ColorMunki pretty much doesn't work at all

My original intent was to make profiles for my paper and ink combinations and write a nice little blog entry that discussed some of the finer points of the product that people have been asking about but have not been covered in any of the professional reviews published to date. I figured I'm smack in the middle of the target audience and I could impart some good "first time user" impressions.

Photography is my hobby, not my profession. Like many people, I don't make as many prints as I'd like simply because consistently getting output that matches my monitor is virtually impossible without good profiles, and like most people whose profession isn't photography, I can't justify spending $1500 on an i1 Photo. I know many people in the same situation, so the timing on the $500 ColorMunki solution is just about perfect.

I was also pretty confident that even if there were a few gotchas with the ColorMunki, I could probably work my way through them. I've written over 40 printer drivers for Mac OS X for both inkjet and laser printers and I also work with a couple of engineers who know color science inside and out. Given the glowing pre-release reviews of the ColorMunki and the resources at my disposal, success should have been guaranteed.

I assumed the ColorMunki package would do the following, given the audience it's targeted at and the verbiage on the website:

• Give clear instructions on how to correctly print targets for scanning. If you can't print the target correctly, you aren't going to make good profiles.
• Make good profiles. Maybe not as good as a $1500 i1, but profiles that are far better than what you get from the printer vendor.
• Provide reasonably detailed instructions for printing images successfully from popular Mac applications. To me, this list would include Photoshop, Aperture, Lightroom, iPhoto, and Preview. If you can figure out how to print with those apps, you should be able to figure out most any other application.

Unfortunately, as much as I was looking forward to writing about how the ColorMunki was solving my printing woes, I can't do that. Instead, I can only document my train wreck of an experience with this product which pretty much fails at every problem it should be solving.

I'll start by recounting my experience with the product so far... For reference, I use a Canon 350D camera with a Canon EF 24-105mm f4 L IS lens. I keep all of my photos in Aperture 2.1 on a 24" iMac running Mac OS X 10.5.2. My printer is a Canon S9000, which is a 6 color 13x19 printer.

Day 1

I started with my Canon S9000 and some Epson Premium Semigloss photo paper. The software instructs you somewhat generically to turn off color management in the print dialog when printing the target sheets, but gives no concrete details on how exactly to do that. I setup the Canon print panels for a Top Quality photo and Color Correction to None and set the Apple standard Color Matching pane to Vendor Matching. It seems like the right thing to do since the software doesn't give any specific guidance.

The first thing I noticed was that many of the target patches appeared to be black or near black. I thought it was odd that it would want that many black patches, but the software didn't complain when I scanned them, so I figured it must be okay. However, I found that my first print was much darker than I expected, had severely crushed shadows and blacks, and skin tones suffered from a yellow-green tint. I made a couple more prints and attempted to fiddle with the ink controls in the Canon print panes to compensate but made no progress at all. The plain print I had made with all settings yielded a print better than any of the prints with the new profile.

I then decided to switch to a slightly different paper I had sitting around, Epson Semigloss Colorlife. Immediately I noticed that many of the patches that printed so dark on the previous paper were actually varying shades of almost black and dark brown. The print from this profile had a bit more shadow detail but was still too dark overall and suffered from the same yellow-green tint in skin tones.

Thinking I had made some progress I then ran the optimization process where you load the image you are printing and it makes a new target sheet with patches designed to improve the specific colors in that image. This yielded a profile where many blues suffered some sort of horrible color change to dark gray.

At this point I've wasted half a pack of decent paper, a fair amount of ink and made absolutely no progress at all. Time to admit defeat and go to bed.

Day 2

I haul the ColorMunki into the office to see if the problem is something specific to my home system. For this test I used an 8 core Mac Pro, Canon Pixma ip4200, a recent 6 color HP photo printer (exact model escapes me), and a Konica-Minolta color laser printer. Surely I will find that I can get one of these printers to kick out a nice photo.

