#1884 (Bug) calibration very long on Ubuntu and GNU/Debian

+1 0

Closed as Won't Fix
Component: ArgyllCMS
Created by Pascal Obry

Last modified

I found that the calibration is very very long. Asking for very fast calibration (advertised as < 20 minutes) it took me more than one hour. This is strange and possible a recent bug as I don’t remember having to wait this long before. In the log I have found something that could explain this: To be more precise the first patches are always fast. Then at then of the batch it gets longer and longer. For example at the start:

Patch 1 of 32 DE 0.842087, W.DE 0.842087, W.peqDE 0.842087, OK ( < 0.900000)
Patch 2 of 32 DE 0.557485, W.DE 0.557485, W.peqDE 0.738723, OK ( < 0.900000)
Patch 3 of 32 DE 0.446446, W.DE 0.446446, W.peqDE 0.424255, OK ( < 0.900000)

Looks good, then later for patch 28:

Patch 28 of 32 DE 0.812517, W.DE 0.812517, peqDE 1.115908, Repeat
Patch 28 of 32 DE 0.641367, W.DE 0.641367, peqDE 0.938590, Repeat
Patch 28 of 32 DE 0.549064, W.DE 0.549064, peqDE 0.879711, Repeat
Patch 28 of 32 DE 0.224360, W.DE 0.224360, peqDE 0.522544, Repeat
Patch 28 of 32 DE 0.153838, W.DE 0.153838, peqDE 0.289116, Repeat
Patch 28 of 32 DE 0.093847, W.DE 0.093847, peqDE 0.326822, Repeat
Patch 28 of 32 DE 0.165225, W.DE 0.165225, peqDE 0.280074, Repeat (got worse)
Patch 28 of 32 DE 0.226527, W.DE 0.226527, W.peqDE 0.151552, OK ( < 0.234887)

What I don’t get is that on the 3 try DE is < 0.9, no? Last, the context: Ubuntu 15.04 & 15.10 Debian sid Latest version of DispCal in all cases (v3.1). But I had the same issue some time ago with a previous version. Anyway, let me know if you want me to conduct more testing or if you need more information.

5 comments on “calibration very long on Ubuntu and GNU/Debian”

  1. Asking for very fast calibration (advertised as < 20 minutes) it took me more than one hour.

    What’s your measurement instrument? Which version of Argyll CMS are you using?

    I see nothing out of the ordinary in the calibration messages.

  2. I have tested with a ColorMunki and a Spider3. Same result.

    Argyll CMS is version 1.8.3.

    When you said nothing out of ordinary, does that means that the retry for a DE < 0.9 even if the DE value seems below is ok? Maybe I’m not reading the log correctly…

    Thanks for the quick reply.

  3. I have tested with a ColorMunki and a Spider3. Same result.

    Both ColorMunki and Spyder3 are not very fast, so depending on the tone curve target and display black level they’ll take quite some time near black.

    When you said nothing out of ordinary, does that means that the retry for a DE < 0.9 even if the DE value seems below is ok?

    Yes.

    Patch 28 of 32 DE 0.153838, W.DE 0.153838, peqDE 0.289116, Repeat
Patch 28 of 32 DE 0.093847, W.DE 0.093847, peqDE 0.326822, Repeat
Patch 28 of 32 DE 0.165225, W.DE 0.165225, peqDE 0.280074, Repeat (got worse)
Patch 28 of 32 DE 0.226527, W.DE 0.226527, W.peqDE 0.151552, OK ( < 0.234887)

    The repeat threshold for patch #28 is 0.234887 as per the log above. peqDE 0.326822 exceeds that, so it retries. The next peqDE 0.280074 is lower, but W.DE 0.165225 has increased (“got worse”), so it retries again.

  4. Ok, so maybe I should try without the “black point compensation” to see if it makes a difference in speed?

  5. Black point compensation is a profiling option (and should only be used for simple shaper + matrix profiles, not cLUT), it doesn’t influence measurement speed. The calibration tone curve does though, so if you have e.g. aimed for gamma 2.2 (the default) then setting it to sRGB instead may already be a little faster. The screen brightness also plays a role because higher brightness means higher black floor and thus instruments that struggle near black should have an easier time (but ultimately the brightness you aim for should depend on the viewing environment).

Comments are closed.