D3D9 effect (hlsl)(转)

 

转:http://blog.csdn.net/leonwei/article/details/8212800

effect其实整合了shader和render state的控制两大部分内容

9.1 effect文件基本框架

part1 :shader state包括全局变量 shader数据结构定义,shader的实现

D3D9 effect (hlsl)(转)

part2 :texture and sampler state,这个通常是全局变量,用于shader中,通常要定义在shader的实现之前

D3D9 effect (hlsl)(转)

part3 其他的render state设置,写在technique的pass里,这才是effect文件的主入口

D3D9 effect (hlsl)(转)

effet中任何一个变量后面都可以接一个语意:XXX

1.可以使用的变量类型

Data : Buffer Scalar Vector Matrix Sample Shader Texture Struct UserDefined

9.1.1参数类型

hlsl中使用的参数有两种:Uniform和 varying

Uniform存储在常量缓存中的,是对shader的每一次调用值都是一样的,可以在D3D中访问的 前面可以加uniform(可不加),通常是全局的

varying:只对一次调用起作用的,是来自vertext stream中的,通常要有语意的修饰,通产是写在shader的输入输出参数里的

9.1.2 sampler纹理采样器

要包含三个信息

  • A texture :纹理物体(tex0 tex1 tex2...NULL)
  • A sampler (with sampler state)如
  • A sampling instruction

例如定义这样一个sampler

 
sampler s = sampler_state 
{ 
  texture = NULL; 
  mipfilter = LINEAR; 
};
 
就可以在shader中使用
float2 sample_2D(float2 tex : TEXCOORD0) : COLOR
{
  return tex2D(s, tex);
}
来采样纹理
9.1.3 shader的函数体

函数通常要包含这样几部分:

  • A return type
  • A function name
  • An argument list (optional)
  • An output semantic (optional)
  • An annotation (optional)
float4 VertexShader_Tutorial_1(float4 inPos : POSITION ) : POSITION
{
   ...
}
struct VS_OUTPUT
{
    float4  vPosition        : POSITION;
    float4  vDiffuse         : COLOR;
}; 
 
VS_OUTPUT VertexShader_Tutorial_1(float4 inPos : POSITION )
{
   ...
}

9.2D3D9中使用effect

1.首先从effect file创建effect(编译),使用函数

D3DXCreateEffectFromFile( 
        pd3dDevice, 
        str, 
        NULL, // CONST D3DXMACRO* pDefines,
        NULL, // LPD3DXINCLUDE pInclude,
        dwShaderFlags, 
               NULL, // LPD3DXEFFECTPOOL pPool,
        &g_pEffect, 
        NULL );
 

dwShaderFlags

    • Skipping validation, 如果已知这是没有问题的 那么就跳过检查
    • Skipping optimization 有时开启优化会导致debug 困难,就可以跳过优化
    • Requesting debug information to be included in the shader so it can be debugged.

这个函数的调用实际上是在从文件编译出一个effect

你还可以利用工具fxc.exe离线的编译出一个effect,他会直接编出一个汇编的shader指令,这样在程序中就不用编译了。发布产品时这样做应该更好。

2.调用一个effect来render

ID3DXEffect::Begin set technique

ID3DXEffect::BeginPass set pass

ID3DXEffect::CommitChanges提交对当前pass的各种参数修改,要在draw之前调用

ID3DXEffect::EndPass

ID3DXEffect::End

一个典型调用

D3D9 effect (hlsl)(转)

3  effect 的参数设置(指effect文件中定义的全局变量(uniform))

设置与获取

m_pEffect->SetValue

m_pEffect->GetValue

函数的第一个参数都是变量的名字

 
不同的effect 可以共享参数:如果一个effect变量定义为shared,并且多个effect创建的时候使用同一个pool(D3DXCreateEffectFromFile的第六个参数)就可以共享这个参数。
 
创建参数block,可以像使用render state的block那样 创建parament state,这时是通过创建Parameter Blocks的方式。把effect 的参数设置操作写在
m_pEffect->BeginParameterBlock();与m_pEffect->EndParameterBlock();
之间。得到的block可以用在effect 上

4 用户注释

用户可以在effect pass 或者某个变量上附加一个用户的注释。他其实就是一个string或numeric格式的变量。格式

 
texture Tex0 < string name = "tiger.bmp"; >;意味着给tex0 附加一个注释,告诉这个texture使用的纹理文件的路径。注释可以理解为用户在D3D层面去扩充effect 文件的定义。

ID3DXBaseEffect::GetAnnotation orID3DXBaseEffect::GetAnnotationByName可以获取

5 preshader技术

对于shader中的常量计算,D3D会将其提取出来,在shader执行之前,先由CPU计算,这样会提高GPU的效率,防止每个顶点或pixcel都重复这个同样的运算,使不使用preshader可一在编译时指定优化的flag,关闭使用参数D3DXSHADER_NO_PRESHADER。所以基于这个技术我们可以在shader中做常量矩阵运算,不用担心这个运算会在每个顶点处做一次。

9.3 语法备忘

9.3.1 语义

VS:

输入:

BINORMAL[n]

Binormal

float4

BLENDINDICES[n]

Blend indices

uint

BLENDWEIGHT[n]

