|Original post by rbarris|
The wording on the Blizzard slides was intentionally terse (I never like presentations where the presenter is just reading the words on the slides); if you have any questions about the content in that segment I would be happy to try and respond to them.
There are a couple of new features that I'd like to know a bit more about from a game developer's perspective:
-I've read the GPU Gems chapter on using transform feedback in D3D10 for increasing the number of morph targets you can blend between for facial animation. Any more ideas on areas where Transform Feedback will be useful?
-Does using Transform Feedback make it harder/easier to balance feeding the pipeline? Is it just a case of "well, I'm fill bound so I can afford a few extra feedback loops at the vertex stage"?
-I'm pretty sure I understand what this does, and how it improves use of occlusion queries by removing the need for the app code to handle the result. It sounds like it'll make using occlusion queries successfully even harder to understand, but the implementation much easier once you understand how to do it. Any tips? Is there any maximum number (or suggested 'best' number) of draw calls that can be queued up as part of a conditional render?
-any tips on handling rollout of apps using different profiles? I've read a bit on this on the opengl.org forums, but I'd be interested in a game developer's perspective.
Edit: almost forgot GLSL [smile]
Removal of Built-In uniforms
-as GLSL has nothing like HLSL/FX's 'semantics' the built-in uniforms were handy for exchanging shaders between apps. Does the ARB have a better solution, GLFX maybe? (sorry, I've been unable to find anything on the recent GLFX update)
-is this for point-sampling?