want to confirm. Two different charts are created from the same preset. One without -K and one with. This difference is visible before printing the charts. I have placed them side by side overlapped so the difference can be seen. Since they are different charts they will print and measure differently.

 

[itsab1989]

again from claude:

@ajaytanna Yes, confirmed, and your overlay shows it nicely: in each pair the -K patch is a little different from the one without (most visibly the cyan in 13 and the magentas in 3 and 4). That is the calibration curves already applied to what goes on paper. So the two charts print differently and measure differently, exactly as you say.

What matters is that they are two different measurements of two different things, so they must never be mixed:

* **The -K chart** measures "the printer **with** your calibration". Its profile is the one you then run Apply Calibration on, and that calibrated profile is the one your RIP uses (with the RIP's own linearisation off).
* **The chart without -K** measures "the printer **without** calibration". Use it only if you are not going to use the calibration at all.
* Don't combine readings from the two in one profile (no averaging or refining across them), and when you check or verify a profile, measure a chart made the same way as the one the profile came from.

ChromIQ keeps each one in its own run, so as long as you build from the run whose chart you printed, you are fine.

 

[ajaytanna]

"That is the calibration curves already applied to what goes on paper", says claude.
So I ask, why then is the need to apply it again? Claude is wrong here. It needs to reassess.

 

[itsab1989]

"That is the calibration curves already applied to what goes on paper", says claude.
So I ask, why then is the need to apply it again? Claude is wrong here. It needs to reassess.

answer:

@ajaytanna Fair challenge, so I checked it again against ArgyllCMS's own source code and documentation rather than repeating myself. The answer holds, but I explained it badly: the calibration is never applied twice to the same thing. It is applied once to every print that goes to the printer. The chart is one print; your images are other prints.

**Follow one patch through, with your numbers:**

1. **Chart (-K):** the patch is defined as 100-68-0-48. printtarg prints it calibrated, as 100-61-0-42. You measure the colour that comes out; call it X.
2. **Measurement file:** it records "100-68-0-48 gives X", with the *original* value. (ArgyllCMS printtarg.c: the page is drawn from the calibrated value `cdev`, while the chart file's rows are written from the original `cols.dev`.)
3. **Profile:** colprof learns from that: "to get X, send 100-68-0-48". In other words, the profile assumes the calibration will still be applied after it.
4. **Printing an image without Apply Calibration:** your image asks for colour X, the profile answers 100-68-0-48, and that goes to the printer *uncalibrated*. The printer prints what 100-68-0-48 prints on an uncalibrated printer, which is not X.
5. **With Apply Calibration:** the same curves are added to the profile's output, so it answers 100-61-0-42, the printer prints X, and it matches the chart.

So -K calibrates the chart print, and Apply Calibration calibrates every later image print. Each print is calibrated exactly once. Leave Apply Calibration out and only the chart was calibrated: the profile then describes a calibrated printer while your images are printed uncalibrated.

ArgyllCMS's own documentation describes exactly this workflow for a printer that cannot calibrate itself: apply the curves to the test chart with printtarg -K, and afterwards "the calibration curves will need to be applied to an image being printed either explicitly when using cctiff ... OR by creating a version of the profile that has had the calibration curves incorporated into it using the applycal tool" (Scenarios.html, print calibration).

**When it really would be twice:** if your RIP applies the .cal itself. Then use Embed only (-I) for the chart and do not use Apply Calibration.