Sign in to follow this  
Cathbadh

Transform Feedback, Parallel rendering

Recommended Posts

Cathbadh    100
I've been looking at deferred shading as a way of lighting my scenes initially because it would allow me to do SSDO and motion blur rather simply, but it got me thinking of other potential advantages, and I want your take if they are actual advantages. For one, with deferred shading I would be able render the scene from the camera and all the lights (for shadow-mapping) in parallel, as in on a per object basis. That is, for each mesh in the scene, render it to camera space, and then render it to all the light spaces and reflection spaces before going to the next object. I'm thinking that I could get better cache-hit ratio this way? Then I was thinking that I could expand on this and perhaps reduce some redundant calculations in the shader (at the cost of some memory) to use feedback buffers to store the result of some complex calculations the first pass through instead of recomputing them over and over for each pass. For example, skeletal skinning. I could calculate the positions in modelspace of each vertex given the skeletal orientation and data for the camera-space render pass, and store it into a feedback buffer so that the lightspace passes can just refer to this buffer for the positions instead of doing the expensive skinning computations over again. As far as I can see, these abilities are only afforded by doing deferred shading. Are these worth implementing, or am I just fooling myself? I could imagine that drawing a single mesh and then switching FBO render target and shaders all the time might incur a lot of flushes of the pipeline, but I'm not sure.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this