I noticed the valuable input from Mark regarding the tests and the fact that the media were not the same across both devices. While I totally understand the comment regarding the fact that the media was not the same on both devices for the optical density test and the additional fact that proofing paper was used for the gamut analysis, the test was set up in such a way as to provide a fair comparison with equivalent papers from the vendors' own brand media library, thereby allowing a reader to see what the results would be achieved. We could of course have chosen a media that worked on both devices (and that would be an excellent test that we would happily take on if requested to do so) but in this instance we would not have known if a third party media was more of a 'perfect fit' for one vendor than the other and hence we could have been adding in a bias to one vendor over the other, which is exactly what we try to avoid happening.
Kind regards
David Sweetnam
Director of BLI Research and Lab services, EMEA / Asia
Hi David,
Yes I understand the reason why you used each manufacturer's OEM papers in the test of their printers. It does show what each achieves in "its own backyard" so to speak, which may be of interest to some people, so it's not a case of it being a "wrong" approach. The issue is whether it is the appropriate approach relative to the purpose. If the purpose is what I just mentioned, fine. However, if the purpose is to compare the inherent capabilities of the printers themselves, then it is not fine, because paper makes a huge difference to outcomes. So if we just want to know what the printers themselves are capable of, it is essential to neutralize the paper variable by using the same media in each machine.
Then the question is what media? Again, purpose driven. If the purpose is to show the maximum black and the maximum gamut each printer can produce, I would select a well known very high gamut paper such as Ilford Gold Fibre Silk, or Canson Baryta Photographique or their likes, because of their outstanding properties in these respects. Then I would use the best available profiling package to make custom profiles for each paper based on the rendition of the profiling targets from each of the test printers - because let us always remember, those gamut diagrams and volume readings come from the profiles, not the prints themselves. So the results also depend very much on the quality/properties of the profiles. However, if my purpose were not to see maximum quality possible, but maximum quality on more difficult papers ( such as most matte papers), then of course using a fine matte medium makes sense.
Turning to your results, as I said, I can't help thinking something unusual is going on. Your gamut volumes for both printers using the glossy photo paper just seem very low, relative to my experience with my Epson 4900, which has a very similar print engine to the 9900. As well, while I have no experience measuring gloss profiles for the Canon 8400, given how competitive these machines are, it would surprise me to see such a large difference of gamut volume between my 4900 and your Canon 8400 for the gloss paper. The proofing paper results are within expectations.
Now - one more thing which one seldom sees outside of printing bureaux, and I have been playing around with lately (curiosity, I'm not a commercial service) - is accuracy of the results. Very often accuracy is not the purpose. It may well be for various scientific, medical, forensic and marketing purposes, but for most "fine-art" and casual photography it is not. Where it is, one likes to know whether the prints really reflect the file values. That of course is basically a test of the profile quality, and whether the combination of the profile and the printer can deliver an accurate result. I do this (and I have no doubt there are more sophisticated ways) by round-tripping a purpose-built file of the Gretag-MacBeth ColorChecker (24 patch version) and calculating the dE values from the differences between the file numbers for the GMCC and spot readings of the 24 patches off the prints once they are dried. You may or may not be surprised by how good or how bad these dE values can be! I think in professional printer evaluation comparisons this may be an interesting test to conduct if the client were prepared to pay for it.
So there you have an extended explanation of my concerns with those aspects of the tests. I do a fair bit of product testing for reviews published on this website, so I am well aware of numerous conceptual difficulties and pitfalls one encounters along the way doing this kind of work - those who know what they're looking for and have tried it will fully understand the conundrums and the issues, so I appreciate that your work is not easy. All the best.
Mark