Blend weights

float

COLOR[n]

Diffuse and specular color

float4

NORMAL[n]

Normal vector

float4

POSITION[n]

Vertex position in object space.

float4

POSITIONT

Transformed vertex position.

float4

PSIZE[n]

Point size

float

TANGENT[n]

Tangent

float4

TEXCOORD[n]

输出

COLOR[n]

Diffuse or specular color

float4

FOG

Vertex fog

float

POSITION[n]

Position of a vertex in homogenous space. Compute position in screen-space by dividing (x,y,z) by w. Every vertex shader must write out a parameter with this semantic.

float4

PSIZE

Point size

float

TESSFACTOR[n]

Tessellation factor

float

TEXCOORD[n]

Texture coordinates

float4

PS

输入:

COLOR[n]

Diffuse or specular color.

float4

TEXCOORD[n]

Texture coordinates

float4

VFACE

Floating-point scalar that indicates a back-facing primitive. A negative value faces backwards, while a positive value faces the camera.

float

VPOS

The pixel location (x,y) in screen space. To convert a Direct3D 9 shader (that uses this semantic) to a Direct3D 10 shader, seeDirect3D 9 VPOS and Direct3D 10 SV_Position)

float2

输出:

COLOR[n]

Output color

float4

DEPTH[n]

Output depth

float

9.3.2数据的访问方式

vector

bool bVector = false;int1 iVector = 1;float3 fVector = { 0.2f, 0.3f, 0.4f };double4 dVector = { 0.2, 0.3, 0.4, 0.5 };

vector <bool, 1> bVector = false;vector <int, 1> iVector = 1;vector <float, 3> fVector = { 0.2f, 0.3f, 0.4f };vector <double, 4> dVector = { 0.2, 0.3, 0.4, 0.5 };


// Given
float4 pos = float4(0,0,2,1); pos.z // value is 2
pos.b // value is 2

// Given
float4 pos = float4(0,0,2,1);
float2 temp; temp = pos.xy // valid
temp = pos.rg // valid temp = pos.xg // NOT VALID because the position and color sets were used.

float4 pos = float4(0,0,2,1);
float4 f_4D;
f_4D = pos; // write four components f_4D.xz = pos.xz; // write two components
f_4D.zx = pos.xz; // change the write order f_4D.xzyw = pos.w; // write one component to more than one component
f_4D.wzyx = pos;

matrix

A matrix can be initialized when it is declared:


float2x2 fMatrix = { 0.0f, 0.1, // row 1
2.1f, 2.2f // row 2
};

Or, the matrix type can be used to make the same declarations:


matrix <float, 2, 2> fMatrix = { 0.0f, 0.1, // row 1
2.1f, 2.2f // row 2
};
  • The zero-based row-column position:
    • _m00, _m01, _m02, _m03
    • _m10, _m11, _m12, _m13
    • _m20, _m21, _m22, _m23
    • _m30, _m31, _m32, _m33
  • The one-based row-column position:
    • _11, _12, _13, _14
    • _21, _22, _23, _24
    • _31, _32, _33, _34
    • _41, _42, _43, _44
    • [0][0], [0][1], [0][2], [0][3]
    • [1][0], [1][1], [1][2], [1][3]
    • [2][0], [2][1], [2][2], [2][3]
    • [3][0], [3][1], [3][2], [3][3]

However, array accessing can read a multi-component vector.


float2 temp;
float2x2 fMatrix;
temp = fMatrix[0] // read the first row

Just like vectors, naming sets can use one or more components from either naming set.


// Given
float2x2 fMatrix = { 1.0f, 1.1f, // row 1
2.0f, 2.1f // row 2
};
float2 temp; temp = fMatrix._m00_m11 // valid
temp = fMatrix._m11_m00 // valid
temp = fMatrix._11_22 // valid
temp = fMatrix._22_11 // valid

As with vectors, reading more than one matrix component is called swizzling. More than one component can be assigned, assuming only one name space is used. These are all valid assignments:


// Given these variables
float4x4 worldMatrix = float4( {0,0,0,0}, {1,1,1,1}, {2,2,2,2}, {3,3,3,3} );
float4x4 tempMatrix; tempMatrix._m00_m11 = worldMatrix._m00_m11; // multiple components
tempMatrix._m00_m11 = worldMatrix.m13_m23; tempMatrix._11_22_33 = worldMatrix._11_22_33; // any order on swizzles
tempMatrix._11_22_33 = worldMatrix._24_23_22;
9.3.3 定义shader类型

PixelShader = compile ShaderTarget ShaderFunction(...);

VertexShader = compile ShaderTarget ShaderFunction(...);

ShaderTarget :shade_model 例如vs_2_0

...就是shader中的uniform参数

9.4编译shader

在VS(2012以下)中直接DEBUG是不能发现shader 的语法错误的,这需要使用fxc工具

常用语法是

fxc /T fx_2_0 file.fx

Will compile the entire file with vs and ps 2.0.

fxc /T ps_2_0 /E PixelShaderFunction file.fx

Will compile PixelShaderFunction with ps 2.0.

9.5调试

可以使用dx sdk中带的pix

上一篇:SNMP协议详解


下一篇:LIS(nlogn) POJ 3903 Stock Exchange