Luminous Landscape Forum
Raw & Post Processing, Printing => Printing: Printers, Papers and Inks => Topic started by: johnbirch on January 04, 2013, 04:01:28 pm
-
What is everyone using to make your own printer ICC profiles? In particular, what hardware are you using. I use an Epson 4900 printer.
Thanks,
Jeff
-
i1 Pro spectro with ArgyllCMS software (it's free!).
-
ColorMunki with ArgyllCMS (the Xrite software for the ColorMunki is total cr.. for making print profiles).
cheers
afx
-
Due to my slight obsession I had been using i1Pro + i1match + Atkinson targets, i1Pro & ColorMunki + ArgyllCMS (also tested basICColor print in the meantime), upgraded to Spectrolino+Spectroscan + ArgyllCMS (Argyll generates very smart hexagonal patches for Spectroscan).
I ended up with i1Publish Pro2. I like it for convenience, quality of improved i1Pro2 spectrophotometer, nice perceptual rendering intents, ease of use, and - last but not least - a stylish carrying case.
-
I've been getting excellent results using ColorMunki Photo for printer and display profiles.
Sal
-
What is everyone using to make your own printer ICC profiles? In particular, what hardware are you using. I use an Epson 4900 printer.
Thanks,
Jeff
I use Datacolor SpyderPrint. Makes great profiles, easily, at reasonable cost.
http://spyder.datacolor.com/en/portfolio-view/spyderprint/
A review by Keith at:
http://www.northlight-images.co.uk/reviews/profiling/spyder3print_sr.html#media_check
John
-
I use Datacolor SpyderPrint. Makes great profiles, easily, at reasonable cost.
http://spyder.datacolor.com/en/portfolio-view/spyderprint/
The "problem" is sometimes it doesn't depending on a number of factors, mostly because it isn't a real Spectrophotometer like the ColorMunki which will also deal with display calibration (even odd display technology like LED or wide gamut).
-
The "problem" is sometimes it doesn't depending on a number of factors, mostly because it isn't a real Spectrophotometer like the ColorMunki which will also deal with display calibration (even odd display technology like LED or wide gamut).
That is BS, as Keith covers in his article(s)....and he is a user of xrite printer profiling products....primarily for the automated scanning.
Andrew, you and others, have made this elitist, promotionally biased statement before. I have NEVER seen it backed up by fact. As Jeff would say....show me the facts...!!!
John
BTW...do not muddy a discussion of printer profiling with display profiling. While colormunki does both, the SpyderPrint is dedicated to print. A separate, dedicated Spyder offering is for displays. Also, last I knew, xrite "crippled" any advantage the Colormunki hardware might have had by limiting the software provided...no upgrade to greater capability provided. If you want that, you need to but the full xrite kit.
-
Just went to the ArgyllCMS website to see howto use this software with my ColorMunki. Wow way over my head. Command line interface? Confused as to how this creates a profile?
-
Here is just one example, when i get to home base, i will provid others:
http://www.imaging-resource.com/ACCS/S3P/S3P.HTM
Sent from my iPhone
-
Just went to the ArgyllCMS website to see howto use this software with my ColorMunki. Wow way over my head. Command line interface? Confused as to how this creates a profile?
It's not as complicated as it looks. I had a good time learning how to cast a spell to create a profile, calibrate a display or check the OBA content in a paper ;)
Start here:
http://argyllcms.com/doc/Scenarios.html
http://www.argyllcms.com/FCMS2010_ArgyllTute.pdf
-
Here is just one example, when i get to home base, i will provid others:
http://www.imaging-resource.com/ACCS/S3P/S3P.HTM
Sent from my iPhone
This article is almost 4 years old. Please read Keith article,recently updated, which covers the EZ targets that provide self analyzing reading to reduce bad reads.
The way I read the article was that the problem was errors in reading....having nothing, other tan conjecture, to do with the hardware.
-
Couple quotes from Keith's article: http://www.northlight-images.co.uk/reviews/profiling/spyder3print_sr.html#media_check
"...I've looked at the Spyder3Print system (and its predecessors) several time in the past and always found it easy to use, with a great deal of useful support and explanatory information.
If I'd had a minor quibble it would be the 'one at a time' patch measurement approach and its susceptibility to errors if you were not careful enough.
I realise that this is a personal quirk, not helped by the fact that I've got automated measurement equipment here, that is ten times the price of the Datacolor system.
The new SR sensor makes it a lot easier to take simple consistent measurements. you can use it in the previous 'patch at a time' mode if you like, or try one of the new scanning modes, where you move it across the paper.
That said, poor measurements make poor profiles, so it's nice to see the new 'EZ' print targets where the software will do its best to alert you to mistakes..."
"...You may see arguments on forums about the relative merits of different measuring devices - I take the view that if the results are good then that's the main feature. Also, like heated camera debates, a majority of those arguing won't actually have the devices in question. I've got examples of every device reviewed on this site and am happy to answer more detailed questions about anything I've written..."
Keith has been doing hands on, detailed, and fair testing and review of monitor and print calibration devices for years. Read his reviews of x-rite devices, including ColorMunki.
John
-
Thanks to everyone for the suggestions.
Jeff
-
That is BS, as Keith covers in his article(s)....
Wow, what crawled up your....
