On Samedi 04 Février 2006 03:24, Graeme Gill wrote: > Hmm. I guess I wasn't expecting anyone to leap into the middle of the > pool straight off. At the current point in time, only a single headed > configuration is supported - sorry about that! (I'll update the > documentation to reflect this current limitation.) > > Even if I expanded the software to be aware of multiple screens (which I > was intending to do at some later stage), I'm not sure this will help > you a great deal, as (from a lot of discussion I've read on the > colorsync list), there are currently issues with operating system > software, video card drivers, and video cards supporting different LUT > on two screens. Apparently, it works sometimes on OSX, and very rarely > on MSWindows. I don't know where X11 stands in this regards, and I'm not > currently is a position to test it easily. > > I'm also under the impression that X11 has two ways of doing multiple > screens, one of which makes it difficult to detect what display things > are on. > > Can you set your DISPLAY variable (within the shell you're running > Argyll from) to point at the CRT ? If you can, then running > dispcal/dispread should wind up on the right screen. It certainly > won't work if the two screens appear as one to the utilities. As I only need to calibrate my CRT, I patched the dispwin.c file to force a virtual screen width in order to get the patches displayed at the center of the CRT screen (line 119 and following). A good solution could be to provide cli options to let the user enter the position and size of the color patches... > > To make some more tests, I reduced the resolution of the CRT screen, > > so patches are now displayed at the right of the LCD panel. > > > > 2) If the brightness of the LCD is too high, I get the following error > > during 'dispread' use: > > > > Sample read failed due to misread > > > > If I reduce the brightness (3 or 4 levels), it works. Why? Could it be > > a hardware problem? > > Sounds like it. I was under the impression that the DTP92 is only > suitable for CRT displays though. I believe that the DTP94 works > with LCDs (I don't have one, and I think it is USB, although it might > do serial as well.) I've never tried using the DTP92 on an LCD, > I've used the Spectrolino on an LCD, which works for both. > > > 3) During 'dispcal' calibration, a lot of patches remain green. And > > the calibration is totally wrong. > > Which patches are you talking about ? The test patches ? What do you > mean "remain" ? > > The test pattern starts out with extreme devices values (RGB white & > black), and then does a series of neutral (or near neutral ramps), that > should improve with each round. > > > Also, if I use 'dispread' with a .cal file, I > > > > get the following error: > > > > Dispread: Error - docalibration failed with return value -1] > > This indicates you had the -a flag set. If you #define DEBUG in > dispsup.c, recompile and re-run, you might get more detailed error > information. You where right: the DTP92 only works on CRT. With patches on the CRT, all is now working fine. I attached a quick made profile of my CRT. Does it look like a correct profile ? (it is named LCD, but it is my CRT...) > > 4) Is there a way to reset the LUT values after 'dispread' have > > modified them ? > > It shouldn't leave them in too poor a state. In the next release there > is a new utility "displin" that creates a linear ramp .cal file, that > can then be fed back into dispread, to reset the display LUT values, and > you can then kill dispread, to reset the display. (Attached is a linear > .cal file). Thanks from the .cal file. > Again, I'm not sure if I can help you, since this isn't my build script. > If you were attempting to compile using JAM, I could probably steer you > in the right direction. > > > Tools from the binary package works fine. > > Since I compiled these, I might be able to help you if they were > misbehaving. Compilation with jam works, but I had add '-f ../Jambase' option... > If we can get that far, it might be possible to work on how to access > each screen individually. Great if you can, but as I said, I only care about the CRT, used to display the pictures. The alternate solution I proposed (option to let the user give the position and size of patches) is OK for me. Guys, thank you very much for your help :o) Regards, -- Frédéric http://www.gbiloba.org