Torben, Thanks for addressing my questions and providing that information.
We've relied on the projectors and associated software (Mersive) to do the distortion correction and edge blending but a cheaper (monetary-wise) approach is always worth investigating... Thanks, -Shayne -----Original Message----- From: osg-users-boun...@lists.openscenegraph.org [mailto:osg-users-boun...@lists.openscenegraph.org] On Behalf Of Torben Dannhauer Sent: Tuesday, August 21, 2012 4:15 AM To: osg-users@lists.openscenegraph.org Subject: Re: [osg-users] post-rendering warping and off-axis projection Hi Shayne, > > On the distortion correction, is there really any need for pixel level > distortion? That would seem very expensive. I would think that > textureCoodinate distortion is more than sufficient with a reasonable mesh resolution. I haven't seen the need for pixel level distortion in a spherical dome display. > distortionNG supports vertex distortion and texCoord distortion, pixel wise distortion is planned but not yet realized because we do not have a use case for pixel wise distortion. Theoretically it is the most flexible distortion but it thwarts a lot of OpenGLs image quality enhancements like anti-aliasing and makes the image noisy, so you need to apply a filter at a post rendering stage. It could be realized in shaders as it is done in the current( old) osgVisual distortion module, its perfomance is aceptable. With vertex(mesh) distortion and texCoord-distortion you can achieve identical results but I personmally prefere the vertex distortion for the follow reason: If you configure distortionNG to display the distortion Mesh, you can use thes mesh to setup/control each channels correct alignment, whereas with textCoord you cannot do that. > > Does osgVisual support an interface for providing asymmetric viewing > frustums for each channel? What I mean by this is can I specify the > clipping bounds (i.e. rotation, widening or shrinking) for each channel frustum so that I can effectively tile the channels in a dome display optimally? > You can apply any projectionmatrix OSG/OpenGL allows, it does not introduice any restrictions. > > How does osgVisual deal with edge blending of the channels or do you rely on the projectors to do this? > The blending is performed in the distortion Modules, the projectors are COTS devices, only a better color management could be useful to ensure an identical color temperature on each channel. -- So why is distortionNG still marked as experimental if it works like a charm?! Maybe I should explain the current situation: The above mentioned 6 channel helicopter simulator uses a proprietary commercial setup system: It is camera based and its output is the blendmap, the projection matrix and the distortion matrix used by distortionNG. It works like a charm, but it is very expensive. For the academic usage osgVisual aims for, a cheaper solutions should be available: Therefore I'm also working on a mouse based setup method. This is very easy for vertex/texCoord distortion setup, but it is not so easy for the projection Matrix or the blend map. The manual setup mode is still unter construction. The manual setup should be done in the following order (not yet finishend, it may contain errors) * Position the projectors to cover every desired sqare inch of the screen, ensure that there is a reasonable overlaping of the projectors to use for blending purposes. * In each channel display the distortion mesh and align it with the mouse and setup a regular mesh structure across all channels on the whole screen. ensure that each distorted channel covers a regular, ideally rectangular area from the simulator eye point perspective, this simplyfies the design of the projection matrix. * Not yet solved because problematic: The blendmaps must be "designed" interactively with the mouse. The blend function between black and wite must not necessarily be linear, but maybe it can be simplyfied be regarding it as linear. The black blending edges are simple, they are the edge of the channel projector (blending maps are not distorted!) . The inner (white) blending edges have to be defined by mouse. If required the non-linear blend function must be definable. * Not yet solved because problematic: the projection Matrix has to be defined. To set it up correctly, the position of each channel relative to the simulators eye point must be determined. A cheap theodolith may be required. Finally the projection matrix has to be defined by this measured angels relative to eypoint. How to do this interactively I have no clue (yet) :) The distortionSet itself is a container which can by saved and restored (like a path file of the animation manipulator) once it is configured by one of the above methods. Finally one thing to mention: Lots of this tasks may be easy - I simply do not know because my time is limited due to my PhD and my daily work (both not in this area) and it is hard to get free hours and anough power to work on distortionNG. I hope this changes dramatically after my PhD is finished. Once distortionNG is functional I'll try to submit it to core OSG. Thank you! Cheers, Torben[/b] ------------------ Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=49382#49382 _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.or g _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