BTW...do not muddy a discussion of printer profiling with display profiling. While colormunki does both, the SpyderPrint is dedicated to print. A separate, dedicated Spyder offering is for displays.
At what additional cost? How's that muddying the water?
On B&H the SpyderPRINT system costs $319. It doesn't have provisions to profile a display or projector. On the same site, the ColorMunki is $449. So right off the bat, you're saving money on the ColorMunki. That is mudding?
The ColorMunki is a true Spectrophotometer. A 10nm LED device that produces spectral data. The SpyderPRINT is said to be a Spectrocolometer (whatever that is, a colorimeter? A Spectrophotometer? A dessert topping and floor wax)? Does it really produce spectral data?
The SpyderPRINT package comes with profile tuning sliders, ask yourself why (lots of users apparently need to use them to get a desired profile, not necessary with ColorMunki). And does the edit affect both parts of the profile table? What's the cost in terms of time, ink and paper to tune a profile?
The SpyderPRINT seems to expect a 729 patch target, one patch at a time, the ColorMunki only requires 100 patches and it can measure in a scanning mode. Faster and as importantly, you're getting something like 100 samples per second, averaging the measured data. One patch at a time? I did that back in early 90's with a ColorTron and let me tell you, it's no fun. Yes, the ColorMunki requires two sets of targets (50+50) to be measured and perhaps someone who's really good at measuring a single patch could read 729 faster than someone could scan 100 IF you include the time you don't have to sit there waiting for the 2nd print to dry before you measure it. But if time is money, you can do something else while the 2nd target dries. Either way, you'll be spending far, far less time measuring with the Munki and getting more data averaging.
But how is the quality of the Munki? My friend George Jardine the Lightroom guy has a Munki and built profiles on two papers, asked me to build profiles using i1Profiler, an iSis XL, 1729 patches plus 2505 optimization patches and we compared the output from both. Mine was a little better but not by much. Of course you can't take MY word for it, ask George as he's got a set of prints too (http://mulita.com/blog/)
Look, if someone wants me to compare the 729 one patch a a time profile (ugh) to the iSis using the same process as I did for George, loan me a unit for a few days. I'll even have George print the same images.
This article is almost 4 years old.
What's the saying about those who neglect history? OK, more recent offerings, not that this has as high a bearing on the two products as the items presented above.
http://stonerosephotos.com/blog/2010/08/fun-with-spyder3print-sr/
However, the system is not without its pitfalls. I learned several of these the hard way. In fact, my initial efforts were absolutely abysmal; and I finally wound up raising a ticket with Datacolor, the suppliers of the S3P product.
http://www.sicip.net/articles/spyder3print-page1.htm
This is a tedious process which takes about 3 seconds per patch. A big target of 729 patches would therefore take around 40 minutes to complete. It is mind-numbing work but you have also to keep your wits about you. Initially we worked far too fast, missed patches and mis-read others. The resulting profiles were shown to be poor by examining a Granger Chart, using the soft proof facility of Photoshop (see screen grab). Suitably chastened, we slowed right down and worked very slowly and methodically. The yellow patches were the most difficult to align as there was little differentiation between them and their white, dividing strokes. The other problem to watch for is that you must make sure you do not scrape the feet of the colorimeter over the unmeasured portion of the target. The drag of the USB lead is a perpetual source of irritation when moving the instrument around. The measuring guide is one of the better ones we have tried but its roller feet were so sticky that they started by ripping the front of the print clean off!
Further (to be fair):
We tested the Spyder against our (very expensive) spectrophotometers and found it to be accurate enough for everyday professional photography use. The claim on the box of the Spyder being a tool for 'production professionals' is a little far-fetched.
http://www.amazon.com/review/R1BYG69711BCT7/ref=cm_cr_pr_viewpnt#R1BYG69711BCT7
Prior to purchasing this I had been using X-Rite's Pulse Color Elite System which was VERY EASY to use and made accurate printer profiles. Unfortunately, our unit has died out after 6 years, and so we opted to try this device from Colorvision. I cannot give good reviews for this device simply because reading the patches of the profile is tremendously difficult. One has to learn a very precise timing when reading the patches otherwise one gets it wrong. Even after getting a hang of the timing, and the patches seem to have been read correctly, one notices that some of the patches read are gray. In other words it registered some white spaces in between color patches and read it as a color. With the X-Rite model,reading patches was no problem at all and one could get the hang of the timing after a couple of tries. Also, the X-rite system was smart enough to know what was a color patch and what was the space in between the color patches. The spyder can't make a distinction. Ok, I know some people might say you need to take more time to get the hang of it. But the fact of the matter is, its supposed to be easy and you're supposed to be able to read patches in minutes. It has now been some hours of repeated trial and error and I have not been able to get one single row of patches that i am confident is accurate. With X-Rite I finished all the patches and created a profile in less than 10 minutes without a sweat. So far this device is wasting so much time. The only other option is to read it measure it patch by patch which will be so tedious.
I should have stuck with X-Rite.
Issues with blue shifting:
http://connect.dpreview.com/forums/post/34535290
I get the magenta shift in the blue skies but I do not see this shift in photoshops softproof (AR: seems the proofing table in the profile need some work!). I do see it in the print preview of the epson driver though, but I can never trust the driver preview because its never accurate. (AR, he's right, don't look there).
Of course this isn't their profile engine (or so they say) it's something about OOG colors (funny):
Sky blues are just easiest to notice because (a) they tend to take up physically large areas of images, and (b) we find it easier to visually accept a shift that's cyan.
David Miller (of datacolor)
No such issues with the i1P engine which ColorMunki uses… In fact, ask a few other color geeks who hang out here how they find the new i1P color engine.
-
I spent hundreds of hours and probably 200 A4 sheets of Ilford Smooth Pearl trying (to no avail) to make a satisfactory print profile for my Canon Pro9000 using the Spyder Print 3. I could never get Blue/cyan right nor remove a magenta cast in greys.
On my fruitless journey I corresponded with Keith Cooper at Northlight and with David Miller at Datacolor. The best David and I could come up with was that the Apple print path somehow screwed the printing of the targets (both EZ and earlier) - by applying a profile (was it 'Generic RGB', was it sRGB? was it ....?) to the target image.
I finally gave in and bought a Colormunki and produced an excellent profile and excellent screen print match immediately. 50+50 patches. No print path problems there! I've subsequently used an i1Pro with i1 Profiler with even better results on both the Canon 900 and my Epson 3880.
Given some users' apparent satisfaction, my conclusion was that my copy of the Spyder 'spectrocolorimeter' was stuffed, indicating relatively poor quality control on the Spyder Print gear. An alternative explanation is that Datacolor's software doesn't play well with MacOSX - given that most of the folk I know who swear by Spyder seem to be Windows users, that is plausible, but hardly satisfactory if true.
Cheers
Tim
PS: I've still got the thing in a box if anyone wants it!
-
The best David and I could come up with was that the Apple print path somehow screwed the printing of the targets (both EZ and earlier) - by applying a profile (was it 'Generic RGB', was it sRGB? was it ....?) to the target image.
And this was printing through their application? If so, how is this Apple's fault?
Like the blue's shifting to magenta, it's not their engine, it's OOG colors or something else. FWIW, blues shifting magenta is a somewhat common occurrence or it was years ago. Good profiling engines shouldn’t be affected. I've seen none of this with the i1P engine (or earlier the Monaco engine).
-
Wow, what crawled up your....
What got at my "..." was your total bias towards x-rite, with absolutely no acceptance of other methods...not just here, but over the years. You will note that I did not bad mouth x-Rite products...only complained about your bad mouthing Spyder. Lots of posts can be found where people express problems with any product. Extracting them to bad mouth is bad form, Andrew
Issues with blue shifting:
http://connect.dpreview.com/forums/post/34535290
I get the magenta shift in the blue skies but I do not see this shift in photoshops softproof (AR: seems the proofing table in the profile need some work!). I do see it in the print preview of the epson driver though, but I can never trust the driver preview because its never accurate. (AR, he's right, don't look there).
Of course this isn't their profile engine (or so they say) it's something about OOG colors (funny):
Sky blues are just easiest to notice because (a) they tend to take up physically large areas of images, and (b) we find it easier to visually accept a shift that's cyan.
David Miller (of datacolor)
First, your reference post is not David Miller's, nor is the first paragraph of you quote.
In your last paragraph "quote" of David's was: "All printer profiles are affected, in different ways, in terms of handling out-of-gamut colors... and there are many colors which are out-of-gamut, including many difficult shadows when you're trying to profile fine art matte papers. Sky blues are just easiest to notice because (a) they tend to take up physically large areas of images, and (b) we find it easier to visually accept a shift that's cyan." A lot different than the way you abstracted it.
Why didn't you include all of David's posts in this thread
such as....
http://connect.dpreview.com/forums/post/34542710
http://connect.dpreview.com/forums/post/34531925
http://connect.dpreview.com/forums/post/34472619
http://connect.dpreview.com/forums/post/34463950
http://connect.dpreview.com/forums/post/34463380
http://connect.dpreview.com/forums/post/34463364
http://connect.dpreview.com/forums/post/34332888
If readers want the truth, not your slanted quotes, they can see that Datacolor tries to give honest, unbiased support for the average user...
Some interesting (but not too long) quotes, that you "missed", are:
"...My guess is that you're printing through the Relative Colorimetric intent, or Perceptual, rather than the Saturation intent, which is what we recommend. I've discussed this in other DPReview threads in the past.
Use the Saturation intent instead, for both softproofing and printing, and I think you'll find that this fixes the issue, assuming that you've printed the target correctly and that the profile is built properly.
If you'd like, email a copy of your measurement file to me at dmiller@datacolor.com so that I can look at them directly. Use the File:Open Data command in Spyder3Print to open the folder that contains your measurement files (they're all small text files with .xml extensions)..."
"...Yes, for artificial gradients, this is how gamut mapping in Spyder3Print profiles works. There's some color mixing in the Perceptual intent, and even more in Relative Colorimetric, that results in some loss of saturation, and that's what you see happening in the color bars.
If you put the same color bars through the rendering intents in other, non-Spyder3Print profiles, you'll see changes and other interesting things happening there, too...:-)..."
"...I spent several days going over ElectronVolt's question (which was the start of this thread) in a support ticket on our web site.
One thing that should be set straight: there wasn't a "horrible purple cast in his blue skies at all. I was able to determine this by looking at sample images he sent me.
