Thank you! We all owe you a debt of gratitude for your
contribution. What follows are some comments and suggestions in no
way intended to diminish the value of what you have done already:
No problem! I can't say my ego is tied to a web site, so all comments and criticisms are welcome.
1. I was struck by the high contrast ratio for the Canon i9900 with
Canon Photo Pro paper. In my quick scan, I did not notice any other
printer with a higher number. On the other hand, you only addressed
Epson pigment ink printers. In fairness to Epson, you might cover
an Epson dye printer such as the 1280.
Not having the 1280 was an omission. It was on the original list, which I happily lost during the wee hours one night. I'll get some 1280 plots for both NCA and automatic modes uploaded before long.
2. As others have suggested, I would like to see the ability to
compare two printers in terms of gamut. Not sure what conclusions
one could draw safely, but it still would be nice.
It sure would. Realistically, however, I just don't see having enough time to do this in the near future. Interactive web programming is not something I am familiar enough with to do at any reasonable speed.
As a practical
matter, you already provide the ability to compare contrast ratios.
Intuitively, that should be a very important comparison. (Is my
intuition correct?)
It is for some purposes. As I mentioned in a reply above, I hesitated to even provide this metric. Having high contrast capability is always a good thing - that's why B&W printers worshipped to the gods of DMax. All things being equal, a higher contrast range makes for prints that offer more visual pop. The Noritsu Mytis (a strange dye-sub/inkjet hybrid) is one example. The color range is not the greatest, but the exceptional DMax and range in the shadows makes for prints that leap off the page. Nonetheless, neither color nor contrast range are the be-all and end-all of digital printing. Having good ink linearization and limiting, along with a balanced color gamut, makes for high quality, visually pleasing colors.
3. Obviously, your website is a great promotion for your business.
In terms of assessing whether I should buy profiles from your
company, two questions are obvious:
a. How different is one instance of a particular printer (e.g.,
Canon i9900) from another? In other words, how tight are the
manufacturing tolerances? Do you have any quantitative data?
(Obviously, I am trying to understand whether your profile is
likely to be more accurate for my printer than the one provided by
the manufacturer.)
We do not have a huge number of i9900's profiled yet - a total of 9 - but the numbers look comparable to previous printers. The average printer-to-printer variation is about 5 Delta E-94, with the occasional color up to 10 or more. For reference, the typical contract proof to print match for commercial acceptance is an average of 3 dE and a maximum of 6.
My recommendation is always to evaluate the manufacturer's own profiles first. Canon uses oddball matrix shaper profiles that sacrifice absolute accuracy for smooth transitions. In many instances, these profiles are good enough. If you are happy with their performance, by all means keep your money in your wallet. If you see an overall color cast, or there is a defined color range (usually reds with Canon printers) that is far off, a good custom profile can certainly help.
b. How much does a particular instance of a printer change over
time? In other words, how often does one need to update printer
profiles? Do you have any quantitative data on how things change
over time?
That depends on the printer. We have the most data for Fuji Frontier, Noritsu, and Agfa printers. With all three, we pooled our data with measurements provided by the printer manufacturers. Our findingswere that after six months, our instruments could easily detect a systematic color shift (1-2 dE), but your eyes would be hard pressed to see the change. After a year, changes in the printer make for visible shifts in output color (2-5 dE on average), but still not that bad. That was where our recommendation (also now being provided by Noritsu, Agfa, and Fuji) of a six-month re-profiling schedule for maximum accuracy came from. For many labs, we tie in new profiles to software and hardware upgrades.
Epson and Canon printers have proven to be remarkably stable. Epson's experience with ink formulation changes a few years back obviously taught them something. The exception is the "silent upgrade" Canon made to their PPP paper. This made for a distinct change in output characteristics (avg. of 5 dE). Photographers with critical color requirements are left needing new profiles for PPP. Aside from such gaffes, both Epson and Canon machines seem to be able to go for years without a systematic drift of more than a few Delta E-94. That is remarkable. We see the occasional printer that varies much more; this likely indicates a hardware flaw.
HP printers are more likely to vary with ink/print head changes. HP's paper itself is also inconsistent from batch to batch. Our data points to this perhaps being the result of different paper manufacturing lines - the HP paper surfaces we have measured look to fall into distinct "clumps" of spectral characteristics. We do not get a large HP printers or papers for measurement, so this is equal parts speculation and actual data.
The printer stability champ are the Fuji Pictrography printers. They go for years with no more color shift than what you get between morning and afternoon (less than 1 dE average). One profilie is all that is needed until the printer is put out to pasture or has major repairs.
--
Ethan Hansen
http://www.drycreekphoto.com/ 
Yes, that is a lot of work. Fortunately wishing is free!