为什么这个图像不是水平渐变？

这个问题不是我的问题，而是更多为什么我们会得到这个结果。该图像是使用以下着色器渲染两个立方体而生成的：

片段：

uniform mat4 u_ModelMatrix;
uniform mat4 u_ViewMatrix;
uniform mat4 u_ProjectionMatrix;

in FS {
    vec4 pos;
} fs;

out vec4 out_Color;

void main()
{
    out_Color.rgb = vec3(u_ProjectionMatrix * u_ViewMatrix * fs.pos).xxx;
}

顶点：

layout (location = 0) in vec4 in_Pos;

uniform mat4 u_ModelMatrix;
uniform mat4 u_ViewMatrix;
uniform mat4 u_ProjectionMatrix;

out FS {
    vec4 pos;
} fs;

void main()
{
    fs.pos = u_ModelMatrix * in_Pos;
    gl_Position = u_ProjectionMatrix * u_ViewMatrix * worldPos;
}

My understand is that as we get the projected fragment coordinates by multiplying by MVP, we should have vec3(u_ProjectionMatrix * u_ViewMatrix * fs.pos) equals to the window screen in X and Y and equals to the depth in Z. So, using xxx as rgb should show an horizontal gradient where models are drawn, with full black at center of the window (as right are negative X) and full white to the right. This is globally the result, but the color does not only depend on X, as we can see that there is vertical lines in the images that have pixels of different colors. Why does it happen ?