With his images: no matter which printer profiles he softproofed and printed with, there was always a -change- in the skies, because they were out-of-gamut light blues. He was comparing 3 different profiles: a Spyder3Print profile; the Epson standard profile; and also a profile built by Cathy's profiles (created with GretagMacbeth ProfileMaker according to the internal tags in the profile, which he sent me a copy of to examine). With the Epson profile and Cathy's profile, the color of the sky shifted noticeably towards cyan. The software that created the Epson standard profile, and Cathy's profile, both have an out-of-gamut mapping algorithm that pushes the skies, in these particular images, in that direction.
With the Spyder3Print profile, for these particular images, the shift is in a different direction; a bit more magenta, rather than cyan. In all cases, it's a question of what out-of-gamut mapping is doing, but there is always a change.
In ElectronVolt's case, we were able to change how the skies softproofed, and printed, by making a small adjustment in one of the Advanced editing sliders in Spyder3Print. That's one way to make a change, and for most people, it's going to be the easiest.
The concept of pre-shifting selected hues in the target images before printing them, which is technically more complex. You have to load the .tif versions of the target images into Photoshop; tweak them; and so there are more steps to go through, and more places where you can do something else wrong that would cause a problem..."
I will follow with another post on another item.
John
-
And this was printing through their application? If so, how is this Apple's fault?
Like the blue's shifting to magenta, it's not their engine, it's OOG colors or something else. FWIW, blues shifting magenta is a somewhat common occurrence or it was years ago. Good profiling engines shouldn’t be affected. I've seen none of this with the i1P engine (or earlier the Monaco engine).
As you know more than most, improperly printing the targets via an application that cannot turn off all modifications will cause improper profile creation.
Still doesn't support your contention of the difference in the spectrometer vs Spyder.
-
What got at my "..." was your total bias towards x-rite, with absolutely no acceptance of other methods...not just here, but over the years.
IF you want me to bitch about X-rite, I've got pages of text to follow. I let them have it here on my original review of i1P, and I've got plenty of negative things to say about their UI, some of the totally silly marketing moves they've made (search for my posts on ColorMunki Smile). How screwed up the iSis is in terms of changes in Revs to the hardware. What I will tell you is once you build the profiles, they are quite good.
"...My guess is that you're printing through the Relative Colorimetric intent, or Perceptual, rather than the Saturation intent, which is what we recommend.
That's pure BS. If you want to believe it, fine. A good profile should produce acceptable results with Saturation, or RelCol or Perceptual and this is something we have to look at to decide, image by image which to use. IF you are forced to use ONE rendering intent, someone isn't following ICC spec.
it's a question of what out-of-gamut mapping is doing
Yes, bad OOG mapping!
-
As you know more than most, improperly printing the targets via an application that cannot turn off all modifications will cause improper profile creation.
I ASKED how the target was printed. IF it was printed though their application, then what they wrote was total BS. And lots and lots of my customers and people here print targets outside a host profiler application ON THE MAC with no issues whatsoever. I've probably built at least a few hundred+ profiles from people all over the world this way.
-
This article is almost 4 years old. Please read Keith article,recently updated, which covers the EZ targets that provide self analyzing reading to reduce bad reads.
The self analyzing feature has some serious limitations. A dual measurement method and checking for consistency using a delta E spreadsheet in Excel is wise. Please read on...Here goes:
I almost fear to wade into this spyderprint 3 discussion, because both good and bad experiences with the Datacolor abound for very real reasons. There's some problems with Colormunki as well, so if anyone is truly serious about routinely building custom profiles, I'd suggest saving some more pennies and stepping up to better hardware/software than either Colormunki for Datacolor Spyderprint 3.
Let me get Colormunki out of the way first, and I base my conclusion solely on a basic understanding that Colormunki requires iterative targets (i.e, print one target, measure, then print a second target with values that are based on the acquired readings of the first target, then measure again). I have not used a Colomunki personally, so maybe there's away around this iterative issue, but here's the problem with it. IMHO, the ColorMunki is perfectly fine for pigmented ink prints that have very little short and long term drift, but ALL dye-based inkjet prints need days (and at moderate to high humidity levels) in order to settle down into sufficiently stable colors for an accurate system profile. This whole short term drift issue with dyes isn't just a Colormunki problem. It is also one of the big challenges I face when attempting light fade experiments on dye-based print samples. One has to let the samples "cure" thoroughly prior to testing in order to get something resembling a stable T=0 colorimetric baseline for the sample. You can force a dye-based print faster into reasonable color stability by incubating the print in a high humidity chamber (pretty easy to make), but an overnight incubation is still going to pose practical problems for an iterative profiling process. That said, "good is good enough" color accuracy (within average delta E of 4 or 5, worst fliers in vivid colors at 6-15) may be fine for some, but I think one could argue folks who are happy with this level of dye-based print performance may also be very happy with generic profiles and have no need for custom profiling.
The spyderprint3 profiling process relies on one print session to make the targets, not two, so it deals with that drift issue in the same way as other higher end profiling packages like Eye-one, iSis, etc. You let the target dry for those several days to improve accuracy, then make the profile. The Datacolor spectrocolorimeter or whatever you want to call it is indeed reasonably accurate for its price point. I found it to agree with my venerable Gretag Spectrolinos on average within delta E = 2, better than 1 on neutrals except for high OBA content media whitepoint (which are an agreement problem with many high end spectros as well), and off in some of the vivid green values as much as delta E= 6. But that kind of accuracy is fair enough to not be the source of any serious profiling error. Humans don't pick out vivid color delta E= 6 color errors very easily when those vivid colors are residing within complex color fields (i.e., most full color photographic prints). The real problem/challenge with Spyderprint 3 is, IMHO, almost entirely OPERATOR ERROR when trying to achieve the measuring accuracy that the unit is truly capable of. It's not trivial to measure a few hundred patches by hand without making one or two mistakes, especially in strip reader mode, but those couple of bad measurements can really screw up profile quality. The SpyderPrint strip reading mode uses a primitive timing method to take the patch readings. The unit senses the pulse from the white separation lines between patches, but then counts to 0.5 seconds (as I recall) and then reads the patch. If you are way off on your timing (look at the "correct traveling speed" in the datacolor video) you will read the wrong patch and the unit's self analysing feature will catch it. This feature seems to be based on an expectation of what the nominal color should be, red, say for example, and if it doesn't read something crudely within a red hue for a colorimetric value it then flags the patch. However, if you are close to the right speed but a little too slow, the instrument's aperture can accidentally integrate a little of the prior patch and or white separator line into the reading. Likewise, if you go too fast, the unit's aperture may read with the aperture straddling the adjacent following patch and/or separator line. Depending on how good or bad the operator's chosen travel speed is, the unit's read error will go up, but not so much as to flag the self analyzing software check all the time. If you think about it that, this self-check feature has to have wide error limits because if one knew exactly what color to expect out of the printer, one wouldn't even need to profile the printer!!!
So, I solved the data color OPERATOR ERROR problem and got consistently repeatable results with SpyderPrint3 by some operator practice and more importantly by measuring the target twice, using the Datacolor softwares import/export text file feature to output a text file readable in Excel, and then using a spreadsheet (email me at aardenburg imaging if you want a copy of it) to make a delta E comparison of the two sets of measurements. The spreadsheet flags the unwanted errors, I remeasure those patches, added the corrected values in the spreadsheet, let the spreadsheet output a corrected and averaged text file that the datacolor software will then import to build a very good profile. On this point of dual measurements and before too many folks cry foul, especially those who think the patch measuring is wickedly time consuming, I can only say many years of using color instrumentation have taught me that good data takes time to collect and VERIFY. Even with an EyeOne or Spectrolino, I use this completely independent dual measurement technique. And in routine light fade testing at AaI&A, we make several thousand measurements each month, and we do catch the occasional error!... an error that would sometimes screw up the profiling tone curves. For these reasons, I think being somewhat anal retentive about one's target measurement is a good idea no matter what instrument you choose.
The last piece of advice I've got for SpyderPrint 3 owners or potential owners, is that compared to a true spectrophotometer, the SpyderPrint 3 unit takes single patch readings very fast, not nearly the 1.5-2 seconds I need for an EyeOne or Spectrolino measurement time. The Spyder3 is so fast in single patch mode, expecially if you move it with one hand and use the computer keyboard to "click" the measurement with your other hand, such that I finally came to the personal conclusion that the strip mode feature was a "nice in concept" but largely irrelevant. I could move faster and with more accuracy in single patch reading mode with the SP3. Once you accommodate the instrument's measuring idiosynchracies, the data color profiling software is really quite nice and reasonably full featured.
Oh yeah, and one more thing. The Datacolor profiling engine does still suffer from the classic LAB color model "blue turns purple" problem which other profiling packages have learned to overcome by fudge factors in the perceptual rendering mapping. Nonetheless, you see this blue-purple bias in the softproof, and it's not so hard to correct if you are working in a monitor calibrated viewing environment, so again, for the price paid, I was very personally satisfied with the Datacolor Kit. It does accomplish what none of us reading this forum should lose sight of, ie., it gives the end-user a much more calibrated printing environment than one is going to achieve relying solely on canned profiles from media manufacturers. Is it as good as a high end profiling hardware/software package? Well, no, so as a stated at the beginning, if you are a die-hard printmaker spending lots of money on printers, inks, and media (like me) you will want to forego both Colormunki and Datacolor and move to higher end profiling kit.
cheers,
Mark
http://www.aardenburg-imaging.com
-
IMHO, the ColorMunki is perfectly fine for pigmented ink prints that have very little short and long term drift, but ALL dye-based inkjet prints need days (and at moderate to high humidity levels) in order to settle down into sufficiently stable colors for an accurate system profile. This whole short term drift issue with dyes isn't just a Colormunki problem.
Some justifiable criticism of the Munki is that indeed, there's some time between sheet 1 and sheet 2. The software does pop up a "wait this long" progress dialog for dry down but as you point out, that isn't enough time for dye based inks. Personally even with pigmented inks, I prefer to let the print dry down for 24 hours. On the other hand, you only have to measure 100 patches in total and the results using this small a number of colors to measure is very impressive. Anyone been playing with these products long enough to recall a produce many years ago that only required the measurement of (if memory serves) 16 patches? Didn't work, made awful profiles. When X-rite claimed they had a process to make a 'good' profile with only 100 patches, I was super skeptical. But as I said, the end results get very close to the quality I saw with nearly 4000 patches (2500 being a 2nd iterative target).
