Fun with Spyder3Print SR

A month or so ago, I purchased a Spyder3Print SR system. I’d already been using a Spyder3 Elite v4 to calibrate my monitors, and I’d been pleased with the features of that system. I was especially happy with the Elite v4’s capability to map out calibration differences across the surface of the monitor. Until I used this new version, I wasn’t aware of just how far off the upper corners of my iMac’s display really were. They are much cooler than the middle of the screen. By calibrating to the native color temperature and gamma of the display, I was able to achieve a much more uniform result across the screen than by picking 6500K and gamma 2.2. Until I saw that in action, I don’t know if I would have guessed how valuable the approach was.

However, that’s a different story…

For now, back to the S3P. I was motivated by the rather modest price of the S3P, and by its recent upgrade to support strip reading. As it turned out, the strip reading feature became of less interest to me when I finally began to use it; but I’m getting ahead of myself. The S3P system, sensor included, is easily a 10th the price of much more high-end devices. The question in my mind was how it would perform relative to any manufacturer’s standard profile for printer, paper, and ink.

I think that the short answer is pretty well. 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. The first paper/printer combination that I profiled was Hahnemühle Fine Art Baryta paper for my Epson 3880 with Ultrachrome K3 Vivid Magenta inks. I wanted to compare prints from my profile against those from Hahnemühle’s ICC profile from their web site for the 3880. I also decided to try out Epson’s built-in color management (CM) assuming the recommended media type of a luster photo paper. The HN FAB paper has been reviewed in many other places on the tubes of the Internet (TOTI). Some people like it; some don’t. As they say, a choice of paper is highly subjective. HN FAB is quite heavy weight, quite warm, and highly textured. I wanted to try it for B&W images, to achieve a certain warm tone. So, profiling it for color gamut might seem a little off the mark, and in fact some of my initial color results through me because I imagined my problems were related to the paper. This proved not to be the case.

To be specific, I was getting rather ghastly grey skin tones with the profile I’d created. Now, I’d printed targets from within the S3P s/w itself, with color management disabled as I thought the instructions had described. (Of course, I’d totally messed this up. I’ll tell how in a moment.) Frankly, people all looked like vampires or something. Initially, I had been using the 4-page “EZ colors plus grays” target set. As advertised, this target set was quite simple to get good results with the strip reading approach, or so it seemed. After thrashing around with several really poor prints on the HN FAB paper, I decided to try out Epson’s Premium Photo Paper Glossy for a compare/contrast. Now, this paper is much cooler than the FAB and has an extreme gloss surface. My prints on this lacked any of the warmth of the FAB, and human subjects appeared to have recently arisen from the grave. Simply awful results.

I went ahead and tried the built-in CM on the Epson for the Epson paper and the results were glorious. To say that I was more than a little disappointed in the results of my efforts would be putting it mildly. I had also begun to discover another problem with the S3P system: very frequently the sensor would read paper white for high quality papers as nearly black.

This issue seemed to have to do with the calibration of the sensor. In the base of the sensor is a small, white disk. When properly calibrated, this white reference disk will measure at an Lab luminance of about 90. Paper white for the various high quality papers I was using was anywhere from 95 to 98 or so. Occasionally, it seemed that the calibration function failed, and the difference in luminance of 7 to 8 or so between the reference disk and paper white was messed up. My guess is that calibration in this case produced too high a value, say 92 or more, and adding the difference for paper white of say 8 or more, caused the internal arithmetic to overflow; that is, effectively drop the digit ‘1’ in ‘101’, for example, leaving just the ‘1’.

More likely, the internal representation for luminance (or the RGB equivalent) is in some fixed point binary representation. If the calibration value is off (too high effectively) then adding in the difference for paper white will yield an overflow, and the measurement comes out looking like black. For example, 250 + 10 in binary would be 4 expressed as a positive 8-bit integer. (Or more correctly, 0xFA + 0x0A = 0x04, expressed as 8-bit positive hexadecimal numbers.)

Anyway, this is a serious pain in using the S3P sensor. After having some discussions with Datacolor’s tech support people, it seems that keeping the reference disk very clean is one part of the problem. Another part is that the base for the sensor has 4 small plastic tabs to protect the reference disk from contacting the sensor head unless pressure is applied. On the particular unit that I have, the amount of pressure needed to make good contact seems to be very high indeed. Frankly, I clipped back two of the four tabs so that this stopped being a problem for me.

So, recognizing this calibration issue, I always start a target measurement process by (1) calibrating the sensor, (2) using the measurement tool to check the luminance of the reference disk, and (3) measuring paper white. If I get an overflow, and paper white measures as black, or as sometimes happens if only one or two color channels blows out and an odd saturated tone appears (say magenta), then I repeat until the calibration is good. I have found this to be an absolutely essential first step to check the white calibration. Otherwise, one risks wasting a good deal of time and effort in measuring targets.

Now, with the so-called “EZ” targets, paper white is the second measurement patch on the first page, right after ink black. With this target set you get almost immediate feedback that something is wrong with your calibration. On the other hand, as I have discovered to my chagrin, with the “Expert Targets” paper white doesn’t show up as a test patch until page 3 or 4. By this time, one has already spent a long time accumulating useless measurements that have to be pitched. Trust me, you don’t want to blow that much of your life on work that you’ll never be able to use. Learn from my mistakes.

Big note to readers: test your white reference calibration first!

Make sure the white disk is clean. Make sure you’re applying sufficient pressure to the sensor head to get it to fit properly and squarely in the base. Measure paper white first.

