Hello Jan,
Thank you for your extensive reply.
I actually use OIS in a framework (apart from rendering) and found it
quite useful in my environment (distributed, yet input is local), so I
simply plugged it in my renderer.
Of course I use the keyboard events provided when working on osg-based
applications, but not all of them are ;-)
But VRPN seems like another tool I'd might use later (when some clients
are driven by external input, such as driving wheels, trackers etc).
Btw: The OIS Implementation for the SpaceMouse does use DirectX on
windows, never tried it on Linux so far. In the old implementation we
indeed used the SDK sample, which was based on directX too, but never
worked to our satisfaction.
cheers
Sebastian
On Mon, Feb 11, 2013 at 7:22 PM, Sebastian Messerschmidt
<[email protected]
<mailto:[email protected]>> wrote:
Hello Jan,
Actually VRPN looked a bit too bloated for me. Also I didn't find
a simple example grabbing keyboard events, directly accessing
keyboard state etc.
Is there some good starting point into VRPN?
Well, it is a different approach than OIS. OIS is a for local
applications using local resources, VRPN is based on a client-server
concept, where you have one server handling the device and several
clients connecting to it at the same time (even over network). I
suggest to first have a look at this short paper:
http://www.cs.unc.edu/Research/vrpn/VRST_2001_conference/vrst_vrpn_paper_reprint.pdf
- it explains the architecture and rationale behind the design.
The advantage of VRPN is network transparency (tracker/device doesn't
need to be on the same machine as your application - useful for larger
setups), device abstraction (no need to make your app depend on
various proprietary SDKs that often conflict, change, etc), large
amount of supported devices that can be swapped by changing the config
file, no need to modify your application at all, etc.
It all comes down to what exactly you need. If you need only a small
desktop application that will always support only one type/brand of
the 3D mouse, you are probably better off using the vendor SDK. If it
is something where you need to be flexible with regards to the hw
available at the deployment site, VRPN is much better.
If you don't need to write custom servers, the client API is very
simple, there are good tutorials here:
http://www.cs.unc.edu/Research/vrpn/vrpn_getting_started.html
If you want to interface e.g. a keyboard, that would be typically
exposed as a vrpn_Button class, sending you the scancodes of the
individual keys. However, that is probably not the method you want to
use to make a text editor (it is a bit rudimentary). A more complex
device, like the SpaceNavigator gives you both vrpn_Analog (for the
joystick axes) and vrpn_Button (for the buttons) classes. That can
then be routed on the server using a config file into the
vrpn_Tracker_AnalogFly virtual tracker that will make a full 6DOF
tracker returning position and orientation in space for you. Look at
the example vrpn.cfg that comes with the library. Your application
doesn't know anything about the tracker emulation - it simply talks to
a vrpn_TrackerRemote instance. If you point that instance to a
Polhemus or Vicon server instead, it will work "automagically",
without you having to change anything in your code to support a
completely different tracker.
BTW, If you need keyboard events and such, OSG supports that already,
why to duplicate that functionality? OIS doesn't give you anything
more in that regard.
In my case, I am usually relying on either OSG event system or Qt to
provide the handling of mouse and keyboard and use VRPN for everything
else - gamepads, wiimotes, 3D mice, all kinds of 3D tracking systems,
etc.
Regards,
Jan
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