So, I solved the data color OPERATOR ERROR problem and got consistently repeatable results with SpyderPrint3 by some operator practice and more importantly by measuring the target twice, using the Datacolor softwares import/export text file feature to output a text file readable in Excel, and then using a spreadsheet (email me at aardenburg imaging if you want a copy of it) to make a delta E comparison of the two sets of measurements.
It took X-rite awhile to get this to work out too. I recall the very early i1 wasn't easy to scan without getting an error to rescan the row again. But at least there was decent error checking such you were told, do it over again. Over the years, either the product has gotten better or I have in terms of scanning. It does take practice but I've patted myself on the back a few times when I've measured hundreds of patches for a profile using the i1 without ever getting an error. I find the Munki is even a tad easier to scan due to those huge patches!
Question: Do you know if in scan mode the DataColor product averages more than one reading per patch?
The Datacolor profiling engine does still suffer from the classic LAB color model "blue turns purple" problem which other profiling packages have learned to overcome by fudge factors in the perceptual rendering mapping.
So what's this about having to use a Saturation intent with the resulting profiles? And what's your experience with the newer product in terms of the editing sliders? Would this fix the color shift? And does it equally affect the preview table of the profile?
-
And this was printing through their application? If so, how is this Apple's fault?
Like the blue's shifting to magenta, it's not their engine, it's OOG colors or something else. FWIW, blues shifting magenta is a somewhat common occurrence or it was years ago. Good profiling engines shouldn’t be affected. I've seen none of this with the i1P engine (or earlier the Monaco engine).
Precisely what I eventually concluded. Even using ACPU to print targets still did not produce decent profiles using the Spyder Print 3 device and Datacolor's profile generator. (Unlike the Munki and subsequently 1Profiler) That's what led me to the conclusion that either (my copy of) the Spyder hardware was shonky, or that their software didn't play with OS X.
Tim
-
Question: Do you know if in scan mode the DataColor product averages more than one reading per patch?
So what's this about having to use a Saturation intent with the resulting profiles? And what's your experience with the newer product in terms of the editing sliders? Would this fix the color shift? And does it equally affect the preview table of the profile?
I can't unravel the muddy waters of why Datacolor recommends the saturation intent rather than perceptual intent for photographs. Nor did I attempt to fiddle with the sliders all that much because I was pleased with the default results once I got good measurment data with which to build the profile. Presumably, the DataColor software guys elected to bake some secret sauce into the saturation rendering intent, and let perceptual act just like relative rendering. It's a little unorthodox, but then other vendors sometimes skip a unique implementation for saturation mode, and thus all I can say is choose to use softproofing and select the rendering intent which gives best starting point for that image.
By way of a little more explanation, the Datacolor print patches are separated both by white and black separator lines depending on whether white or black gives better contrast to the oncoming color patch. The unit has a very simple photocell detector independent of the colorimeter that looks for this contrast peak as the unit travels along. It then tells the colorimeter to take a single reading 0.5 seconds after the photocell recognizes a white or black line. Hence, the operator does have to learn how fast or how slow to go in order to have that 0.5 second activated single readout land fully within the patch color. The patches are pretty big which give you some allowance for travel speed, but not a whole lot. A new user won't have any idea where to start with an appropriate scanning speed without looking at the video.
I looked at the SpyderPrintStudio kit in late 2010 right after the new strip reading version was released. Funny in hindsight now, but at the time of release the guys who did the operator manual video compression messed up, and the demo was running twice as fast on a typical computer as it should have. So initially, I tried to match my personal operator scanning speed to the video and was getting flagged with all sorts of errors. Then, I took another approach, zeroing in on optimal scanning speed using my spreadsheet results. I then emailed Datacolor tech support and we then had a good conversation about how the strip mode actually works. Datacolor confirmed my finding that in strip mode the unit expects a 1 second per patch travel time (0.5 seconds thus triggering perfectly in the middle of the patch if your speed is also perfect). The actual colorimeter readout and reset time is thus much faster, and that's why spot mode works great, too, especially for operators who can't master the right travel speed. Hence, new operator error can be high with the SP3 gismo, and you can see why there are many forum complaints about SpyderPrint 3 making poor profiles. Garbage data in, garbage data out.
One can certainly make a good case that it's way too much to ask of photographers and printmakers to go through the learning curve I went through to make the SP3 work according to it's design, but I found the ways to make it work well, and there's something good to be said for that. I can list countless other products in the digital printing workflow that have stumped me big time and for many more hours of frustration, including but not limited to computer OS upgrades, printer driver glitches, mystery meat software menus, hardware failures, etc, etc. ;D Many product reviews conveniently ignore all of the balls in the air that must be juggled just to get any of this stuff to "work as advertised".
cheers,
Mark
-
I can't unravel the muddy waters of why Datacolor recommends the saturation intent rather than perceptual intent for photographs. Nor did I attempt to fiddle with the sliders all that much because I was pleased with the default results once I got good measurment data with which to build the profile. Presumably, the DataColor software guys elected to bake some secret sauce into the saturation rendering intent, and let perceptual act just like relative rendering. It's a little unorthodox, but then other vendors sometimes skip a unique implementation for saturation mode, and thus all I can say is choose to use softproofing and select the rendering intent which gives best starting point for that image.
