Opengl Depth Value Transformation

出自 http://www.opengl.org/archives/resources/faq/technical/depthbuffer.htm 

老是忘记。。于是干脆直接粘到这里方便自己找。

Why is my depth buffer precision so poor?

The depth buffer precision in eye coordinates is strongly affected by the ratio of zFar to zNear, the zFar clipping plane, and how far an object is from the zNear clipping plane.

You need to do whatever you can to push the zNear clipping plane out and pull the zFar plane in as much as possible.

To be more specific, consider the transformation of depth from eye coordinates

x_e, y_e, z_e, w_e

to window coordinates

x_w, y_w, z_w

with a perspective projection matrix specified by

glFrustum(l, r, b, t, n, f);

and assume the default viewport transform. The clip coordinates of z_c and w_c are

z_c = -z_e* (f+n)/(f-n) - w_e* 2*f*n/(f-n)

w_c = -z_e

Why the negations? OpenGL wants to present to the programmer a right-handed coordinate system before projection and left-handed coordinate system after projection.

and the ndc coordinate:

z_ndc = z_c / w_c = [ -z_e * (f+n)/(f-n) - w_e * 2*f*n/(f-n) ] / -z_e

= (f+n)/(f-n) + (w_e / z_e) * 2*f*n/(f-n)

The viewport transformation scales and offsets by the depth range (Assume it to be [0, 1]) and then scales by s = (2ⁿ-1) where n is the bit depth of the depth buffer:

z_w = s * [ (w_e / z_e) * f*n/(f-n) + 0.5 * (f+n)/(f-n) + 0.5 ]

Let's rearrange this equation to express z_e / w_e as a function of z_w

z_e / w_e = f*n/(f-n) / ((z_w / s) - 0.5 * (f+n)/(f-n) - 0.5)

= f * n / ((z_w / s) * (f-n) - 0.5 * (f+n) - 0.5 * (f-n))

= f * n / ((z_w / s) * (f-n) - f) [*]

Now let's look at two points, the zNear clipping plane and the zFarclipping plane:

z_w = 0 => z_e / w_e = f * n / (-f) = -n

z_w = s => z_e / w_e = f * n / ((f-n) - f) = -f

In a fixed-point depth buffer, z_w is quantized to integers. The next representable z buffer depth away from the clip planes are 1 and s-1:

z_w = 1 => z_e / w_e = f * n / ((1/s) * (f-n) - f)

z_w = s-1 => z_e / w_e = f * n / (((s-1)/s) * (f-n) - f)

Now let's plug in some numbers, for example, n = 0.01, f = 1000 and s = 65535 (i.e., a 16-bit depth buffer)

z_w = 1 => z_e / w_e = -0.01000015

z_w = s-1 => z_e / w_e = -395.90054

Think about this last line. Everything at eye coordinate depths from -395.9 to -1000 has to map into either 65534 or 65535 in the z buffer. Almost two thirds of the distance between the zNear andzFar clipping planes will have one of two z-buffer values!

To further analyze the z-buffer resolution, let's take the derivative of [*] with respect to z_w

d (z_e / w_e) / d z_w = - f * n * (f-n) * (1/s) / ((z_w / s) * (f-n) - f)²

Now evaluate it at z_w = s

d (z_e / w_e) / d z_w = - f * (f-n) * (1/s) / n

= - f * (f/n-1) / s [**]

If you want your depth buffer to be useful near the zFar clipping plane, you need to keep this value to less than the size of your objects in eye space (for most practical uses, world space).