// Get texel size
float2 texelSize = 1.0 / shadowMapSize;
// Offset the texcoords by half a texel
float2 offsetTex = texelSize / 2;
// Object depth in shadow space
float objDepth = min( vProjCoords.z, 0.99999 );
// Get four shadowmap samples at the corners
float4 tap;
// X O
// O O
tap.x = tex2Dproj( DepthSampler, float4( shadowMapCenter + float2( -offsetTex.x, offsetTex.y ), objDepth, 1 ) ).x;
// O Y
// O O
tap.y = tex2Dproj( DepthSampler, float4( shadowMapCenter + float2( offsetTex.x, offsetTex.y ), objDepth, 1 ) ).x;
// O O
// Z O
tap.z = tex2Dproj( DepthSampler, float4( shadowMapCenter + float2( -offsetTex.x, -offsetTex.y ), objDepth, 1 ) ).x;
// O O
// O W
tap.w = tex2Dproj( DepthSampler, float4( shadowMapCenter + float2( offsetTex.x, -offsetTex.y ), objDepth, 1 ) ).x;
// Get the fractional xy position of the fragment within the texel
float2 fpos = frac(vProjCoords.xy * shadowMapSize);
Stupid question, but have you tried inverting the order that the taps appear in the lerp intrinsic?
(i.e. float r1 = lerp(tap.y, tap.x, fpos.x) becomes float r1 = lerp(tap.x, tap.y, fpos.x) and so forth)
EDIT: Wow, that looks horrible. Hopefully you can get the gist of it, though.
clb: At the end of 2012, the positions of jupiter, saturn, mercury, and deimos are aligned so as to cause a denormalized flush-to-zero bug when computing earth's gravitational force, slinging it to the sun.
Try to set MagFilter = Point; / pID3DDevice->SetSamplerState(st, D3DSAMP_MAGFILTER, D3DTEXF_POINT); for your shadowmap texture
I ran into some similar visual arfifact with shadowmap & lerp when I was using LINEAR filtering on the texture.
David Tuft held at Gamesfest in 2010 a nice talk on shadow mapping, where he covered some basic problems and provided little code fragments to get pass them.
It's a one hour talk, which is definitely worth listening to, especially if you're just starting with shadow mapping. Slides and audio track can be found here at the MSDN download center.
Have fun with that. ;)
Try to set MagFilter = Point; / pID3DDevice->SetSamplerState(st, D3DSAMP_MAGFILTER, D3DTEXF_POINT); for your shadowmap texture
I ran into some similar visual arfifact with shadowmap & lerp when I was using LINEAR filtering on the texture.
exactly - also had a similar problem once - and point sampling did solve it.
I believe the issue is that you're looking at the wrong pixels for your taps. You're sampling the 4 pixels in the diagonal corners around the source pixel, like this:
(X represents source pixel, O represents sampled pixel)
O O
X
O O
When in fact the sampling should be like this:
XO
OO
(*note X is also part of the tap)
If you look carefully at the original code you linked, the offsets are different in yours.
[size="2"]Currently working on an open world survival RPG - For info check out my Development blog:[size="2"] ByteWrangler