Keith quotes Datacolor as saying:
""Saturation Intent: This intent favours colour saturation over hue or lightness when determining how to print out of gamut colour; those colours that the printer can’t reach, which must be replaced somehow with achievable colours. This intent offers the most satisfying inkjet colour in terms of keeping high colour areas in your images looking like you remember them. For smaller gamut devices, it may be more important to avoid Hue Bias, by starting with Relative Colorimetric.
Relative Colorimetric Intent: This intent makes the most accurate, literal, choices for replacing out of gamut colours. It is often used for prepress and graphic design work, for matching spot colours, and for other non-photographic work. Brilliant colours may be significantly less saturated than with Saturation intent on inkjet printers, but may better retain their hue with other types of devices.
Perceptual Intent: Datacolor’s Perceptual intent falls somewhere between Relative Colorimetric and Saturation intents, and is best used when an intermediate result is needed. Use in cases where Hue Shift with Saturation intent is objectionable, but the low colour brilliance with Relative Colorimetric is problematic.
Absolute Colorimetric Intent: This is a special intent for proofing purposes, where the off white paper of the final printing device needs to be emulated on the whiter paper of the proofing device. Unless you are emulating newsprint output on your desktop printer, you probably will never need this intent."
It helps to remember why you are doing all this - to produce better prints of photographs. Don't forget to go out and take some as well. ;-)"
As far as I know, perceptual stays perceptual, same with relative. They do offer an option to create a profile where perceptual is replaced by saturation, relative stays relative. This is for people who like the Spyder saturation intent and want to use the profile with Lightroom, where Adobe, in their wisdom, only support perceptual and relative.
John
-
Why make them at all since the paper firms have experts who all have more experience than we do at paper profiles. I'd rather spend my time optimizing my images than creating more work for myself. Can anyone honestly say they get better results on an Epson printer than just using icc profiles provided by manufacturers?
-
Why make them at all since the paper firms have experts who all have more experience than we do at paper profiles.
If only that were true and accurate. First off, there are paper firms that make bad profiles or don't supply them at all. In order for them to provide profiles, they have to either write their own software to do so (forget about that), OR pay a pretty steep licensing fee to the profile creation company (e.g. X-rite).
Can anyone honestly say they get better results on an Epson printer than just using icc profiles provided by manufacturers?
Yes, lots and lots of us can! I have built hundreds of profiles for myself and others who will report better results for a number of reasons (you want to know why?). There are lots of people on the LuLa forums who either build them themselves or have had others build custom profile for them.
-
If only that were true and accurate. First off, there are paper firms that make bad profiles or don't supply them at all. In order for them to provide profiles, they have to either write their own software to do so (forget about that), OR pay a pretty steep licensing fee to the profile creation company (e.g. X-rite).
Yes, lots and lots of us can! I have built hundreds of profiles for myself and others who will report better results for a number of reasons (you want to know why?). There are lots of people on the LuLa forums who either build them themselves or have had others build custom profile for them.
+1
-
If only that were true and accurate. First off, there are paper firms that make bad profiles or don't supply them at all. In order for them to provide profiles, they have to either write their own software to do so (forget about that), OR pay a pretty steep licensing fee to the profile creation company (e.g. X-rite).
Yes, lots and lots of us can! I have built hundreds of profiles for myself and others who will report better results for a number of reasons (you want to know why?). There are lots of people on the LuLa forums who either build them themselves or have had others build custom profile for them.
Andrew....I agree with you.....but, for education purposes, I would like to hear your "you want to know why."
I am not pulling your chain....I would really like to know it, so I could share it with others, particularly my camera clubs.
John
-
I have use ColorMunki. Made some profiles, and still have sample prints where some out of gamut color were messed up. Then I bought SpyderStudio 3 SR. It has both display and print calibration. I am satisfied with the display calibration, but print profiling is hit or miss, definitely better than ColorMunki, but stumbles in difficult areas.
I also tried i1 Photo 2 (or whatever it is called but it is the current version), it did much better job on certain papers, but stumbled on Glossy Canvas. Given that this is the Gold Standard and even manufacturer profiles are also made by this product, I am attributing the problems with user errors. Which makes me think that user error may be the issue with Spyder 3 as well. So I will re-profile some of the difficult papers later on using ez targets.
Colormunki was so straightforward and so limited in options, that I can not think that there was any user error involved.
-
Just as reminder: "The cheapest, most repeatable automated spectrophotometer I know of—is the HP Z3100 printer! And it includes the software to make profiles—and, oh yeah, it can make pretty good prints, too."
http://www.luminous-landscape.com/essays/Z3100-profiling.shtml
-
I am not pulling your chain....I would really like to know it, so I could share it with others, particularly my camera clubs.
Just the differences between the quality of the Perceptual rendering intent for say i1Profiler versus the older offering makes a visible improvement when ink hits paper. IOW, just the software used to build the profiles can play role. When I build profiles with the new software, I also always run the optional 2505 patch gray optimization which further improves the results but it is subtle and I suspect there's a world of people who wouldn't notice the difference unless pointed out.
