Ben Goren wrote:
Does the calibration really need to be saved to a file at all, then? Why not just keep it in memory and never write it to disk?
It's because of the nature of the Argyll toolset. Each tool is a separate executable process that gets created and destroyed. So to keep the cal. in memory would require some sort of independent server, and that would be complete overkill. Saving the calibration in a file is simply to allow the -N option to work, so that a sequence of tools can be used with one calibration at the start of the sequence.
...and, any chance of adding an option to have chartread (and maybe dispread) pause every N patches for a recalibration? Perhaps even a sliding scale with frequent recalibrations early on, and longer intervals if the instrument appears to have stabilized?
It would be nice to add something like this. I have added it as a manual option to spotread, and a similar thing could be done with chartread, but I'm not sure how it would be managed with dispread.
I'm also thinking it might be helpful to have some sort of an auto-test mode that calibrates, runs the lamp for N seconds, calibrates, repeats, and reports when it's stabilized. And perhaps even an option to have the instrument warm up with the lamp on for x seconds before starting to read the chart.
I'm not sure that would work. Certainly for the i1pro the lamp has it's own thermal dynamics, and it's hard to model (I've tried). The thermal dynamics gets more obvious as the lamp ages (which is why they should/would replace it when sent in for recalibration). For all their disadvantages (lumpy spectrum and lack of UV), LED's have the advantage of not suffering so much from thermal dynamics and aging, and the thermal compensation in the Munki works very well. The munki readings vary imperceptibly from one to the next on the same spot, irrespective of the timing, whereas excellent consistency for the i1pro demands exactly the same timing, or letting it cool down completely between readings. In practice it's not a huge problem, since repositioning errors/variation dominate ie. the paper location variance is of the order of 0.3 delta E for most media, which is larger than the lamp thermal dynamics errors. By letting the lamp cool down between readings and carefully not repositioning the instrument I can get reading consistency of about 0.005 - 0.020 delta E. with the i1pro, similar to the Munki. When it comes to strip reading, there are worse problems (strip edge detection, total integration time, and lamp thermal changes), and once again the LED with thermal compensation has an advantage with regard to the latter. Graeme Gill.