Thanks Sebastian.I could of thought of this. Let me give it a shot. Thanks
again!
Nick
On Mon, Dec 1, 2014 at 9:26 AM, Sebastian Messerschmidt <
sebastian.messerschm...@gmx.de> wrote:
> Hi Nick,
>
> One way to improve this is to place your "root" somewhere near the center
> of your instanced geometry and subtract the center from your instances
> positions.
> This way the coordinates to be multiplied stay small in relation (as the
> big precision loss due to View*Model is done on the CPU side (during cull)
> instead of multiplying big coordinates one the GPU with the inverse.
> So decorate your CustomGeometry subgraph with a transform and modify your
> instances position accordingly.
>
> Hi Community,
>
> me again on this. I have it working, however, it looks really bad when
> using large coordinates (when placed on UTM or Geocentric). The model is
> somewhat distorted due to precission issues. I know how to avoid this, like
> working with double matrices, but when I pass double matrix on the vertex
> shader it kills the performance - from 60Fps to ~1 fps. Bellow is the
> vertex shader. Any hints?
>
> Thanks a bunch as always
>
> #version 150 compatibility
>
> uniform mat4 osg_ModelViewProjectionMatrix;
> uniform mat3 osg_NormalMatrix;
> uniform vec3 lightDirection;
>
> in vec3 vPosition;
> in vec3 vNormal;
> in vec2 vTexCoord;
> in mat4 vInstanceModelMatrix;
>
> smooth out vec2 texCoord;
> smooth out vec3 normal;
> smooth out vec3 lightDir;
>
> void main()
> {
> gl_Position = /*osg_ModelViewProjectionMatrix*/
> gl_ModelViewProjectionMatrix * vInstanceModelMatrix * vec4(vPosition, 1.0);
> texCoord = vTexCoord;
>
> mat3 instanceNormalMatrix = mat3(vInstanceModelMatrix[0][0],
> vInstanceModelMatrix[0][1], vInstanceModelMatrix[0][2],
> vInstanceModelMatrix[1][0], vInstanceModelMatrix[1][1],
> vInstanceModelMatrix[1][2],
> vInstanceModelMatrix[2][0], vInstanceModelMatrix[2][1],
> vInstanceModelMatrix[2][2]);
>
> normal = osg_NormalMatrix * instanceNormalMatrix * vNormal;
> lightDir = lightDirection;
> }
>
> On Tue, Nov 25, 2014 at 12:06 PM, Robert Osfield <robert.osfi...@gmail.com
> > wrote:
>
>> Hi All,
>>
>> On 24 November 2014 at 09:28, Sebastian Messerschmidt <
>> sebastian.messerschm...@gmx.de> wrote:
>>
>>> The example I was looking for was in the submissions, called:
>>>
>>> [osg-submissions] New example : culling and LODing performed on GPU side
>>>
>>> Unfortunately Robert didn't approve or merge the nescessary changes.
>>> Maybe it went unnoticed.
>>>
>>
>> I am still progressing through the submissions backlog and am delighted
>> to say that this wee gem of an example is now checked into svn/trunk - look
>> for osggpucull. I haven't had a chance to spend the time understanding it
>> yet, but the code is there.
>>
>> Robert.
>>
>>
>> _______________________________________________
>> osg-users mailing list
>> osg-users@lists.openscenegraph.org
>> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>>
>>
>
>
> --
> trajce nikolov nick
>
>
> _______________________________________________
> osg-users mailing
> listosg-users@lists.openscenegraph.orghttp://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>
>
>
> _______________________________________________
> osg-users mailing list
> osg-users@lists.openscenegraph.org
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>
>
--
trajce nikolov nick
_______________________________________________
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