You might be using a differing media setting or ink control setting than the manufacturer's setting for their profiles (if they tell even tell you). The paper, ink or even hardware may be different behaving on your unit then the unit used to print the targets for the canned profile.
It is much like the on-going posts here about DNG profiles. Are they better than the one's supplied by Adobe? The answer is almost always yes but it isn't necessarily night and day differences. Adobe's profiles for their camera sample is probably superb. But how is it on yours?
-
Why make them at all since the paper firms have experts who all have more experience than we do at paper profiles. I'd rather spend my time optimizing my images than creating more work for myself. Can anyone honestly say they get better results on an Epson printer than just using icc profiles provided by manufacturers?
The Epson's are much more linear today than when I started using them over 12 years ago, yet across the board, I have not found a manufacturer's profile to be better than those I make myself using iOne Pro and iOne Pro 2. Over the years, there have been a few, but just a few, that are very close.I am now making profiles for my favorite media for a new Epson 7900 and comparing them to the ones from each of the paper manufacturers.
I find that those I create, versus the manufacturer's, provide not only for better color rendition, but better highlight and shadow control and neutrality, density and smoother tonal gradations.
There is often a lot better quality that can be obtained from good digital files than many realize. One way to learn is to look at prints made by master printers, whether analog or digital, and strive for that level of quality.
So, yes, I can say that for me, making my own profiles is superior to nearly all I've seen from manufacturers.
Ed
-
Just the differences between the quality of the Perceptual rendering intent for say i1Profiler versus the older offering makes a visible improvement when ink hits paper. IOW, just the software used to build the profiles can play role. When I build profiles with the new software, I also always run the optional 2505 patch gray optimization which further improves the results but it is subtle and I suspect there's a world of people who wouldn't notice the difference unless pointed out.
You might be using a differing media setting or ink control setting than the manufacturer's setting for their profiles (if they tell even tell you). The paper, ink or even hardware may be different behaving on your unit then the unit used to print the targets for the canned profile.
It is much like the on-going posts here about DNG profiles. Are they better than the one's supplied by Adobe? The answer is almost always yes but it isn't necessarily night and day differences. Adobe's profiles for their camera sample is probably superb. But how is it on yours?
Thanks...
"...always run the optional 2505 patch gray optimization..."
....but you're cheating....with an i1-iSis automated reader :-)
John
-
Why make them at all since the paper firms have experts who all have more experience than we do at paper profiles.
Even if a paper company has a profiling 'expert' on staff, how much time do you think they can put into creating each individual profile, if they have to profile several dozen printers for each paper they sell? Do you think they even produce individual profiels for each Epson/Canon model, or just a profile for each ink-set? How much effort do you think they put into choosing the optimal media type and other printer settings, and how large of a patch set do you think they use? How likely is it that they do the second-pass gray-scale optimization that Andrew mentioned?
-
Thanks...
"...always run the optional 2505 patch gray optimization..."
....but you're cheating....with an i1-iSis automated reader :-)
John
I do it by hand, but only for OBA-free papers, because if I don't use "legacy charts" I can't scan the resulting charts without constant errors. For whatever reason, the test-chart generator for Optimization doesn't lay out the patches to reduce scanning errors like it does for the regular profiling charts.
-
Let me get Colormunki out of the way first, and I base my conclusion solely on a basic understanding that Colormunki requires iterative targets (i.e, print one target, measure, then print a second target with values that are based on the acquired readings of the first target, then measure again). I have not used a Colomunki personally, so maybe there's away around this iterative issue, but here's the problem with it. IMHO, the ColorMunki is perfectly fine for pigmented ink prints that have very little short and long term drift, but ALL dye-based inkjet prints need days (and at moderate to high humidity levels) in order to settle down into sufficiently stable colors for an accurate system profile.
I was working with a friend's Colormunki, and we successfully managed to build excellent profiles while letting the target dry down for 24 hours. One can shut down one's computer, and restart the target scanning and profile building later, skipping the completed earlier steps. It is not exactly straightforward, but it can be done. I don't have access to the munki at the moment, and I can't remember the exact steps. But it can be done.
-
Andrew, you and others, have made this elitist, promotionally biased statement before. I have NEVER seen it backed up by fact. As Jeff would say....show me the facts...!!!
Fact is, a few years back I spend many hours trying to profile my Canon toy printer with the previous Datacolor Prdcut (Spyder 2 Studio I think it was called). Never received anything even remotely satisfactory. DataColor Support could not produce anything useful either.
Had profiling papers printed for X-Rite, sent them off and got working profiles in return.
Case closed.
cheers
afx
-
Regarding the colormunki comment regarding the need to let dye ink made test sheets dry for 24 hours or more, this is absolutely no problem with the colormunki software. You can wait a week or more, and its no problem.
Just shut down the program after the first print, and after the second print. Wait the amount of time you want to let the second print dry, and then start up the print profiler of Colormunki. You get to the first screen, then click on the box that says "Target is already printed" and click next to go to the second target, click Target is already printed, then scan it. No problem. 10 minutes is suggested, but not mandatory in any way.
I just got my colormunki today, and its already improved my monitor. My spyder2express colorimeter just wasn't working right anymore - maybe filters faded or something. The instructions and video are very straightforward and i've been able to make my first printer profile ever - amazing to me.