Now, once I’d got past the white is black issue, I still had vampires for subjects. Again, it took me a discussion with Datacolor’s support folks to straighten this out. I should point out that I’m using Macs running OS 10.6.4. In fact, I tried this out on 3 of them, with the same bad results. It turned out to be a complete misunderstanding on my part as to how to turn off printer color management from within the driver. This was as true for an Epson and a Canon printer that I have. I assumed, incorrectly, that I could achieve this by enabling ColorSync color management in this window:

Do not do this!

Do not do this!

Instead, you must do this:

Select Epson Controls

Select Epson Controls

And this:

Disable color management in the printer here

Disable color management in the printer here

Having finally figured out how to print properly with color management disabled from within the S3P s/w, I could compare the target prints I had been making with the “right” ones. The difference was amazing to say the least. Of course, to my completely untrained eye, the original target sets looked OK. In contrast to the “right” sets, they were completely washed out. Deep blues looked like cyans, etc etc. With proper target prints (and proper white calibration), I began to get good measurements. Then, I began to get reasonable printer profiles.

Of course, I’d begun the exercise with the idea of comparing my custom profiles with those from various manufacturers or third parties. Let me say this about that. I think the custom ones I finally came up with are superior to those available online.

How can I express this? One way would be to try to get some comparison prints online, but I think the differences are subtle enough that this wouldn’t prove much. Instead, I have a different set of observations. If I open the S3P targets into Photoshop, I can soft proof them with either my own profiles or comparison profiles from the appropriate manufacturer. I can also turn on the gamut warning to see which target patches are out of gamut with that profile. For this purpose, I assumed that “absolute colorimetric” was the appropriate rendering intent for this comparison exercise.

Armed with this method, I began to compare/contrast the gamut warnings for my profiles versus those from the various manufacturers. By this time, I had profiles for Kodak papers on my Canon i9900, and for Hahnemühle Fine Art Baryta and Photo Rag Baryta on the Epson 3880, as well as Epson Exhibition Fiber and Premium Photo Glossy for the 3880 too. Now the ICC profile for the Epson Exhibition Fiber is particularly interesting in that the one I employed was done by the good folks at Pixel Genius. You can see a description here.

First, I have to point out that the Pixel Genius site doesn’t actually have a profile for the 3880; so I wound up using the one for the 2880, which uses the same Ultrachrome K3 Vivid Magenta inks. To be frank, as far as this soft proofing exercise is concerned, I didn’t see any significant difference between any of the ICC profiles available from Pixel Genius as far as the K3 VM ink printers were concerned.

Anyway, here’s a summary of my results. In every case, the profiles I made using the S3P device had a broader gamut than the standard profiles from the manufacturers and Pixel Genius as measured by the number of out-of-gamut patches shown by Photoshop’s gamut warning function in soft proofing.

It’s probably worth pointing out at this stage that I graduated from the 225-patch EZ targets to the 729-patch Expert targets. With the Expert targets, I found strip reading to be pretty useless. In spite of the tedium, I’d go through the process of reading patch by patch, carefully paying attention to any obvious errors. A patch reading error was pretty much guaranteed if I moved the sensor too soon after heading the button. The software provides audible feedback with clicks and bells from the computer’s sound system. Occasionally, it seemed that there would be a longer delay in completing the reading. This was random, and maybe driven by priorities within the OS. Who knows?

I found that patch reading errors were easiest to detect once a target page of measurements was completely full, and then turning the display over to show just the measured values. There are three options: the pure intended tones, a split view (in which the upper side of the patch is the intended value and the lower part is the measured value, and just the measured values. The targets have a certain visual pattern, and errors are fairly obvious. My guess is that I would always have about 4 or 5 errors on every page, with no particular rhyme or reason to why. With about 243 patches per page, that’s around a 2% error rate. Maybe I’ll get better. Right now, I think a lot of the issues have to do with the sensor, the software, and the priorities with the OS. My conclusion here is that it is extremely important to review each and every target page for the accuracy of the patch measurements. When in doubt, repeat the measurement.

The S3P s/w has the ability to average measurements together to smooth out measurement errors. Apple has an extremely useful profile characterization tool in the ColorSync Utility. For some of my early profiles, I’d open them up in ColorSync and compare them to the manufacturer’s profile. In this way, it became obvious where there were “glitches” in my profiles, which apparently stemmed from poor target measurements. Since the profile displays in ColorSync shows the Lab values where there are issues, it’s fairly easy to go back to the target set and identify the failed measurement patch. As well, repeating the measurements and averaging is something I’ve found to be very useful in achieving accurate results.

If all this seems like a lot of work, I cannot disagree. It really is. But here is a demonstration of the benefits. These are screen shots of the target pages within PS CS5, with soft proofing enabled for my profile of Epson Exhibition Fiber versus the profile from HN, in each case with an absolute colorimetric intent, black point compensation on, and paper white enabled.

Expert Target Page 1

Expert Target Page 1

Expert Target, S3P Profile

Expert Target, S3P Profile

Expert Target, Pixel Genius Profile

Expert Target, Pixel Genius Profile

By my count, there are 78 patches out of gamut in the S3P version and 102 patches out of gamut in the Pixel Genius version. This is a reasonably typical result. I have a lot more work and experimentation to do, but at the present time, I’m quite pleased with the results I’ve been getting. Mind you, all of this is not for the faint of heart. But for the price, the Spyder3Print SR unit seems to be giving a good bang for the buck.

Leave a Reply