Sign in to follow this  

Problems with shadows in deferred renderer

Recommended Posts

I'm making a simple deferred (classic deferred) renderer in learning purpose. Most things work ok but I suddenly got stuck when trying to implement shadow mapping.

The idea is:
when rendering the lightvolume (e.g. a sphere for a point light) in order to calculate light contribution for that light I also read the shadow texture corresponding to that light. (that has been rendered just before the light pass).
I sample the shadow texture with a point sampled from the gbuffer that I transform with the shadow matrix.
Then I compare that sample points z value to the z-value of the sample position.

It should be simple enough but it just dont work for me so I have to ask here both if the general idea is right and if the actual code does what I intend.

The GBuffer:
This is a FBO with attachments for albedo, position etc. position is calculated in the attribute write pass like this:
vec3 vPosition = vec3(uModelViewMatrix * vec4(inPosition, 1.0));
vPosition is then directly written to the gbuffer in the frag shader.
so It is in eye (camera) space but without perspective transform.

The shadow map:
Mat4f shadowPerspective = Mat4f::Perspective(45.0f, 1.0, 0.1f, 500.0f);
the lightpos + lightdir is concatenated into lightViewMatrix
the shadow matrix is then:
Mat4f shadowMatrix = Mat4f::Multiply(shadowPerspective, lightViewMatrix);
By checking the rendered texture for the shadow I can tell that this works as expected. (I attached a color tex to see that the output looked like it should from the lights point of view)
By this I conclude that the shadowMatrix is correct.

In the light pass:
I first concat the inverse of the eye(camera) and the shadow matrix (I have verified that inverse works alright)
shadowMatrix = Mat4f::Multiply(inverseCameraMatrix, shadowMatrix);
this shadow matrix is then sent to the light shader (and called "uShadowMatrix" in the calculatins below)

The point light vertex shader:
I transform the light sphere to clip coord (camera space with perspective transform)
gl_Position = uProjectionMatrix * uModelViewMatrix * vec4(inPosition, 1.0);

in the light frag shader:
First I extract the eye position from the gbuffer (this is eye space without perspective transform)
vec2 samplePoint = vec2(gl_FragCoord.x/uBufferSize.x, gl_FragCoord.y/uBufferSize.y);
vec3 position = texture2D(uPositionTex, samplePoint).rgb;
Now I want to transform this position into the shadow space
vec4 shadowspacePosition = uShadowMatrix * vec4(position, 1.0);
and remove the perspective transform from the shadow matrix
vec3 shadowcoord = / shadowspacePosition.w;
I then sample the shadow texture to get if it is in shadow or not:
vec3 shadowSample = texture2D(uShadowTex, shadowcoord.xy ).rgb;
if((shadowSample.z) < shadowcoord.z) shadow = 0.0;

But this doesn't work. Have been looking at my code for more then one day now but I just can't seems to get it right at this time. Is there something very obviously wrong with my thinking here?
I would very much appreciate your help!

Best regards

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