|
From: Radek <tru...@o2...> - 2008-07-23 00:03:41
|
Hello, I have some questions regarding the possibility of having subpixel accuracy in drawing apps (like GIMP or so). I'll try to describe the problem the best I can: Lets say I run GIMP with an image viewed at 25% zoom, that is, one pixel on display correnponds to 4x4 square on the image. Using the tablet I draw a curve with a pencil tool (1x1 square brush without antialiasing). I zoom image to 100% and see that the curve is not smooth, but consists of short, straight sections either vertical, horizontal or diagonal, like if it was snapped to 4x4 pixel grid. I know that this is because GIMP gets the stylus coordinates with accuracy limited to the resolution of my display. If there was a way to make it receive (and properly interpret) stylus coordinates with an increased precision (so that it would be limited by tablet's resolution rather than display's), it would work as it should. I've been googling to find a solution for this, but I've only found out that it's a Xorg limitation and nothing can be done unless Xorg supports it. Despite that, I think that there can be some workarounds. One that came to my mind is to create some kind of virtual display with a size in pixels close to the working area of tablet (so that the drawing window of GIMP could be sent there), and put it on the screen zoomed out. What do you think about it? Is it even possible? I want to say here I'm not a linux guru nor a pro coder, so probably I coudn't fix it myself. But this issue is rather important for me, because it feels a little pointless of having a tablet of a high resolution while not being able to take any advantage of it. Or maybe I'm just missing something? Best regards, Radek |
|
From: Alexia D. <ale...@gm...> - 2008-07-23 06:31:10
|
On Wednesday 23 July 2008 03:03:34 Radek wrote: > I'll try to describe the problem the best I can: > Lets say I run GIMP with an image viewed at 25% zoom, that is, one > pixel on display correnponds to 4x4 square on the image. Using the > tablet I draw a curve with a pencil tool (1x1 square brush without > antialiasing). I zoom image to 100% and see that the curve is not > smooth, but consists of short, straight sections either vertical, > horizontal or diagonal, like if it was snapped to 4x4 pixel grid. Please include an image. What you describe right now sounds pretty normal for a 1x1 not anti-aliased brush. > I know that this is because GIMP gets the stylus coordinates with > accuracy limited to the resolution of my display. Nope, no it does not. Gimp 2.5 does limit accuracy in some cases tho, for performance and cutting noise. It is a bit of code that could use some tuning, and perhaps you are willing to test? > If there was a way to > make it receive (and properly interpret) stylus coordinates with an > increased precision (so that it would be limited by tablet's resolution > rather than display's), it would work as it should. Leaving it entrely to tablet is pointless and generates noise, because you simply aren't able to control you stylus with that precision, but that limit tho related to display resolution should go unnoticed for the user. > I've been googling to find a solution for this, but I've only found out > that it's a Xorg limitation and nothing can be done unless Xorg > supports it. Xorg supports it. Gimp supports it. Please explain what you expect. > Or maybe I'm just missing something? Yep, missing something. |
|
From: David De La H. G. <dav...@gm...> - 2008-07-23 21:35:21
|
2008/7/23 Alexia Death <ale...@gm...>: >> I've been googling to find a solution for this, but I've only found out >> that it's a Xorg limitation and nothing can be done unless Xorg >> supports it. > Xorg supports it. Gimp supports it. Please explain what you expect. > Well, bearing in mind that some versions of Xorg, have the device coordinate rescaling problem, the workaround for which in linuxwacom drivers significantly reduces tablet resolution. |
|
From: Radek <tru...@o2...> - 2008-07-27 00:28:51
|
On Wed, 23 Jul 2008 22:35:19 +0100 "David De La Harpe Golden" <dav...@gm...> wrote: > > Well, bearing in mind that some versions of Xorg, have the device > coordinate rescaling problem, the workaround for which > in linuxwacom drivers significantly reduces tablet resolution. > I'm using Ubuntu Hardy with default Xorg, which has this bug, so at least I know the source of my problem. On Wed, 23 Jul 2008 09:31:08 +0300 Alexia Death <ale...@gm...> wrote: > > Xorg supports it. Gimp supports it. Please explain what you expect. > > > Or maybe I'm just missing something? > Yep, missing something. > So, Xorg *supports* it, but it's not working. As far as I know, the whole 1.4 branch of X server is affected, and the fixed version won't be included in a stable release any time soon. What can I do to make it work? Do I need to build Xorg unstable from sources? |
|
From: David De La H. G. <dav...@gm...> - 2008-07-27 03:09:26
|
Radek wrote: > So, Xorg *supports* it, but it's not working. As far as I know, the > whole 1.4 branch of X server is affected, and the fixed version won't > be included in a stable release any time soon. Well, the more gung-ho distros (see: fedora 9) are already shipping 1.5 release candidates*, and there's one in debian/experimental and ubuntu/intrepid too, but: Errr. It looks like the xorg xserver 1.4.2 release actually included a version of Magnus' rescaling fix. However, the linuxwacom configure script doesn't know that. It probably should? If you're running xorg xserver 1.4.2**, configure linuxwacom with ./configure --disable-quirk-tablet-rescale and you get hires coords. Though I guess it's possible that there are other problems with 1.4.x that merit the quirk setting, Magnus and/or Ping would know better. * Though there may be new bugs there, of course... http://lists.freedesktop.org/archives/xorg/2008-July/037528.html "[xinput] coordinate system of root/x/y" from Thomas Jaeger ** should be able to do an "X -version" at the command line, it outputs server version and exits rather than trying to open the display. |
|
From: David De La H. G. <dav...@gm...> - 2008-07-27 03:14:32
|
David De La Harpe Golden wrote: > * Though there may be new bugs there, of course... > http://lists.freedesktop.org/archives/xorg/2008-July/037528.html > "[xinput] coordinate system of root/x/y" from Thomas Jaeger > That should be "[xinput] coordinate system of root_x/y" |