X-Rite i1Profiler Review
By Andrew Rodney
This is the RGB Printer Profiling screen. To the left side are the various assets.
At the bottom are the Printer Profiling Workflow buttons that move the user from window
to window to step though the various processes that result in the creation of a profile.
Notice the separation of Patch Set and Test Chart.
Here you see the Test Chart workflow for just configuring the page layout after selecting a patch set.
It’s been a very long wait, but the next generation professional ICC profiling suite from X-Rite called i1Profiler (i1P) has been released. Before getting into a short review, it’s useful to take a look at the “recent” history of professional level color management solutions. For years, there have been two popular high-end, competing packages: GretagMacbeth ProfileMaker Pro 5 (PMP) and X-Rite Monaco PROFILER 4. I don’t recall the exact release date of each, but both date back to 2004. Being 2011, this represents eons in software development. In 2004, CS1 was the current version of Photoshop and Lightroom 1.0 was three years from release.
In 2006, X-Rite acquired GretagMacbeth and later released an impressive “entry level” profiling package called ColorMunki. X-Rite later released the ColorChecker Passport for creating DNG camera profiles. Since then, only minimal software updates for PMP and PROFILER were released. Today, X-Rite has attempted to take the best of both of those products and produce a new high-end package. Did they succeed? Was it worth the wait? To answer those questions, it’s necessary to take a quick look at these older competing packages.
Monaco PROFILER had a simple wizard based User Interface (UI) for guiding users through the various tasks of building ICC profiles. The color engine, the heart of the profile creation software, produced excellent and smooth color transitions. It’s my preferred package for the creation of CMYK output profiles. PMP used a modular software approach to divide up the added functionality and complexity provided in this package: target generation, color analysis tools, and profile editing. The PMP color engine isn’t known for producing the smoothest color gradients; yet, many preferred this package and the profiles it generated. I adopted this package for creating RGB output profiles. PMP had additional advanced functionality that was somewhat unique and useful. I could measure an illuminant for the generation of an ICC output profile. Much of the data used to build PMP profiles is stored inside the profiles making it very easy to regenerate new profiles with differing settings.
Both products had strengths, weaknesses, and a loyal following who preferred one product to the other. Clearly the challenge for X-Rite was making both groups happy by incorporating the best of both products into one new package. The old saying, “You can’t please all the people all the time“, is appropriate in terms of version 1.0 of i1P. There is legacy functionality from previous generation products that isn’t available in i1P, such as the ability to create ICC profiles for scanners and digital cameras. Some customers will disapprove, but, in X-Rite’s defense, the number of users who need these options today is presumably low. In my experience, the creation of ICC camera profiles has been often ineffective, unlike the creation of other ICC device profiles. Customers who upgrade from PROFILER or PMP to i1P will have access to their legacy software for the creation of these profiles. There are some useful profile analysis tools I use in PMP MeasureTool that are missing in i1P, but I have access to that module. One could argue that in 6+ years since PMP or PROFILER was released X-Rite could and should have incorporated every legacy function in i1P. I can’t disagree!
I will concentrate on reviewing the creation of printer profiles within i1P. While this product has a workflow area for profiling projectors and displays, I did very little testing of this functionality. I prefer to use a “smart monitor” that fully controls this process for calibration and profile creation. There is good news about i1P in how it builds printer profiles. The new color engine, based on the ColorMunki (called i1prism), produced superior quality profiles from both RGB and CMYK devices compared to PROFILER or PMP. I tested CMYK profiles generated by PROFILER and i1P for several digital presses; printing a large sample of images and synthetically created analytical targets, then visually comparing the output. In these tests, the same measured data was used for the generation of all profiles. The output using the profiles from i1P was, in every case, superior and preferable. A scum dot* issue when printing with PROFILER generated profiles has been eliminated in i1P. I tested RGB output profiles using an Epson 2880, 3880 and Canon iPF6300; all printing on Luster paper. I imported the same measured data into PMP and i1P for these tests. I preferred the prints using the i1P profiles on all three printers. The new i1Prism engine produces output profile quality, which is visually superior to both of the older X-Rite packages: smoother gradients, better gray balance, better tonality.
* For those not familiar with the term scum dot, when converting RGB values of 255/255/255 to CMYK with these profiles, the resulting value was not 0/0/0, but rather something like 0/1/1. A dot was produced where none should result from the Perceptual table in PROFILER profiles.
I prefer to build custom RGB targets for printer profiles and this is one reason I gravitated to PMP. I was told that the new patch generation engine in i1P was superior to PMP so I ran a test building targets using the same number of patches in each. I specified the generation of a 1728 patch target in PMP and i1P, measured each, and built a profile in each packages. I’m pleased to report that the i1P target produced superior profiles from both package. For example, on the Epson 3880, I could see slightly better (purer) red rendering in my Printer Test File image. While subtle, it appears this newer patch generation produces better data for profile creation. For PMP users who have built custom targets in the past, I’d recommend you rebuild new targets using the i1P target generator and, of course, build new profiles using the improved i1Prism engine.
