Possibility to convert csv-file with only L*A*B* values into a ArgyllCMS ti3 file

[Epatcola]

Thanks both for the work you have done on this device. I had a CR30 delivered today.

I think with care you could go smaller than 15mm square patches. The hole in the bottom of the CR30 is only 5mm.

If I do end up using the CR30 for printer profiles I think I would print a collar for it something like this

and a rule with notches matching the patch pitch.

 

[pharmacist]

I think printing a special mall that uses the adjacent patches as kind of reference to clearly define the position of the scanning hole is feasible, but since it can do spot reading only the whole process will take much more time then strip reading. However: the older Spyderprint3 also supports spot reading only and the 729 patch expert target would take more than an hour to read all patches, especially when including the 239 patch enhanced gray target too.

After several tests with @Ink stained Fingers we came to the conclusion that with a small patch set (likewise 192) the results are already very good. Don't forget the old GretagMacbeth package has a very small target using 45 patches to give you a rough printer profile, but that is a bit too small to my taste. With the larger 225 patch "High Quality Target" of the Spyderprint package there is a substantial improvement in print quality up to the point that is more than enough for most users, since modern printers are already very linearly predictable in their behaviour. My target has 228 for the standard filled and 247 patches for the completely filled A4 target. The patch size is large enough to place the CR30 without any collar to focus on an individual patch. But with smaller patch size it will become necessary in order not to read the colour of the adjacent or previous patch.

the python executable made by @x64 is very smart to use the ti2 instead of the ti1 file as a reference. Changing page size or regenerating the ti2 file will change the patch randomization by which the position of each patch size is configured, thus making it a smart choice to do so. Changing paper size or patch size and the ti2 file is changed accordingly.

 

[Ink stained Fingers]

This CR30 appears to be quite an interesting spectro; is there any calibration mode - or a reference tile as with the XRite spectros ?

Let me compare the profiles with a profile I did some time ago with the Action glossy paper on my ET-8550 running with Epson 106 inks; I was using an A5 patch sheet with 283 patches at that time.

These plots show how the profiles compare at a mid luminance at L*=50 as an horizontal cut with

light blue - I1Pro1 scan
orange - CR30 scan
dark blue - I1Pro2 scan

The overall gamut volumes are pretty much identical, there are areas where one or the other profile is just slightly wider

The overall look is the same at a darker luminance of L*=25 , we are here pretty much at the bottom of the gamut volumes, all 3 profiles are pretty much identical.

There are slight differences at the bottom tip - here at L*=15, the black points are between 10 to 12.

I think tha's quite an interesting overlap of these profiles - created with different instruments on different printers with (possibly) different inks.

The question remains what the impact is on prints - specifically for the differentiation of dark grays - I'm using a gray patch sheet like this one which directly shows me at which level gray differences become visible, and it is interesting to see the differences in print when you change the rendering intent and BPC - black point compensation - on or off

 

[pharmacist]

yes it has a white tile to calibrate on:

it is magnetically attached on the spectro.

 

[Epatcola]

I think printing a special mall that uses the adjacent patches as kind of reference to clearly define the position of the scanning hole is feasible,

With a collar like I suggested and a patch pitch notched rule you would only have to align the rule for each column the CR30 would just be slid along to the next notch.

yes it has a white tile to calibrate on:

My calibration patch reads 89.44 -0.71 0.12 (D65 source) is yours the same?

The documentation is pathetic.

I just noticed the back of the calibration cap also reads 89.44 -0.71 0.12. The device detects the magnet in the cap and button presses become white calibration operations. You would think it could do a black calibration at the same time by not turning on the LEDs but it doesn't.

A black calibration initiated in the windows application seems to be stored in the device and retained over power cycles.

 

[pharmacist]

ok today I tried a newer 247-patch target and unfortunately there is an error when executing colprof:

The message:
warning: colprof: Warning - gammap internal: input bounds has changed !....

@x64 Could it be there is a conversion error from mapping the ti2 file to the csv file to generate the ti3 file ?

Maybe you can have a look at those 2 files if you have some time and try to generate the target yourself:

 

[pharmacist]

With a collar like I suggested and a patch pitch notched rule you would only have to align the rule for each column the CR30 would just be slid along to the next notch.


My calibration patch reads 89.44 -0.71 0.12 (D65 source) is yours the same?

The documentation is pathetic.

I just noticed the back of the calibration cap also reads 89.44 -0.71 0.12. The device detects the magnet in the cap and button presses become white calibration operations. You would think it could do a black calibration at the same time by not turning on the LEDs but it doesn't.

A black calibration initiated in the windows application seems to be stored in the device and retained over power cycles.

Mine reads a white point of the tile: 91.22 -1.09 0.17

probably not a concern, as each tile is bound to each individual spectro.

 

[x64]

ok today I tried a newer 247-patch target and unfortunately there is an error when executing colprof:

View attachment 17603

The message:
warning: colprof: Warning - gammap internal: input bounds has changed !....

@x64 Could it be there is a conversion error from mapping the ti2 file to the csv file to generate the ti3 file ?

Maybe you can have a look at those 2 files if you have some time and try to generate the target yourself:

Something with the gamut mapping the source ICC, when running colorprof with no mapping (no -S/-s) I can create an ICC with that dataset (although it does not look correct).
I'm not sure what that target looks like, is there enough dark patches for contrast? the issue is mostly low contrast in the measurement set from what it seems like looking at the logs.

 

[pharmacist]

I will try to reread the target again, because maybe the problem is on my side. Normally the distribution of the test samples should be ok, as the targen.exe argument -G is enabled (good distribution). I will give it a try again and report back.

Meanwhile I could generate a profile using my Epson SC-P900 that can produce an exceptional good black point and has a particular mango-shape towards violet (because it has a violet ink increasing the gamut in this area giving a mango-shape pointed bulge in the blue/violet region and the magic is still there:

The CR30 spectro does registrates the very good black point see below:

And here the mango-shape bulge feature of the violet ink that is also well registered by the CR30:

Above target is generated with the 228-patch target. I will try to redo with a slightly larger target to see how it behaves. Note: 228 patches is more than enough for modern printers as those printers are very predictable in behaviour.

 

[Epatcola]

probably not a concern, as each tile is bound to each individual spectro.

I noticed the calibration cap has a serial number matching that which the device reports to the application. Our white calibration values differing shows the device has has a factory calibration for itself and associated calibration cap.