Horde3D

Thanks for the archive! Since I am currently getting more concrete about the problem again I will certainly take a look at your code. Your short documentation gives a good overview!



Personally I like TortoiseSVN very much. It can also do diffs on arbitrary non-versioned files directly from Windows Explorer.



Enabling/disabling depth writing is already possible through the shader contexts but right - depth/alpha testing configuration is still missing.

Ok, since everybody asked for it I have realized the pipeline as resource now.

A camera node requires a pipeline resource and the render function has a new parameter, the camera used for rendering. The setActiveCamera/getActiveCamera functions were removed. This will also make it easy to call the render function several times with an additional start node as proposed by DarkAngel.

Does the camera own the pipeline resource?
Or does it reference it?

Too bad render targets in materials isn't as easy. I'm still messing around trying to find the fastest / cleanest / safest way. Profiling says ugly == better, but I know better.

The camera references the pipeline resource which in turn owns the render buffers. This is conceptually the same as for all other resources. Geoemtry for example is referenced by a mesh and owns the GL vertex buffers.

There will we one new feature which will help to back up this system. It is resource cloning, a function that duplicates an existing resource. So you can create a private pipeline copy for each camera if you want. This will finally also make it easy to do software skinning (ass opposed to hardware skinning in the shaders) since every model can get its private geometry that it can modify.