阴影贴图:将世界空间像素投影到光空间

我正在用延迟着色编写阴影映射.

I'm writing shadow mapping in deferred shading.

这是我的定向光深度图(正交投影):

Here is my depth map for directional light (orthogonal projection):

下面是我的全屏四边形着色器，用于在光视图空间中渲染像素的深度:

Below is my full-screen quad shader to render pixel's depth in light view space:

#version 330

in vec2 texCoord;
out vec3 fragColor;

uniform mat4 lightViewProjMat; // lightView * lightProj

uniform sampler2D sceneTexture;
uniform sampler2D shadowMapTexture;
uniform sampler2D scenePosTexture;

void main() {
    vec4 fragPos = texture(scenePosTexture, texCoord);
    vec4 fragPosLightSpace = lightViewProjMat * fragPos;

    vec3 coord = fragPosLightSpace.xyz / fragPosLightSpace.w;
    coord = coord * 0.5 + 0.5;

    float lightViewDepth = texture(shadowMapTexture, coord.xy).x;
    fragColor = vec3(lightViewDepth);
}

渲染尺寸为1280x720，深度图的尺寸为512x512，看起来像是在场景中重复了深度图.我认为我对协调的预测是不正确的.我想看看从像素到光的深度是否正确.有人可以给我一些建议吗?

Render size is 1280x720, depth map's size is 512x512 and look like it is repeating the depth map in my scene. I think my projection for coord isn't right. I want to see if the depth from pixel to light is correct. Could someone give me some suggestions?