Argyll CMS printer profile workflow update

[pharmacist]

Roy, From my brief experience of the results of both CM and Argyll, I think you would be equally happy with either.

Hi Emu,

At the moment am using the colormunki for my print profiles and very happy with the results using IS inks. Also, it is quick and easy to use, and produces excellent profiles despite the small number of scanned patches..

I had been previously using the Spyderprint 4.2.3 and was not happy with the results.

I am reluctant to change to Argyll and endure the learning curve unless I can be reasonably sure it produces better profiles.

From what I am reading so far the difference is so small as to be almost unnoticeable.

Roy

Hi Roy,

actually the advantage of the ArgyllCMS method that reading is done in a single step. The profiles are equally good, better than the CM method (using 2 optimizations) in the grays. Is it difficult: not really. I made a batch file called makeprofile so I get an interaction for input for internal name, file name etc instead of using cryptic descriptions.

Out-of-gamut colours are bit better reproduced perceptually compared to the CM-profiles (more vivid). ArgyllCMS allows you to create profiles for others and not bound to the software license of X-rite. Also the two-step method of method makes it very inconvenient to make profiles of others as second target is depending on the readings of the first target.

I had a SpyderPprint too before and the results are very inconsistent with the dreaded blue-to-purple shift in the profiles. Despite the hue-manipulation of the target does eliminate this problem at the cost of making purples more blue...

 

[mikling]

Just this week, I bit the bullet and got an i1pro2 spectro and decided to give Argyll a shot. So far I am impressed. I should be, cause that spectro cost an arm and a leg. I had a Datacolor as well. When using pigment ink, Datacolor is not bad but when used with dye inks, you need to master the editing function within the software otherwise your results will not be good. It took me a while to figure this aspect out.
For the little experience I've since had with Argyll, this is what I have noted. There are two methods of generating targets. One involves printing patches out, measuring them and then generating the ICC. The other is to produce an ICC as in previous BUT use that profile as a base from which to generate a second set of targets. In comparing the results between the two, I am sold on the second method thus far.
My first foray was to generate 2310 patches with 256 greys. The results were quite good but I could see some issues involving the B&W linearity with radial tests. The flesh tones were bang on. Never seen better and there was just a touch of some casts in the gray ramp in certain spots.
I then used the initial profile as the base, generated a second set of targets again with 256 grays and 2310 total. With this profile, the results were much better. The radial patterns were near perfect and the shadow detail had improved. Stunning considering it was a decade old Epson R200 that still showed grain due to dot size.
I am now trying to determine how many initial patches would suffice to get good results when used as a base before creating the second target set.
I have not played around with the FWA functions as yet. However, all my profiles will now be made with Argyll and posted on my site soon as time permits. All of them will have at least 2310 patches in the data set. I hope to have profiles and each with specified light conditions.

Now what I don't know is if the increased performance over the Datacolor is principally due to the top end spectro from Xrite or the software itself. It doesn't matter anymore. I'm sold on the combination.

One thing I noted was the tremendously long compute time Argyll took to generate the profile compared to Datacolor which generates profiles near instantly.

 

[mikling]

Just ran another set of conditioning profiles for a Canon MP500. It's old now but keeps chugging along. The profile was a random 882 patch set on two pages set up for the i1. Let it dry and compared it to the Epson R200 printed with the much larger set of targets in the profile.
Colors matched very well to each other. Slightly slightly more warmth to the Canon MP500. The Canon did not display as much grain. ( Grain only to be seen with magnifying glass) but here is where the key differences were. Shadow detail.The Epson R200 is easily visibly superior especially on B&W. Again, the colors match in general but the difference was in the details.
I will run another set with the MP500 using the 882 set as a conditioning run. This set will have the same parameters as the Epson R200. After this run, the differences will be to the machine and no longer to the profile software and hardware.

 

[mikling]

On another run, I produced a conditioning profile for the Pro-100 and despite the limited number of patches, it already beats the Epson R200. This might be testament to the advances in the Pro-100. Already, it easily is the best output I have seen this printer produce. Next run on the Pro-100 is the standard 256 grey targets and 2316 patches.

 

[pharmacist]

Hi Mike,

Could please write down and share the two steps method with us ?

 

[pharmacist]

Hi Mike,

Could please write down and share the two steps method with us ?

