作业7 简单抗锯齿出现毛边

[oxine]

原来的代码是这样的:

float x = (2 * (i + 0.5 ) / (float)scene.width - 1) * imageAspectRatio * scale;
float y = (1 - 2 * (j + 0.5) / (float)scene.height) * scale;

真的很简单，把下面代码中的0.5改成get_random_float()

float x = (2 * (i + get_random_float() ) / (float)scene.width - 1) * imageAspectRatio * scale;
float y = (1 - 2 * (j + get_random_float()) / (float)scene.height) * scale;

原来的代码是在像素的中心点采样的，现在我们在整个像素中随机采样，思路类似SSAA, 得益于spp的存在，都不用多加一层循环了。

现在锯齿是没有了，但是不可避免的出现毛边，一开始我还以是因为spp不够，拉到512，边依然是毛糙的。

我的EPSILON = 0.01，测试下来好像与epsilon无关。

This post has received 1 vote up.

Attachments:

You must be logged in to view attached files.

[oxine]

有条件的同学试着跑跑看吧，我想不明白了…

[arc]

直接修改的话会导致每个像素都取到某一个固定的位置上，产生毛边是正常的。
你需要在计算每一个样本的时候都计算一个随机的位置，这样才会产生平滑的结果。

This post has received 1 vote up.

• This reply was modified 4 years, 6 months ago by arc.

Attachments:

You must be logged in to view attached files.

[oxine]

对哦，有道理！我的xy在spp的循环内都是不变的，我太蠢了先前没想到，谢谢大佬解答！

• This reply was modified 4 years, 6 months ago by oxine.

[oxine]

修改起来很方便，把xy以及射线的生成挪到spp循环内部就行了，这样可以节省一次循环

for (int k = 0; k < spp; k++){
	//generate primary ray direction
	float x = (2 * (i + get_random_float()) / (float)scene.width - 1) * imageAspectRatio * scale;
	float y = (1 - 2 * (j + get_random_float()) / (float)scene.height) * scale;
	Vector3f dir = normalize(Vector3f(-x, y, 1));
	auto index = j*scene.width + i;
        framebuffer[index] += scene.castRay(Ray(eye_pos, dir), 0) / spp;
}

在低spp情况下也能获取不错的效果

Attachments:

You must be logged in to view attached files.

[Author]

Posts