Creating targets in i1P is overly complicated due to its design of separating the Patch Set (the number and size of each patch) and the Test Chart (the page and layout the patches flow) in two different Printer Profiling Workflow and thus UI areas. (See figure 1.) This means I had to move from Patch Set to Test Chart and back numerous times to define how I want to build a target. To this day, I can’t understand the logic of this strategy. Attempting to build a defined page filled with the maximum number of patches is an exercise in frustration. Fortunately, once you build a test chart, it’s easy to load and use.
During the months I tested i1P, I was creating custom ICC print profiles for customers. I would send these customers PMP generated profiles and then import the same measured data into i1P and have that package build a profile with roughly the same settings. I sent both profiles to customers only telling them that the profile labeled “v2” was different and asked them to comment on which profile they preferred. Every customer provided the same feedback; they preferred the i1P profiles. Consequently, the new i1Prism engine alone is providing output profiles that my customers and I visually prefer. Add superior target generation and the profiles are even better. That does make me a happy dog.
There is one more interesting new feature in i1P that deserves mention. I call it the “Holy Checker.” X-Rite prefers to call it the more politically correct “ColorChecker Proof” (wimps). The i1P package ships with a new and smaller half page sized 24-patch ColorChecker that has cutouts in the target so you can see through it (see figure 2). After creating an ICC profile, an option in i1P opens a workflow called ColorChecker Proof. In this area, a profile is used to convert an internally stored document of a ColorChecker to the output color space. I can send that data to a printer from within i1P or export the document. Once the print is made, the concept is to place the holy checker over this print and use it as a visual reference to inspect the output quality of the profile. I have no idea what rendering intent, CMM, or other options are used to make this color space conversion. This is a necessary data point that should be specified. Further, since the ColorChecker workflow requires printing on 8“x11” paper, it would have been useful if i1P output two rendering intents per page, one with a Perceptual and one with a Relative Colorimetric intent and labeled each. Due to the large number of options one can specify when building the perceptual table in i1P, it would have been useful to see these effects on a print.
At the top is a screen capture of the ColorChecker Proof workflow in i1P.
At the bottom is a ColorChecker Proof placed on top of an Epson print.
Where X-Rite dropped the ball with the ColorChecker Proof workflow is providing a method to improve a profile based on patches that may not adequately match. While I can set the software to put a hash mark over a patch that falls out of printer gamut, thus providing an indication of why that patch may not match, this process then stops in its tracks. There is no way to set this ColorChecker Proof workflow to regenerate new color patches in order to update the profile for a better match. That would make this process really useful. In fact, the new profile iteration process, first introduced in ColorMunki, could be a waste of engineering resources, but more about that later. I hope X-Rite takes this excellent concept and evolves it further to produce a better match between a print and a known color reference. It’s a good start and it’s useful for those that need to see the outcome of the profile instead of relying only on numerical statistics. The process needs to guide the user into producing better settings for profile generation and result in superior profiles.
While i1P makes improved output profiles and targets, what’s the bad news? Primarily, the user interface and mode of operation seem unpolished. Instead of using a wizard-based approach as implemented in PROFILER, or a modular approach employed in PMP, X-Rite’s software team has invented a new scheme that I found confusing and, in some cases, just flat out frustrating. Creating printer targets, as discussed, is an example of the poor user experience from questionable design decisions. Much of the UI breaks Macintosh user interface guidelines. Many settings that should remain sticky between launches don’t and I believe there are too many settings that are of questionable usefulness. Some workflow modules operate differently than others making the consistency and ease of use ineffective.
Like PMP, i1P stores a lot of user generated data that can be used to rebuild new targets, test charts, profiles and so on. These are called Assets and are stored in an area along the left side of the UI. The concept being just drag and drop an Asset onto workflow buttons for that area to accept data and update various parameters. Assets operate like presets in Lightroom, but with a haphazard approach. Often the results of dragging and dropping an Asset result in impartial update of the parameters needed or expected. In i1P, the new data Assets take on a cornucopia of file formats to describe data such as targets, measured data, saved workflows, measured illuminants; dpfx, mxf, txf, sxf, pwxf, pxf, oxf, axf, cxf and txt. These Asset files live deep inside the Application Support folder on the Macintosh, within dozens of sub-folders, making it a bit of a undertaking to sync and update multiple workstations (hint, Symbolic Links and Dropbox work well). It’s initially confusing and a little overwhelming.
