dispcalGUI 0.7 @ rwx³

My yesterday held workshop for calibrating and profiling monitors using dispcalGUI+ArgyllCMS in Nuremberg was a nice experience. Around ten people from the openSUSE Conference gathered, all being eager to do something for their health of colour vision. We wanted to create ICC device profiles and collect them for later publishing. Following is a small report and review from the workshop.

We had available a i1display, a DTP94 and a i1pro as measurement devices, which no of the attendees owned privately. Installation on openSUSE-11.4 went pretty smooth. ArgyllCMS is in the multimedia:photo repository and dispcalGUI is in multimedia:color_management. Both are easily searchable through the http://software.opensuse.org URL. One hacker had all packages installed in advance and could start instantly with the i1display. With the default settings in dispcalGUI appeared a small terminal and required to adjust native monitor settings or continue with calibration. We pressed number 7 and continued with the calibration part. This lasted relatively long. dispcalGUI and in the background ArgyllCMS took quite some work to iterate over the calibration for four times with lots of "regression getting worse" style messages. This indicated to us, that it reached an end for improvement or the application where not satisfied according to each persons like. The profiling and installation finished after quite some time and the new calibration and profile could be used in colour managed applications. In Gimp the monitor profile was not used by default. The Colour Management tab in Gimp's preferences needs to manually enable the system profile, which is a troublesome exercise to figure out. Instead colour managed applications should look first at the system profile. Overriding the system profile by default is dangerous for a good user experience.

The next people became ready and used the DTP94 in between. Unfortunately there where four people failing to use it. What looked like a defect DTP94 device cable of my best colorimeter, was observed by three users as continuous USB device ID switching. One could spot the reason to a conflicting UDEV rule of libmtp. I filled a bug report on SourceForge including the patch of that user. With that fix the DTP94 worked again. I am happy to can use the DTP94 in the future.

I figured out how we can speed up the calibration, by reducing the calibration quality, which was nice in such a big group. One attendee had a repeated black switching of his laptop monitor during the whole process in connection with the nv driver. That persisted even during the profiling stage, which makes no sense to me. I suggested him to install the nvidia proprietary driver especially for his installed Quadro card.

In several cases dispcalGUI crashed without providing further informations on the command line. We had then to repeat the process. Luckily a longer break allowed us to finalise most ICC profiles. I requested the monitors EDID data block from all people, to have the device information available alongside the profile. This can be done by installing the oyranos-monitor package, which brings relatively many dependencies, like a default profile set and several libraries and modules. oyranos-monitor -f edid -o monitor.edid could then be used to collect that data. Some installations did not work as expected and I have to look into the issue. The new profiles need now further preparation before they can be uploaded to icc.opensuse.org.

It was fun to see how most issues where resolved in the group, and that most people could get a device profile for their usage. Registration of a smaller number of people would be helpful to ensure no overload of capacities. Between, dispcalGUI has a homepage with further detailed informations.