Pages: [1]   Go Down

Author Topic: Rate this profile  (Read 1512 times)

danstart17

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
Rate this profile
« on: October 21, 2020, 03:52:42 pm »

Hi

I've just created a profile with i1Profiler and taken a look at it in Gamut Vision but it's looks a bit strange compared to ones I've created in ArgyllCMS.

Just wondered what your thoughts were on it?

Here's the link: https://drive.google.com/drive/folders/1Ha8X0MTy6iAGh-HI25MTm1oiDbVsuYzB?usp=sharing
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Rate this profile
« Reply #1 on: October 21, 2020, 03:55:15 pm »

First, not all gamut viewers are created equally or show the same mapping:
http://www.colorwiki.com/wiki/Color_Management_Myths_26-28#Myth_26
So, using the profile on a color reference image, HOW DOES IT PRINT?
Proof's in the print.  ;)
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

danstart17

  • Newbie
  • *
  • Offline Offline
  • Posts: 29
Re: Rate this profile
« Reply #2 on: October 21, 2020, 04:24:13 pm »

Hi

Thats fair enough.

I've not actually used it to print yet, I measured the chart tonight while at home and I'll do a print test tomorrow when I get into work.

I thought there might be some magical tool that could tell me if it's any good while not at work.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Rate this profile
« Reply #3 on: October 21, 2020, 08:35:12 pm »

Hi

Thats fair enough.

I've not actually used it to print yet, I measured the chart tonight while at home and I'll do a print test tomorrow when I get into work.

I thought there might be some magical tool that could tell me if it's any good while not at work.

I looked at the profile. It's shape is typical of I1Profiler profiles with 3k patches. Black point is around L=5 and White point of L=95. But, as Andrew noted, you really can't tell how goo a profile is just examining its shape.

One thing I did notice is that the AtoB transition along the device neutral axis is rather lumpy and from RGB 0,0,0 to 35,35,35 the L* increases very slowly. This might indicate too much ink. You can test this by printing a neutral gradient from L*=0 to 30 or so. If you don't see a smooth gradient or there is a shift in hue it's an indication the inking is too heavy for that paper. However, 3k patches will overcome some degree of non-linearity so you might well be fine. If your printer has adjustable ink loading you might get somewhat better results is you reduce the inking a bit but you would then have to re-profile. I wouldn't bother though unless you see problems in the dark shadows of prints.
Logged

JRSmit

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 922
    • Jan R. Smit Fine Art Printing Specialist
Re: Rate this profile
« Reply #4 on: October 22, 2020, 02:35:48 am »

Hi

I've just created a profile with i1Profiler and taken a look at it in Gamut Vision but it's looks a bit strange compared to ones I've created in ArgyllCMS.

Just wondered what your thoughts were on it?

Here's the link: https://drive.google.com/drive/folders/1Ha8X0MTy6iAGh-HI25MTm1oiDbVsuYzB?usp=sharing
This profile is not ok.
What patch size did you use?
It looks to me as some patch read errors occurred. Perhaps also not enough drying time. Also posibility you moved the i1Pro2 too fast.
To avoid patch read errors tha patch should be the measurement opening plu2 2mm around, so for a I1Pro2 (or 3 Std) it is 4.5 plus 2x2mm is 8.5mm
The drying time should really be 24 hrs or so.

Logged
Fine art photography: janrsmit.com
Fine Art Printing Specialist: www.fineartprintingspecialist.nl


Jan R. Smit

JRSmit

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 922
    • Jan R. Smit Fine Art Printing Specialist
Re: Rate this profile
« Reply #5 on: October 22, 2020, 07:00:34 am »

Somehow this profile is also incomplete, blackpoint tag is missing, chromatic adaptation tag is empty.
What settings of i1Profiler did you use?
Logged
Fine art photography: janrsmit.com
Fine Art Printing Specialist: www.fineartprintingspecialist.nl


Jan R. Smit

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Rate this profile
« Reply #6 on: October 22, 2020, 02:33:07 pm »

It's an ICC V4.3 hence no bkpt tag, with 37x37x37 3DLUTS (Highest I1Profiler accuracy) 16 bit.

However, the profile curves are very distorted but I1Profiler fits them reasonably well with an average dE2000 of 0.40 measured LAB v AtoB PCS LAB. Normally this is around 0.2. Interestingly, looking only at slices through L*, the a*b* sections don't look awful. But a line chart of the profile indicates multiple large areas of distortion indicating the profile is working very hard trying to linearize the printed colors.

No idea what is causing the distortion.  Really bad inks not designed for this printer or mix up of inks. Really quite weird.

Here's a line graph grab from I1Profiler:
« Last Edit: October 22, 2020, 02:36:29 pm by Doug Gray »
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Rate this profile
« Reply #7 on: October 22, 2020, 11:11:37 pm »

On a hunch I checked the profile made from I1Profiler using the ICC profile embedded RGB spectral patch measurements against a new profile made from the same data and settings.  It should match. It doesn't. Comparing the AtoB Lab values of both profiles from the same 3000 patch RGB values the deltaE2000 is .37.