I praise X-Rite for adopting this idea of storing user generated data in i1P, but there are too many file formats, and a lack of backwards compatibly may affect existing workflows. If I want to build a new target in i1P and use it in MeasureTool it’s possible but it’s a bit of trial and error to accomplish. It’s not possible to load PMP targets into i1P, it will only accept the newer txf file. Older i1Share measured illuminants used to build printer profiles in PMP can’t be used in i1P as there is no backwards compatibility with older cxf data files. I can load CGATS measured data files from PMP into i1P using drag and drop over the Measurement workflow button. I can then regenerate profiles using the new i1P engine, although X-Rite recommends using a target generated with i1P for best results. For PMP users, regenerating new profiles from older measured data is easy. Doing the same from a PROFILER saved session file is also possible, but is quite an involved process using X-Rite ColorPort to save CGATs data. It is too bad i1P couldn’t just import legacy data formats from the earlier X-Rite products. I suspect when customers see the quality of the i1P color engine, some will need to easily rebuild profiles from existing data.
One of the most interesting new features I wanted to test was the optimization of profiles first introduced in ColorMunki. After creation of a profile, it can be improved by generating a new set of patches, loading Pantone color swatches, or even extracting colors from an image. I built a profile for the Epson 3880 using 1728 custom patches with the new chart generator and the resulting profile was quite good. But, could it be improved? The ColorChecker Proof already discussed seemed to indicate a few patches could better match. I built a new target using an additional 3000 patches and then imported an image of the ColorChecker to extract its 24 colors. This process isn’t as effective as I’d like. I expected the software to recognize the unique 24 colors in the ColorChecker and extract them for this target, but that was hardly the case. I had to specify the highest value for an extraction (a mere 30 patches for some reason), and then manually remove the incorrect colors. Also somewhat inexcusable in such a product, images loaded into this workflow are not color managed so they do not preview correctly.
I printed this 3024 patch target, measured it and used this data to build an optimized profile from the original, but I saw no difference on the print. After converting a target of 918 color patches using the original and optimized profile, and making color lists in ColorThink Pro from this data, I was shocked to see the max deltaE difference between the two profiles was less than 1. In other words, the optimization didn’t provide any visible or useful differences in this test. I was told by X-Rite that the concept behind profile optimization is to improve profiles initially built from a small number of color patches. That makes sense for the audience using a ColorMunki, who will only measure 100 patches. For a high-end package like i1P, why would anyone build a profile with so few patches and then essentially start the process over again? It takes far less time and effort building a 1700 patch target in the first place. I was hoping this optimization process would improve gray balance for printing neutrals by importing such patches, much as Bill Atkinson did back in 2005 when he created the Epson 7600 and 9600 printer profiles. Bill used an optimization process of his own design to greatly improve how grays map to output color space, a novel idea at the time and one I had expected to see in i1P. I hope X-Rite can improve both the extraction of more colors from images and allow us to build better profiles using optimization in future versions.
Another irritation is i1P provides many profile building options that don’t appear to produce any benefit. When creating ICC profiles, there are no less than four different Chromatic Adaptation choices. So, which to use? Chromatic Adaptation isn’t applied to profiles built using the Standard D50 illuminant, the default, yet this option is still visible and I suspect it will just confuse users. i1P has sliders with scales that seem inappropriate for the results they produce. For example, there is a new Smoothness slider for the creation of the output profile. It scales from 0 to 100. One would expect, at the very least, a profile generated at zero, 5, 10, or 15, would produce results that can be seen. Otherwise, make a far smaller gradation of the slider scale. Point in fact; this and many of the provided sliders produce tiny, almost insignificant differences from each end of the slider scale. Building profiles with both extremes for the Smoothness slider produced results that were less than a max deltaE of 1, meaning the differences in output was invisible. While some output devices may exhibit behavior that benefit from all the possible settings, finding the right one will be very time intensive. In PMP, there was a very handy Batch function whereby I could pick a number of different profile parameters, load them into a dialog box and walk away while all were built. Then, I could test the settings and evaluate their qualities. i1P has even more profile building options than PMP or PROFILER, yet no batch processing of profiles. Another disappointment for those making CMYK profiles is the lack of custom black curve editing as can be adjusted in PROFILER. This is an advanced, but necessary, feature that users such as myself require when building CMYK profiles for certain types of presses.
Overall, the look and feel of i1P feels like a version 1.0 product, arguably a late beta release. For example, I can easily drag and drop a RGB profile or Asset into a CMYK workflow and the software will go about building a new profile from the data when clearly it should not. The resulting profile is an RGB output profile, but the settings for black generation in the CMYK workflow affect the profile. The lack of such error checking like this is seen in several areas throughout the application. Yet the underlying color engine and the qualities of the profiles it generates are such that I can’t imagine using either PMP or PROFILER to build my output profiles. After months of testing, I can just get past the most egregious UI design discrepancies, but I hope X-Rite will fix these visual aggravations and bugs in the next dot release. If you own either PMP or PROFILER, you’ll want to consider this upgrade as the profile quality is impressive. Lets just hope that we don’t have to wait too long for the other issues to be resolved; we’ve waited long enough for this new package.
For information about upgrading from ProfileMaker Pro, PROFILER and i1Match, see:
For Information On Upcoming Andrew Rodney Workshops
In Vancouver, BC, See Below...