per fragment lighting and window to camera transforms

Question

I implemented a per fragment lighting and light attenuation using a window to camera transforms:

vec4 ndcPos;
ndcPos.xy = ((gl_FragCoord.xy / windowSize.xy) * 2.0) - 1.0;
ndcPos.z = (2.0 * gl_FragCoord.z - gl_DepthRange.near - gl_DepthRange.far) / (gl_DepthRange.far - gl_DepthRange.near);
ndcPos.w = 1.0;

vec4 clipPos = ndcPos / gl_FragCoord.w;

vec3 cameraPos = vec3(clipToCameraMatrix * clipPos);

Then I use lightPos and cameraPos to calculate lighting. lightPos has a fixed position in world space. I have ViewMatrix to handle a fps-style camera and ModelMatrix. Earlier I was using vertex position instead of window to camera transforms and everything was ok. Now I get a incorrect lighting when I move around objects using a camera. Before uploading to fragment shader lightPos is multiplied by ViewMatrix.

glUniform3fv(dirToLightUniformLocation, 1, glm::value_ptr(ViewMatrix * lightPos));

I have no idea what I'm doing wrong.

Your code looks correct, so the bug is in code that you're not posting. — Nicol Bolas

datenwolf datenwolf · Accepted Answer · 2012-05-08T08:07:07

You have one problem here:

ndcPos.z = (2.0 * gl_FragCoord.z - gl_DepthRange.near - gl_DepthRange.far) / (gl_DepthRange.far - gl_DepthRange.near);

unless you're using a orthographic projection gl_FragCoord.z does not map linear to the depth range, wich is caused by the w-division. Although it is possible (and not difficult) to back-project it, I suggest you just pass the viewspace eye position from the vertex shader to the fragment shader using an additional varying in/out.

per fragment lighting and window to camera transforms

2 Answers