Hi Doug,
How do you use Argyll to create B/W profiles? My understanding from Graeme is that it does not support this feature. I can set Argyll up to read B/W patch sets of various types but have been using the QTR script tool to create the profile. My routine is to use a 51 step gradient as opposed to 21.
Argyll doesn't make B&W only profiles unlike QTR that does. But that doesn't prevent making standard "color" profiles that are used for B&W.
This is actually what Canon does with the Pro1000. They use the same color profiles (Limited to the Canon supplied ones - no custom allowed) but run the data through the B&W driver process.
When I first noticed that I thought Canon was just using the same driver but converting RGB to B&W. This isn't the case. Canon's B&W mode is completely different than the color mode and it makes, like Epson, very neutral prints with only small excursions in a* and b*. The deltaE between the two modes on a neutral ramp is as much as 10 due almost entirely to the a* and b* differences.
The L* delta varies about +- 2 between the two modes and averages under .8 which allows them to use their existing profiles for B&W.
Getting back to Argyll, B&W profiles can be considered a subset of color profiles but where the LUT values for a* and b* are 0 and the output's 3 channels, which will already be close to each other are treated as B&W by the driver. If, in fact, the LUTs produce the same output for B&W PCS Lab values, then the profile is effectively operating as a true black and white profile.
The problem is creating such a profile. Colprof and I1Profiler require RGB values and a color gamut. My solution is to provide them one by synthesizing Lab values that are not along the neutral axis in such a way that they intercept the B&W printer neutrals. This means printing and measuring a 256 RGB neutral patch set. Then the a* and b* of the measured Lab values are set to 0. Next, Lab values are synthesized by adding a grid and creating "colors" to fill out the needed, but unused in B&W, values. Then the data set is loaded by I1Profiler or Argyll's colprof and a profile is created.
And they make great B&W prints since the LUTs don't have to correct for a* and b* variations off 0. Much smoother and more accurate than making B&W using a full color workflow.
Now, if I could find some direct way to print using them w/o the workaround described in my first post.