 

[mikling]

My initiation into Argyll was vastly aided by crenedecotret . He suggested that I might want to try the two step.
I was able to complete it on two printers but I was unable to do so during the targen on the third printer, the Pro-100. It did not lock up but it was taking forever. It was iterating like crazy. I don't know the cause of this but I will attempt again at another time.

Here is what the conclusion of the MP500 test is....the two step improved it very slightly. The shadow details were not markedly improved and here the Epson R200....despite the coarser granularity was superior to the newer MP500...same engine as the iP4200 and MP530. I will post them on my website for anyone to try. So after a certain improvement, the machine design comes to the front. I am actually very surprised at how well the R200 did in the end.

OK, here's the two step as suggested by crenedecotret.

" Now here is another trick... you can TRY to squeeze a bit more out of your 3 or 4 sheets of paper by using a pre-conditioning profile. A pre-conditoning profile would be a profile "somewhat close" to what you are profiling... ex: same printer and ink, but a slightly different paper. Or am OEM profile for the same printer/paper but let's say after market ink. The conditioning profile is simply used for targen to get a rough idea of what your printer can do and help generate more appropriate, useful patches within your "patch budget". I originally thought that this might limit the results, but according to Graeme it does not. Recently I bought some real inexpensive Kodak paper (cheap, nasty stuff) to use for throwaway prints for a school project my 8 year old son was working on. It was practically unuseable. I decided to profile it... I used a profile for Kirkland paper to pre-condition and only used one sheet of patches. The profile was very good. The prints were not awesome, but there were quite useable (no color casts, useable contrast, etc) . Soft proofing was spot on.
To use pre-conditioning, replace the targen command above with this, then continue through the regular workflow.
targen -v -c [precondition-oldprofile.icc] -d2 -G -g128 -f## -e4 [name]
"
Now for those that own the Munki, and were wondering what gives. Here's the lowdown. Just like Argyll, Argyll initially assumes an sRGB shape and appropriately generates patches for it. Here is where the divergence of the two systems take over. Argyll's author is more technically oriented and wants to do it the proper way...a bit geeky? Munki is targeted for the general consumer who does not want to be exposed to the details. I think Munki's optimize routine is actually the same strategy as the Argyll. Xrite's iprofiler is the same algorithm but adds GUI and a few other twists.
The conditioning profile essentially subsequently make each target a meaningful one in the second patch generation. If you produced targets not knowing which ones count, you could be producing a lot of throwaways. It does not reuse the intial patches. Munki on the other hand takes a few more but retains the ones already measured.

If you are not afraid of command line and came up the years with DOS, Argyll is not a problem. If you want the Munki instrument to go places it was never meant to go and Xrite does not want it to go, try Argyll out. It can be as simple or as detailed as you want it to be.

http://www.argyllcms.com/doc/FCMS2010_ArgyllTute.pdf

 

[mikling]

Well I made another run with the Pro-100. All I can say is WOW. What I did was use a prior profile I had for the printer as the precondition profile and then generated 2310 patches with 256 gray iterations. The result is that the shadow details in B&W is so there. In the Northlight test images for B&W, there is absolutely no black crush into all corners of the cave now. Nothing and I mean nothing is compressed. The Pro-100 is superb when combined with great profiles that Argyll can make.
The only thing is I don't know if the CM can perform the same because of this according to Graeme Gill "It also is not quite as good as the i1pro in accurately measuring very dark patches due to stray light being reflected by the aperture surround, and a degree of electrical interference between the LED driver and the spectrometer sensor." But if you want the Munki to give all it's got, try Argyll to the max or just download the future posted profile.
I'll definitely be posting this profile into my profiles menu for the Pro-100. I'll also be testing the illuminant effect and determine if it worthwhile to post profiles for Tungsten, Fluorescent lighting and FWA.

the reason I can't give a sample is that my scanner simply can't capture the tonal resolution of the print.

I just can't see how printing will be better. Just can't. Even with a magnifying glass, no grain either.

 

[Emulator]

2310 patches with 256 grey iterations - that will make everyone jealous.

 

[mikling]

Actually it is not that hard with the CM to do that if you followed Pharmacists routine. It's a small investment towards better prints for the cost of time and a few more sheets. And.....once you're profiling, you don't count by the sheet anymore. We know that.