Commit Graph

6 Commits

Author SHA1 Message Date
Robert Osfield
43049ebcf1 Moved FBO Extensions into GL2Extensions.
git-svn-id: http://svn.openscenegraph.org/osg/OpenSceneGraph/trunk@14586 16af8721-9629-0410-8352-f15c8da7e697
2014-12-09 18:30:28 +00:00
Robert Osfield
74f91037a7 Further work on Bound class/Node::getBound() and Drawable::getBound() and usage in OSG codebase 2014-05-14 16:01:40 +00:00
Robert Osfield
4174d72a52 2014-05-14 10:19:43 +00:00
Robert Osfield
4b9aa1fd24 Replaced deprecated osg::Geometry::set*Binding() usage. 2013-06-27 09:54:12 +00:00
Robert Osfield
ceb97fe230 From Christian Buchner, "Here is a strongly overhauled version of the original osgoit ("order independent transparency") by Mathias Fröhlich. I called this version myosgoit. It looks very nice, just build and run it!
This version adds:

- an encapsulation of the entire Depth Peeling procedure into a class (not currently a scene graph node) for easier integration in other projects.

- compositing with opaque (solid) geometry is possible and the opaque model is only rendered once. This needs to performs some depth buffer blitting between FBOs.

- mix and match with GLSL shaders in the transparent objects is possible, as demonstrated with a 3D heat map intersecting an opaque truck model.


Some Drawbacks:

- the display framebuffer does not receive any depth information from the compositing camera. This could be fixed by compositing with a GLSL shader and writing to FragDepth."

From Robert Osfield, ported the code to work under Linux and without the automatic ref_ptr to C* conversion.
2013-06-25 11:13:50 +00:00
Robert Osfield
00f004fc38 From Mathias Froehlich, "I have now put together what I have for the order independent transparency or
short oit. This rendering technique is also known as depth peeling.

Attached is the example that makes depth peeling work with the fixed function
pipeline. Ok, this is 'old fashioned' but required for our use case that
still has to work on older UNIX OpenGL implementations as well as together
with a whole existing application making use of the fixed function pipeline.
I can imagine to add support for shaders when we have that shader composition
framework where we can add a second depth test in a generic way.

This does *not* implement the dual depth peeling described in a paper from the
ETH Zurich.

This example could serve as a test case for the feature that you can on the
fly remove pre render cameras that you made work a few time ago.
It is also a test case for the new TraversalOrderBin that is used to composite
the depth layers in the correct blend order.
This example also stresses your new texture object cache since you can change
some parameters for the oit implementation at runtime.

You can just load any model with osgoit and see how it works.
Use the usual help key to see what you can change.

There is already an osgdepthpeeling example that I could not really make sense
of up to now. So I just made something new without touching what I do not
understand."
2010-07-12 11:30:15 +00:00