This is typical of I1Profiler's bug where a patch set is generated (3000) and immediately used to print targets and scan measurements with a spectro. I1Profiler incorrectly handles the fractional data created by the generate patch process. In some cases it truncates the data and in other cases it rounds it.

Always, Always save any new RGB generated patch data first as an RGB patch set in the default format. Then reload the patch data. It will no longer have fractional values and I1Profiler will then generate correct prints, tiff files, and profiles.
Logged

JRSmit

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 922
    • Jan R. Smit Fine Art Printing Specialist
Re: Rate this profile
« Reply #8 on: October 23, 2020, 02:46:01 am »

I do not use v4 icc profiles, so I was unaware that is why the blackpoint tag is not there.

I agree with the statement to always use a saved patch set.

And would add to also use patch sets with near neutrals. In i1Profiler these are f.i. 1733 or 2033 or 2371 patches.
This together with patch sizes of at least 8.5 mm square, profiling with a i1pro2 or i1pro3 std has become very repeatable and reproducable.
Logged
Fine art photography: janrsmit.com
Fine Art Printing Specialist: www.fineartprintingspecialist.nl


Jan R. Smit

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Rate this profile
« Reply #9 on: October 23, 2020, 10:55:33 am »

And would add to also use patch sets with near neutrals. In i1Profiler these are f.i. 1733 or 2033 or 2371 patches.
This together with patch sizes of at least 8.5 mm square, profiling with a i1pro2 or i1pro3 std has become very repeatable and reproducable.

Yep, totally agree. Extra near neutrals greatly improves profiles. -Doug
Logged

Lessbones

  • Full Member
  • ***
  • Offline Offline
  • Posts: 170
Re: Rate this profile
« Reply #10 on: October 26, 2020, 03:07:04 pm »

Yep, totally agree. Extra near neutrals greatly improves profiles. -Doug

Is there some kind of formula you guys have come across to get to those patch numbers, or is that more a trial and error thing?  I've had some favorite numbers in the past, but it's always seemed really weird to me that a "professional tool" like i1profiler wouldn't expose to the user the methods behind adding and subtracting near neutrals...
Logged

JRSmit

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 922
    • Jan R. Smit Fine Art Printing Specialist
Re: Rate this profile
« Reply #11 on: October 27, 2020, 03:33:10 am »

Is there some kind of formula you guys have come across to get to those patch numbers, or is that more a trial and error thing?  I've had some favorite numbers in the past, but it's always seemed really weird to me that a "professional tool" like i1profiler wouldn't expose to the user the methods behind adding and subtracting near neutrals...
Yeah, the formula is : change the number of patches and observe what happens. Also known as trial and error. Just joking.
You are right, the logic xrite uses in i1profiler for the near neutrals in patch-sets is an unknown. Nor do I know if there is a better solution for near neutrals.
But sofar these patch numbers seem to work for me. Having said this, I currently use 2371 patches for measurements with i1pro2/3. Spread over 4 pages at max width and patch size 8.5x8.5mm.


Logged
Fine art photography: janrsmit.com
Fine Art Printing Specialist: www.fineartprintingspecialist.nl


Jan R. Smit

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Rate this profile
« Reply #12 on: October 27, 2020, 04:44:46 pm »

Is there some kind of formula you guys have come across to get to those patch numbers, or is that more a trial and error thing?  I've had some favorite numbers in the past, but it's always seemed really weird to me that a "professional tool" like i1profiler wouldn't expose to the user the methods behind adding and subtracting near neutrals...

The I1Profiler's "Smart Patch Generator" automatically creates near neutrals to fill in gaps. To get maximum near neutrals it's best to use a count that is just before it rolls over into a larger grid spacing. For example an 8x8x8 grid spacing with the maximum near neutrals added is 581 patches. There are 512 patches in the grid and 63 near neutrals. I1Profiler always adds 6 solid color patches and the end. They are there to provide a quick density quality check to detect gross error.

You can see visually where the near neutrals are added and when they roll over for a higher grid count.

Here's a partial list of max near neutrals patch counts:

517
581
653
734
815
905
1005
1105
1215

2553

4357

Logged

Ethan Hansen

  • Full Member
  • ***
  • Offline Offline
  • Posts: 148
    • Dry Creek Photo
Re: Rate this profile
« Reply #13 on: October 27, 2020, 11:15:44 pm »

Is there some kind of formula you guys have come across to get to those patch numbers, or is that more a trial and error thing?  I've had some favorite numbers in the past, but it's always seemed really weird to me that a "professional tool" like i1profiler wouldn't expose to the user the methods behind adding and subtracting near neutrals...

I did an analysis of optimum patch counts for i1Profiler some years ago. The RGB patch generator hasn't changed, so these remain a good starting point.
Pages: [1]   Go Up