Stormii
30-Oct-2008, 02:11
Best Deferred value storage practices in shader model at least 3.0 I have read starcraft2 paper that suggests to use VPOS to get rid of some division when getting world position form view...
I'm using 16-bit floating point render targets for everything, through it introduced slight frame rate drop (was on 8-bit before)
Now im trying to figure out stuff i don't actually know to well right now like best practices or some suggestions on storing depth/position and renewing position from stored depth, also i can get depth in ton of ways, currently its worldPosition.z / worldPosition.w; And that i guess gives me some non-linear, depth, ranging form 0.0 to 1.0? Any suggestions welcome! :oops:
I'm using 16-bit floating point render targets for everything, through it introduced slight frame rate drop (was on 8-bit before)
Now im trying to figure out stuff i don't actually know to well right now like best practices or some suggestions on storing depth/position and renewing position from stored depth, also i can get depth in ton of ways, currently its worldPosition.z / worldPosition.w; And that i guess gives me some non-linear, depth, ranging form 0.0 to 1.0? Any suggestions welcome! :oops: