http://blog.csdn.net/cubesky/article/details/38682975

前面发了一篇关于unity Matrix的文章。

http://blog.csdn.NET/cubesky/article/details/38664143

其中对于一般的Matrix可以说应该有一个清晰的了解了。但是对于UNITY_MATRIX_IT_MV这些matrix估计理解起来还是比较有问题。这里再重点描述一下UNITY_MATRIX_IT_MV。

首先,我们看一下unity中Doc中的描述:

UNITY_MATRIX_IT_MV float4x4 Inverse transpose of model * view matrix.

然后我们来看一下UNITY_MATRIX_IT_MV实际的变换意义

The transpose of World2Object is the transpose of the inverse of the Object2World matrix.

  • MV transforms points from object to eye space
  • IT_MV rotates normals from object to eye space

And similarly:

  • Object2World transforms points from object to world space
  • IT_Object2World (which, as you point out, is the transpose of World2Object) rotates normals from object to world space

If it is orthogonal, the upper-left 3x3 of Object2World will be equal to that of IT_Object2World, and so will also rotate normals from object to world space.

上面这里很好的描述了UNITY_MATRIX_IT_MV的使用场景,专门针对法线进行变换。但是为什么法线的变换和定点不一样呢?让我们来看一篇推导的文章。

The gl_NormalMatrix is present in many vertex shaders. In here some light is shed on what is this matrix and what is it for. This section was inspired by the excellent book by Eric Lengyel “Mathematics for 3D Game Programming and Computer Graphics”.

Many computations are done in eye space. This has to do with the fact that lighting is commonly performed in this space, otherwise eye position dependent effects, such as specular lights would be harder to implement.

Hence we need a way to transform the normal into eye space. To transform a vertex to eye space we can write:

	vertexEyeSpace = gl_ModelViewMatrix * gl_Vertex;

So why can’t we just do the same with a normal vector? A normal is a vector of 3 floats and the modelview matrix is 4×4. Secondly, since the normal is a vector, we only want to transform its orientation. The region of the modelview matrix that contains the orientation is the top left 3×3 submatrix. So why not multiply the normal by this submatrix?

This could be easily achieved with the following code:

	normalEyeSpace = vec3(gl_ModelViewMatrix * vec4(gl_Normal,0.0));

So, gl_NormalMatrix is just a shortcut to simplify code writing or to optimize it? No, not really. The above line of code will work in some circumstances but not all.

Lets have a look at a potential problem:

;

In the above figure we see a triangle, with a normal and a tangent vectors. The following figure shows what happens when the modelview matrix contains a non-uniform scale.

Note: if the scale was uniform, then the direction of the normal would have been preserved, The length would have been affected but this can be easily fixed with a normalization.

In the above figure the Modelview matrix was applied to all the vertices as well as to the normal and the result is clearly wrong: the transformed normal is no longer perpendicular to the surface.

We know that a vector can be expressed as the difference between two points. Considering the tangent vector, it can be computed as the difference between the two vertices of the triangle’s edge. If  and  are the vertices that define the edge we know that:

Considering that a vector can be written as a four component tuple with the last component set to zero, we can multiply both sides of the equality with the Modelview matrix

This results in

As  and  are the vertices of the transformed triangle,  remains tangent to the edge of the triangle. Hence, the Modelview preserves tangents, yet it does not preserve normals.

Considering the same approach used for vector T, we can find two points  and  such that

The main issue is that the a vector defined through the transformed points, , does not necessarily remain normal, as shown in the figures above. The normal vector is not defined as a difference between two points, as the tangent vector, it is defined as a vector which is perpendicular to a surface.

So now we know that we can’t apply the Modelview in all cases to transform the normal vector. The question is then, what matrix should we apply?

Consider a 3×3 matrix G, and lets see how this matrix could be computed to properly transform the normal vectors.

We know that, prior to the matrix transformation T.N = 0, since the vectors are by definition perpendicular. We also know that after the transformation N’.T’ must remain equal to zero, since they must remain perpendicular to each other. T can be multiplied safely by the upper left 3×3 submatrix of the modelview (T is a vector, hence the w component is zero), let’s call this submatrix M.

Let’s assume that the matrix G is the correct matrix to transform the normal vector. T. Hence the following equation:

The dot product can be transformed into a product of vectors, therefore:

Note that the transpose of the first vector must be considered since this is required to multiply the vectors. We also know that the transpose of a multiplication is the multiplication of the transposes, hence:

We started by stating that the dot product between N and T was zero, so if

then we have

Which is exactly what we want. So we can compute G based on M.

Therefore the correct matrix to transform the normal is the transpose of the inverse of the M matrix. OpenGL computes this for us in the gl_NormalMatrix.

In the beginning of this section it was stated that using the Modelview matrix would work in some cases. Whenever the 3×3 upper left submatrix of the Modelview is orthogonal we have:

This is because with an orthogonal matrix, the transpose is the same as the inverse. So what is an orthogonal matrix? An orthogonal matrix is a matrix where all columns/rows are unit length, and are mutually perpendicular. This implies that when two vectors are multiplied by such a matrix, the angle between them after transformation by an orthogonal matrix is the same as prior to that transformation. Simply put the transformation preserves the angle relation between vectors, hence transformed normals remain perpendicular to tangents! Furthermore it preserves the length of the vectors as well.

So when can we be sure that M is orthogonal? When we limit our geometric operations to rotations and translations, i.e. when in the OpenGL application we only use glRotate and glTranslate and not glScale. These operations guarantee that M is orthogonal. Note: gluLookAt also creates an orthogonal matrix!

