当前位置:网站首页>反走样/抗锯齿技术
反走样/抗锯齿技术
2022-06-12 22:20:00 【ZEKEGU1997】
走样的根本原因,是用离散的像素点无法正确表达连续的几何体,导致边缘的像素不平滑。
不过当像素越多分辨率越高时,走样现象就越不明显。所以反走样的第一种办法就是在渲染计算时使用更高的分辨率,这种方法被称为超采样抗锯齿(Super Sample Anti-aliasing, SSAA),但是这种方法显然性能开销很大。


如左图所示,通常判断一个点是否属于某个三角面片,是根据像素中心点是否被遮盖,这个中心点也被称为采样点。增加采样点的个数,并根据被覆盖的采样点个数对颜色进行平均,如右图所示,这就是MSAA方法。
MSAA要使用连续的几何信息,而在延迟渲染中GBuffer并没有连续的几何信息,所以通常MSAA不适用于延迟渲染
FXAA利用了图像处理算法,通过检测图像边缘并平滑的办法实现抗锯齿的效果。
TAA的原理和MSAA 大致相同,都是每个像素点有多个采样点。但是不同与 MSAA 的方式,TAA (Temporal Anti-Aliasing) 综合历史帧的数据来实现抗锯齿,这样会将每个像素点的多次采样均摊到多个帧中,相对的开销要小得多。
边栏推荐
- LNMP platform docking redis service
- Unity 常用3D数学计算
- 3.5 setup and teardown of test classes
- 【LeetCode】69. x 的平方根
- Kotlin collaboration process - flow
- [data analysis] data clustering and grouping based on kmeans, including Matlab source code
- Leetcode: the maximum number of building change requests that can be reached (if you see the amount of data, you should be mindless)
- 数据库每日一题---第10天:组合两个表
- [Jianzhi offer simple] Jianzhi offer 06 Print linked list from end to end
- February 27th
猜你喜欢

Ansible PlayBook et ansible roles (3)

【图像去噪】基于三边滤波器实现图像去噪附matlab代码

JVM foundation > G1 garbage collector

【数据分析】基于 kmeans实现数据聚类分组含Matlab源码

PE installation win10 system

Ansible playbook and ansible roles (III)

How to specify your webpage's language so Google Chrome doesn't offer to translate it

JVM foundation > CMS garbage collector

Flutter series part: detailed explanation of GridView layout commonly used in flutter

Su embedded training day13 - file IO
随机推荐
Design a MySQL table for message queue to store message data
JVM foundation - > what is STW?
Ansible-大总结(六)
China barcode decoder market trend report, technical innovation and market forecast
大学期间零基础如何开展编程学习
The kotlin coroutine -- coroutine context and exception propagation
C # reading table data in word
JVM Basics - > how to troubleshoot JVM problems in your project
JS fighting on...
Redis optimization
What are thread scheduler and timeslicing?
【LeetCode】53.最大子数组和
Qt Quick 3D学习:使用鼠标键盘控制节点位置和方向
Kotlin collaboration process - flow
One article to quickly understand whether there are security risks in your network
JVM foundation - > talk about class loader two parent delegation model
JVM Basics - > What are the JVM parameters?
JVM foundation > CMS garbage collector
LNMP platform docking redis service
A puzzle about + =