I'll just cut to the chase and say that I spent about 4 hours making profiles and test prints and all were positively horrible. I was joined halfway through by another engineer and one of our color science guys and even with our combined talents, we could not make anything close to a decent print. Additional optimization targets did not help, and in some cases were horribly worse.

I also experienced an extremely high number of scanning errors at the office that I did not have at home. When you scan each row of targets the software either moves the highlight on the screen to the next row or shows a red ring around the row to alert you that there was a problem and you need to re-scan that row. On the surface this seems like a fine UI but it quickly becomes frustrating when things go wrong.

The scanning process goes like this: Scan the row. Get a red ring. Scan the row. Get a red ring. Scan the row. Get a red ring. Scan slightly slower. Get a red ring. Scan even slower. Get a red ring. Watch the instructional video. Scan slow and be sure to move in a straight line. Get a red ring. Start scanning farther back on the paper. Get a red ring. Start scanning farther back on the paper and continue to the far edge of the paper. Get a red ring. Scan the row just like you did the first time. Magically works.  

Repeat this process for the next 4 rows, print the second target, wait 10 minutes for it to dry, and begin the red ring dance again. Fun!

Day 3

I'm pretty fed up with the ColorMunki by this point and contemplating returning it to my dealer and buying a new lens or something that actually works. I've burned through a huge pile of paper and more ink than I care to think about. However, I've heard that X-Rite has a software update in the works and it should be out this week. I'm going to hang on to the device for a few more days and see if the product actually becomes usable.

What I think is wrong with the ColorMunki...

Making Profiles

As I mentioned above, I had expectations that the ColorMunki package would get you to print a target correctly, make a good profile from the target, and then provide you with enough knowledge to be able to correctly print from a number of applications.

Clearly, the product doesn't give you much help getting the targets printed correctly, and it certainly doesn't generate good profiles based on my usage (nor that of a friend who reviews printers professionally who has had equally bad results with the ColorMunki). However, in studying the piles of targets and test prints i now have, it's not clear if the bad profiles are a result of bad profiling algorithms or if the problem is that the targets are not being printed correctly and that leads to a bad profile being generated. Laying out all of the targets from the various printers and papers I see some pretty large variances in ink density. On some the patches are light and others the darker patches are crushed to near black. However, when scanning the targets the software never indicates that the page was printed with too little or too much ink.

My first thought is that the target pages need to include something like a step wedge in the margin going from black to white with 10-20 steps. Before attempting to scan the first row of targets, the user should be able to look at the steps and be able to tell each step from the next. If you can't tell the last few black steps apart, you need to re-print with less ink. If you can't tell the first few apart you need to print with more. Basically you need to linearize the printer before you get a useful target. I would think this should be an easy addition to the product. If the product is supposed to be handling this already, then I can only conclude the profile generation code is just plain broke or there is something wrong with the device itself.

There also needs to be more specific guidance on how to configure the settings of the most popular drivers to ensure that color matching is disabled. While this sounds difficult, I think the reality is that most people are using one of three brands (Epson, Canon, and HP) and the drivers for each brand don't vary much in terms of settings. For example, most of the consumer Canon drivers have the same set of controls; you can document it once and cover most of their printers. In particular, they need to cover:

• What, if anything, needs to be done with the default profile for the print queue in the ColorSync utility. Should this be set to no profile, left to the factory profile or set to something else?
• What settings to use on the Apple Color Matching panel. Vendor matching or ColorSync? If ColorSync, what profile should be chosen?
• What settings to use on the vendor specific print panels for the top three vendors. Should I make any changes to the ink sliders?

I think I chose the correct settings but I'm not 100% sure. I think users who are less technical (which is the target audience) would have a good chance of making a variety of mistakes.

Scanning Profiles

This part of the process was positively infuriating. I watched the provided instructional videos showing an animated hand smoothly and surely moving the ColorMunki over the targets. I found myself wishing at the end of the video for the hand to then angrily hurl the ColorMunki across the room in frustration, because I surely wanted to.

