From: Jason G. <kil...@gm...> - 2012-01-04 19:33:07
|
On Wed, Jan 4, 2012 at 11:07 AM, Jason alavaliant <ala...@gm...> wrote: > On Thu, Jan 5, 2012 at 7:12 AM, Jason Gerecke <kil...@gm...> wrote: >> On Wed, Jan 4, 2012 at 12:45 AM, Jason alavaliant <ala...@gm...> wrote: >>> On Wed, Jan 4, 2012 at 5:48 AM, Jason Gerecke <kil...@gm...> wrote: >>>> On Tue, Jan 3, 2012 at 1:58 AM, Jason alavaliant <ala...@gm...> wrote: >>>>> Hi, >>>>> >>>>> I know how to calibrate a Cintiq in Linux when it's the only monitor >>>>> attached to the computer. However what if there is a second monitor >>>>> attached as well (twinview using 'Coordinate Transformation Matrix' to >>>>> map to the Cintiq) the values needed for calibration are quite >>>>> different. >>>>> >>>>> For example with the Cintiq I was testing with; >>>>> >>>>> 0, 0, 86680, 54180 >>>>> is the default tablet area >>>>> >>>>> 0, 0, 86680, 62180 >>>>> seems to be the right values for just having Cintiq connected. >>>>> >>>>> However if I have a second monitor attached on the left and map the >>>>> tablet to the right monitor only using Coordinate Transformation >>>>> Matrix values I have to use >>>>> 808, 365, 87419, 54407 >>>>> (worked out very slowly by hand by changing values bit by bit by hand) >>>>> is what is needed to make the calibration line up. I'm sure >>>>> there is a way to calculate these values directly but so far how to do >>>>> it is escaping me. Has anybody out there worked out the sums >>>>> required? >>>>> >>>>> Thanks >>>>> -J >>>>> >>>> >>>> The calibration shouldn't need to be changed if the proper Coordinate >>>> Transformation Matrix is used. The raw coordinates from the tablet are >>>> first scaled according to calibration and then transformed onto the >>>> proper monitor by the X server with the matrix. >>>> >>>> Can you provide the output of the command 'xrandr', as well as the >>>> value of the Coordinate Transformation Matrix? >>>> >>>> Jason >>>> >>> >>> hmm interesting, I think I have the transform matrix correct. I >>> can't give you the exact values (since I don't have the Cintiq on >>> hand, it's not mine, I'm just helping the owner set things up), I use >>> the same formula to workout the values on all computers so on another >>> similar setup I have; >>> >>> left monitor; 1280 x 1024 >>> right monitor; 1680 x 1050 >>> (We are using the nvidia proprietary driver so I don't think xrandr >>> would give you much useful info since it doesn't support randr 1.2 ) >>> >>> with >>> Coordinate Transformation Matrix (119): 0.567568, 0.000000, 0.432432, >>> 0.000000, 1.000000, 0.000000, 0.000000, 0.000000, 1.000000 >>> to map the tablet onto the right monitor. Which looks to work fine >>> with Intuos tablets I normally work with. >>> >>> -J >> >> That matrix looks good to me. I just did some testing to make sure >> things are running as expected over here, and I wasn't able to >> reproduce what you're seeing. Calibration remained correct after >> enabling a second monitor to the left of the Cintiq, once the >> Coordinate Transformation Matrix was updated. You might double-check >> the resolution and position of both monitors in nvidia-settings. If >> the monitors are sharing the bottom edge instead of the top or if >> there is dead space between then, the matrix would be incorrect. >> >> Jason >> > > They are sharing the bottom edge (it's the default for twinview and > what we use on our corporate workstations since we can write one > xorg.conf that works on all hardware automatically for that but lining > up at the top requires positing each screen absolutely which requires > custom per monitor setup to line them up). Should I be adjusting > the matrix values for how our monitors are lined up? Lining up the monitors along the bottom would normally require adjustment, though (now that I think about it for more than a second) it doesn't in this particular case. The monitor that the matrix is mapping to is already 100% of the desktop height, so the math will come up with the same results. If lining up along the bottom created dead space above the monitor, you'd need to add in a factor for the vertical offset. What version of X server and xf86-input-wacom is running? I'll see if I can match my setup more closely to yours. Jason --- Day xee-nee-svsh duu-'ushtlh-ts'it; nuu-wee-ya' duu-xan' 'vm-nvshtlh-ts'it. Huu-chan xuu naa~-gha. |