注:之所以法线不能直接使用UNITY_MATRIX_MV进行变换,是因为法线是向量,具有方向,在进行空间变换的时候,如果发生非等比缩放,方向会发生偏移。为什么呢?拿上面的例子来说,我们可以简单的把法线和切线当成三角形的两条边,显然,三角形在空间变换的时候,不管是平移,还是旋转,或者是等比缩放,都不会变形,但是如果非等比缩放,就会发生拉伸。所以法线和切线的夹角也就会发生变化。(而切线在变换前后,方向总是正确的,所以法线方向就不正确了)。

参考:

http://www.lighthouse3d.com/tutorials/glsl-tutorial/the-normal-matrix/

http://forum.Unity3D.com/threads/_object2world-or-unity_matrix_it_mv.112446/

http://www.cnblogs.com/kesalin/archive/2012/12/06/3D_math.html

UNITY_MATRIX_IT_MV[Matrix]的更多相关文章

  1. UNITY_MATRIX_IT_MV[Matrix] (转载)

    转载 http://blog.csdn.net/cubesky/article/details/38682975 前面发了一篇关于unity Matrix的文章. http://blog.csdn.n ...

  2. 旋转矩阵(Rotation Matrix)的推导及其应用

    向量的平移,比较简单. 缩放也较为简单 矩阵如何进行计算呢?之前的文章中有简介一种方法,把行旋转一下,然后与右侧对应相乘.在谷歌图片搜索旋转矩阵时,看到这张动图,觉得表述的很清晰了. 稍微复杂一点的是 ...

  3. angular2系列教程(十一)路由嵌套、路由生命周期、matrix URL notation

    今天我们要讲的是ng2的路由的第二部分,包括路由嵌套.路由生命周期等知识点. 例子 例子仍然是上节课的例子:

  4. Pramp mock interview (4th practice): Matrix Spiral Print

    March 16, 2016 Problem statement:Given a 2D array (matrix) named M, print all items of M in a spiral ...

  5. Atitit Data Matrix dm码的原理与特点

    Atitit Data Matrix dm码的原理与特点 Datamatrix原名Datacode,由美国国际资料公司(International Data Matrix, 简称ID Matrix)于 ...

  6. Android笔记——Matrix

    转自:http://www.cnblogs.com/qiengo/archive/2012/06/30/2570874.html#translate Matrix的数学原理 在Android中,如果你 ...

  7. 通过Matrix进行二维图形仿射变换

    Affine Transformation是一种二维坐标到二维坐标之间的线性变换,保持二维图形的"平直性"和"平行性".仿射变换可以通过一系列的原子变换的复合来 ...

  8. [LeetCode] Kth Smallest Element in a Sorted Matrix 有序矩阵中第K小的元素

    Given a n x n matrix where each of the rows and columns are sorted in ascending order, find the kth ...

  9. [LeetCode] Longest Increasing Path in a Matrix 矩阵中的最长递增路径

    Given an integer matrix, find the length of the longest increasing path. From each cell, you can eit ...

随机推荐

  1. poj1845(二分快速求等比数列模M和)

    Sumdiv Time Limit: 1000MS   Memory Limit: 30000K Total Submissions: 17039   Accepted: 4280 Descripti ...

  2. 重载(overload)、覆盖(override)和隐藏(hide)

    写正题之前,先给出几个关键字的中英文对照,重载(overload),覆盖(override),隐藏(hide).在早期的C++书籍中,可能翻译的人不熟悉专业用语(也不能怪他们,他们不是搞计算机编程的, ...

  3. C - Common Subsequence

    C - Common Subsequence Time Limit:1000MS     Memory Limit:32768KB     64bit IO Format:%I64d & %I ...

  4. Python菜鸟之路:Python基础-内置函数补充

    常用内置函数及用法: 1. callable() def callable(i_e_, some_kind_of_function): # real signature unknown; restor ...

  5. [note]BSGS & exBSGS

    BSGS (感觉这东西还是要写一下) BSGS主要用于求解形如\(x^k=y\pmod p\)(注意这里p与x互质)这样的方程的最小正整数解的问题 设\(m=\lceil\sqrt p\rceil,k ...

  6. SpringMVC的第一个入门案例

    用户提交一个请求,服务器端处理器在接收到这个请求后,给出一条欢迎信息,在页面中显示. 第一步:导入jar包 在原有Springjar包基础上添加2个jar包 spring-webmvc-4.2.0.R ...

  7. springcloud zuul 使用zuulfilter 修改请求路径和响应头

    最近做项目有一个需求:一个网盘系统,文件存放在分布式文件系统中,之前的文件下载统一走的文件下载服务,现在需要在单文件下载的时候不需要走文件下载服务,而是直接访问文件系统上的路径,响应的时候修改响应头, ...

  8. QT5使用Webkti

    Qt 5.3 使用原来的QT4.8.4项目时QWebView .QWebFrame等类无法编译通过. 出现原因:QWebView .QWebFrame.QWebPage.QWebInspector等这 ...

  9. 4G U盘版64位bitcoin专用挖矿操作系统

    这个操作系统是基于linux的操作系统,采用的ubuntu平台打造,所有的软件都已经安装齐备,是一个bitcoin专用挖矿操作系统,是64位的,对于显卡数量基本上没有限制,前提是你的主板支持足够多的显 ...

  10. ACN经典例题1

    1.韩信点兵 描述相传韩信才智过人,从不直接清点自己军队的人数,只要让士兵先后以三人一排.五人一排.七人一排地变换队形,而他每次只掠一眼队伍的排尾就知道总人数了.输入3个非负整数a,b,c ,表示每种 ...