The root of the problem is that when there is a failure, the software just says "error" without giving the user any feedback on what exactly was wrong. Did I scan too fast? Too slow? Did I not scan the paper before or after the row? Did i tilt the device and let stray light in? Was there a communications error? Which patches does it think were wrong? The user is left to try some random variation on scanning techniques and diagnose the problems on their own. The software should give a lot more feedback in this situation.

Somewhat ironically, there were a number of times where I know I botched scanning a row (caused by the cable snagging on something or the rubber feet on the device itself catching on the edge of the paper and causing it to drift into the next row) but the software didn't detect it. In this case, there is no provision to rescan the row it just accepted. Your choices are to start over from the beginning or keep going. This seems especially important since the second sheet of targets is determined by calculations from the first (according to the documentation). The end result will be 20-30 minutes of wasted time and a useless profile at the end. Again, this seems like something that could be easily fixed in the software by letting the user choose to go back and re-scan the last row (or any row for that matter).

Printing Pictures

Same problem as printing targets. It wouldn't be that hard to add specific examples of how to print from a variety of popular applications on the Mac.

Built In Help

For whatever reason, the bulk of the help for the product is on the ColorMunki site and not in a standard Mac help book. So, if you don't have an internet connection at the moment you need help, you aren't getting any help. In addition, this means that when you click the Help button, Safari will launch with a browser window for the ColorMunki site. If you are using Spaces this likely means all of your photo work whizzes offscreen as you switch to the space that has Safari, so the help is not in the space you need it in. While I realize some smart person thought this was a good idea because they can keep the help on the website up to date, the product has a built in software updater. There is no reason users cannot have up to date help locally in a format that plays well with the rest of the system.

The USB Cable

The provided USB cable is pretty thick and has a relatively inflexible jacket on it. So, after being tied up in a figure eight in the box for shipment, you unravel it and wind up with a long zig-zaggy cable with semi-permanent bends in the jacket that snags on everything on your desk as you move the device back and forth across the targets. Given that the usage of the device involves moving it around on your desk repeatedly you would think a cable with a jacket that resists forming bends that don't want to un-bend would be mandatory. During my usage it hooked my keyboard, mouse, mouse cable, mouse pad, desk corner and the target paper itself. X-Rite should look at replacing this with something thinner and more flexible like the cables Apple uses for the iPod and the new aluminum keyboards. However, this is the least of their problems. Had I not had so many scan failures and needed to scan each target 20 times I would not have noticed.

A Couple of Good Things

To be fair, I did find two two things that the ColorMunki actually did correctly. It did a good job of profiling two Dell LCDs, my 24" iMac and an older 20" Apple Cinema Display. In each case the display looked more neutral and had more linear response on the sample color ramps displayed in the ColorMunki Photo application. In addition, the Photo ColorPicker application which lets you sample colors on objects appeared to work quite well too.

Some notes about the install

The ColorMunki install deposits a fair amount of stuff on your drive:

There are a couple of items of note here.

• There is a colormunkid process that is launched on-demand whenever an application connects to a socket named colormunkidsocket. No idea what that does...
• There is a ninjad process that is launched on-demand whenever an application connects to the ninjadsocket. No idea what that does either, but on my iMac it's running all the time.
• The ColorMunki Photo Tray application, which is buried in their application support folder, is the process that gives you the little ColorMunki status menu in the menubar. Normally, processes that maintain a status menu in the menu bar have LSUIElement=1 set in their Info.plist to keep the process from appearing in the Dock. Otherwise, the user sees an unexpected running application icon in the Dock, but when they click on it no windows appear. It's confusing and it clutters up the Dock. It took me a while to figure out why I had this app running and what it's function was. Turns out it runs the status menu... X-Rite should clean this up.

Wrapping it up

To summarize, the product was a complete failure for me on multiple machines and multiple printers and I cannot recommend it to anyone. My friend has problems as well, so there is definitely something wrong with the product as a whole. The only thing keeping me from returning it at this point is the promise of better software this week. X-Rite is known for having quality products, so it's baffling as to why they shipped the ColorMunki in this state.

I'll be sure to post another entry when the new software arrives.