Which shading method for deferred rendering

I have a simple and small question, just woundering what method of shading would be best suited for deffered rendering. Also could someone provie me with a link to the pro’s and con’s of each shadowing method?

With “shading” you mean “shadowing” ?? Those are two entirely different terms. “Shading” means really everything you can do to color a pixel, but it is usually used to mean something like “lighting”. “Shadowing” means how to compute shadows of a light-source and is in generell only one tiny piece of the whole shading that you do for each pixel.

For deferred rendering you can use any shadowing technique, that you can use in forward-rendering, too. These days “shadow mapping” is the most popular, but in some areas “stencil shadows” are a good solution, too. Depends on what you are doing and what results you expect.

With those two keywords google will give you a lot of information.


Sorry I did mean shadowing both times. Thanks for the information, Im suprized that there is not one stand out method of shadows for deferred shading thought… Anyways I think im going to look itno CSM’s as I’ve heard that a lot of games today use them… Not quite sure how im going to test it all on a 6150LE but ohh well :stuck_out_tongue:

For deferred shading shadow maps are the preferred way because you don’t have to render the geometry an extra time (besides for building the shadow maps themselves, but stencil shadows have the same problem), also there are a lot of tricks you can do to optimize them towards deferred shading.
With stencil shadows you still have to render the whole scene for every light pass, which kinda defeats the entire purpose of it.

You should be able to test CSM/PSSM on that hardware, though maybe without all the optimization possibilities the latest hw will buy you.

I was looking at CSM’s (btw is there anything different between CMS’s and PSSM’s? they seem the same )
I’ve also heard it is possible to combine VSM’s and PSSM’s to create something called PSVSM. I might look into that. Speed is not really a issue yet, The code base im working on will be made for 3.0 / 1.3 hardware once I get it, then I can worry about performance… sadly That requires me to build a whole new computer with no money…

The overall idea with CSM and PSSM seem roughly the same, but they may differ in some details, say in how/where the splits are made.

Just read a good writeup on CSMs over on Nvidia’s OpenGL SDK samples page. There’s a good white paper on PSSMs floating about too, but I don’t have a link handy.

You should be able to combine any of the soft shadow or map warping techniques with the split frustum stuff. Think of each slice as a separate render: anything you can do in a single render you can do in multiple renders. The area where the split hits the fan is at slice boundaries, where it’s usually necessary to introduce a slight overlap to fill the gaps that crop up from time to time.

Thanks, Ill check out nvidia’s CSM stuff, just read the gpu gems 3 article on PSSM’s…
Thanks for the help

Yep, forgot that was in the Gems (same authors).

This looks to be an interesting combination of the frustum split and map warping.


This topic was automatically closed 183 days after the last reply. New replies are no longer allowed.