Hi Werner, Possible the most simple and quickest way for you to set things up is to simple create a new vertex, normal and tex coords for every triangle corner, then use osg::DrawArrays(GL_TRIANGLES, total_no_of_new_vertices) for the PrimitiveSet, and BIND_PER_VERTEX for the binding of the normals. Note, you'll need to create a new Vec3Array for the vertices form the index and orignal vertex data.
This won't be the most efficient route as it's likely to result in duplicate vertices but at least it should be very straight forward. You can make things more efficient later. Robert. On Thu, Aug 26, 2010 at 12:08 PM, Werner Modenbach <[email protected]> wrote: > Thanks Robert, I get closer to it. > > Just for verification: > > If I have i.e. 10 vertexes and I get 20 calls to the callback method each > defining a triangle. > Each call gives me 3 indexes, 3 normals and 3 texcoords. > > Right now I just push everything into the corresponding arrays and at the end > I call: > geometry->addPrimitiveSet(new > osg::DrawArrays(osg::PrimitiveSet::TRIANGLES, 0, osgVertexIndices->size())); > > You say this will not work because: > 1) geometry->setNormalBinding(osg::Geometry::BIND_PER_PRIMITIVE_SET); > expects just 1 normal for the whole thing (?) > 2) Texture coordinates are allways bound per vertex? > > In this case I have to: > 1) initialize a normal array with Vec3(0.0, 0.0, 0.0) > for each triangle callback I have to look on each vertex index , if the > indexes normal is 0 > if yes: just set it > if no: duplicate the vertex at the end of the vertexes and push_back > the normal > 2) do something comparable on the texcoords > > Is that your recommendation? > > Is there any way to bind normals to the vertex indexes? > > Thanks in advance! > > - Werner - > > Am Donnerstag, 26. August 2010 12:06:34 schrieb Robert Osfield: >> Hi Werner, >> >> >> BIND_PER_PRIMITIVE_SET will mean that you'll have one of the >> associated attribute (i.e. colour, normal) per osg::PrimitiveSet >> attached the osg::Geometry, so if you have one osg::DrawElementsUShort >> used for the triangles then you'll need just one of the associated >> attribute for it. >> >> BIND_PER_PRIMITIVE will apply the attribute per primitive in you case >> per triangle. >> >> Tex coords in the OSG are always bound per vertex, there isn't a >> control to alter this. >> >> If you have a vertex array and indices for your triangles the natural >> thing to do would be to use a osg::DrawElementsUShort(GL_TRIANGLES) >> then use push_back(tri.p1) for each index on each triangle. >> >> I would suggest binding the normals per vertex, you'll have to do this >> for the tex coords as well. If a single vertex will have multiple tex >> coords or multiple normals associated with it due to differences in >> the triangles then you'll need to duplicate the vertices and remap the >> indices on the triangles to fit this. >> >> Robert. >> >> On Thu, Aug 26, 2010 at 10:38 AM, Werner Modenbach >> >> <[email protected]> wrote: >> > Thank you Paul for your response. >> > >> > Yes, you are right, the solution isn't perfect at all. But I'm >> > integrating some external code which is just giving me a vertex array >> > and a per triangle callback with vertex indexes and normal and texture >> > coords. >> > >> > And further more I need some screenshots very urgently for a customer >> > presentation. >> > >> > The performance improvement has to wait :-( >> > >> > So I try to analyse why it's not working. And I detected something I >> > don't understand. >> > In my opinion it should give the same result having: >> > 1) one normal per triangle an binding PER_PRIMITIVE >> > 2) 3 times the same normal and binding PER_PRIMITIVE_SET >> > >> > But it doesn't. So I'm confused. >> > >> > Thanks for further hints. >> > >> > - Werner - >> > >> > Am Mittwoch, 25. August 2010 22:41:14 schrieb Paul Martz: >> >> Werner Modenbach wrote: >> >> > Maybe someone can give me some explanation on my stupid little >> >> > problem. >> >> > >> >> > I have a geometry being composed of TRIANGLES. >> >> > I have a vertex-array, a verted-index-array and a normal-array >> >> > assigned to my geometry. >> >> >> >> Off-topic: Do you need to use the vertex-index-array? It is deprecated: >> >> /** deprecated - forces OpenGL slow path, just kept for >> >> backwards compatibility.*/ >> >> void setVertexIndices(IndexArray* array); >> >> >> >> > 3 indexes and 3 normals per Triangle. >> >> > Normal binding is set to BIND_PER_PRIMITIVE_SET. >> >> > >> >> > Unfortunately the rendering shows the geometry flat in curious dark >> >> > colors and some unexpected light behaviour. >> >> > For analysis purpose I set the normal binding to BIND_PER_PRIMITIVE >> >> > and pushed just 1 normal per TRIANGLE. >> >> > Everything looks good except I see the triangles and the geometry >> >> > isn't smooth. >> >> > The next test is pushing the same normal vector 3 times and setting >> >> > the binding back to BIND_PER_PRIMITIVE_SET. >> >> > In my understanding this should give the same optical result as >> >> > before. Unfortunately it doesn't and everything is dark again. >> >> > >> >> > I guess, I misunderstand something with the normal bindings. >> >> > >> >> > Any hint is highly appreciated. Thanks in advance >> >> >> >> I'm not sure why you don't use BIND_PER_VERTEX, as you stated above that >> >> your normal array contains 3 normals per triangle, so you have one >> >> normal per vertex, right? >> >> >> >> BIND_PER_PRIMITIVE would result in flat shading (only one normal used >> >> for each triangle -- which, by the way, also forces OpenGL down the >> >> slow path), and BIND_PER_PRIMITIVE_SET would be even more granular: one >> >> normal used for each PrimitiveSet you add to the Geometry. This would >> >> make all triangles in the PrimitiveSet either uniformly dark or >> >> uniformly lit, depending on the direction of the one normal used for >> >> the whole group. >> >> >> >> For analysis purposes, I suggest using the normals pseudoloader: >> >> > osgviewer dumptruck.osg.normals >> >> >> >> Hope that helps, >> >> -Paul >> >> >> >> _______________________________________________ >> >> osg-users mailing list >> >> [email protected] >> >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.or >> >> g >> > >> > -- >> > TEXION Software Solutions >> > >> > TEXION GmbH - Rotter Bruch 26a - D 52068 Aachen - HRB 14999 Aachen >> > Fon: +49 241 475757-0, Fax: +49 241 475757-29, web: http://www.texion.eu >> > >> > Geschäftsführer/Managing Director: Werner Modenbach >> > _______________________________________________ >> > osg-users mailing list >> > [email protected] >> > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> >> _______________________________________________ >> osg-users mailing list >> [email protected] >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > -- > TEXION Software Solutions > > TEXION GmbH - Rotter Bruch 26a - D 52068 Aachen - HRB 14999 Aachen > Fon: +49 241 475757-0, Fax: +49 241 475757-29, web: http://www.texion.eu > > Geschäftsführer/Managing Director: Werner Modenbach > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
