DirectX10，实例渲染，顶点缓冲区输入

所以，我试图在DirectX10中使用简单着色器进行硬件实例化，其中顶点结构为顶点和实例的顶点和模型矩阵提供位置，纹理坐标和法线。问题在于创建将数据发送到着色器中的模型矩阵的输入布局。用PIX进行调试表明，该矩阵只能得到无意义的值。DirectX10，实例渲染，顶点缓冲区输入

相关代码：

输入布局

m_Tech = m_Effect->GetTechniqueByName("SimpleTech"); 

//Position-normal-texture vertex 
D3D10_INPUT_ELEMENT_DESC posNormalTexVertexDesc[] = 
{ 
    {"POSITION", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, 0,       D3D10_INPUT_PER_VERTEX_DATA, 0}, 
    {"NORMAL",  0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, D3D10_APPEND_ALIGNED_ELEMENT, D3D10_INPUT_PER_VERTEX_DATA, 0}, 
    {"TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT,  0, D3D10_APPEND_ALIGNED_ELEMENT, D3D10_INPUT_PER_VERTEX_DATA, 0}, 
    {"mTransform", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, 0,       D3D10_INPUT_PER_INSTANCE_DATA, 1}, 
    {"mTransform", 1, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D10_APPEND_ALIGNED_ELEMENT, D3D10_INPUT_PER_INSTANCE_DATA, 1}, 
    {"mTransform", 2, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D10_APPEND_ALIGNED_ELEMENT, D3D10_INPUT_PER_INSTANCE_DATA, 1}, 
    {"mTransform", 3, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D10_APPEND_ALIGNED_ELEMENT, D3D10_INPUT_PER_INSTANCE_DATA, 1} 
}; 
D3D10_PASS_DESC PassDesc; 

m_Effect->GetTechniqueByName("SimpleTech")->GetPassByIndex(0)->GetDesc(&PassDesc); 
HR(mp_D3DDevice->CreateInputLayout(posNormalTexVertexDesc, 7, PassDesc.pIAInputSignature, 
    PassDesc.IAInputSignatureSize, &m_InputLayout));

的Shader结构

struct VS_IN 
{ 
    float4 Pos : POSITION; 
    float2 TexCoord : TEXCOORD0; 
    float4 Normal : NORMAL; 
    row_major float4x4 mTransform : mTransform; 
};

缓冲器

void Model::BuildVertexBuffer() 
{ 
    D3D10_BUFFER_DESC vbd; 
    vbd.Usage = D3D10_USAGE_DEFAULT; 
    vbd.ByteWidth = sizeof(MeshVertex) * (UINT)m_Mesh->VertexCount(); 
    vbd.BindFlags = D3D10_BIND_VERTEX_BUFFER; 
    vbd.CPUAccessFlags = 0; 
    vbd.MiscFlags = 0; 
    D3D10_SUBRESOURCE_DATA vinitData; 
    vinitData.pSysMem = m_Mesh->VertexData(); 
    m_D3DDevice->CreateBuffer(&vbd, &vinitData, &m_VertexBuffer); 
} 

void Model::BuildInstanceBuffer() 
{ 
    D3D10_BUFFER_DESC ibd; 
    ibd.Usage = D3D10_USAGE_DEFAULT; 
    ibd.ByteWidth = sizeof(D3DXMATRIX) * (UINT)m_instanceData->size(); 
    ibd.BindFlags = D3D10_BIND_VERTEX_BUFFER; 
    ibd.CPUAccessFlags = 0; 
    ibd.MiscFlags = 0; 
    D3D10_SUBRESOURCE_DATA vinitData; 
    vinitData.pSysMem = &m_instanceData[0]; 
    m_D3DDevice->CreateBuffer(&ibd, &vinitData, &m_InstanceBuffer); 
}

绘制调用

void Model::Draw(Camera& camera) 
{ 
    UINT strides[2]; 
    UINT offsets[2]; 
    ID3D10Buffer* bufferPointers[2]; 

    //Set buffer strides 
    strides[0] = sizeof(MeshVertex); 
    strides[1] = sizeof(D3DXMATRIX); 

    //Set buffer offsets 
    offsets[0] = 0; 
    offsets[1] = 0; 

    //Set the array of pointers to the vertex and instance buffers 
    bufferPointers[0] = m_VertexBuffer; 
    bufferPointers[1] = m_InstanceBuffer; 

    //Set the vertex and instance buffers to active in the input assembler so it may be rendered 
    m_D3DDevice->IASetVertexBuffers(0,2,bufferPointers, strides, offsets); 
    m_D3DDevice->IASetPrimitiveTopology(D3D10_PRIMITIVE_TOPOLOGY_TRIANGLELIST); 

    //Set input layout 
    m_D3DDevice->IASetInputLayout(m_Effect->GetLayout()); 

    //Set 
    m_Effect->SetResources(camera, NULL); 

    D3D10_TECHNIQUE_DESC techDesc; 
    ID3D10EffectTechnique* tech = m_Effect->GetTech(); 
    tech->GetDesc(&techDesc); 

    for(UINT i = 0; i < techDesc.Passes; ++i) 
    { 
     ID3D10EffectPass* pass = tech->GetPassByIndex(i); 
     pass->Apply(0); 

     m_D3DDevice->DrawInstanced(m_Mesh->VertexCount(), m_instanceData->size(),0,0); 
    } 
}

m_instanceData是为每个实例保存D3DXMATRIX的向量。看过SDK附带的例子，我们的输入布局似乎看起来是一样的，但矩阵没有正确传输？

我目前只是发送一个标识矩阵作为测试用途的模型矩阵，但它显示为完全不同的东西。我已经检查过，确保一旦我初始化实例缓冲区后，数据就在向量中。

EDIT1：改变了BuildInstanceBuffer（）方法来

void Model::BuildInstanceBuffer() 
{ 
    D3D10_BUFFER_DESC ibd; 
    ibd.Usage = D3D10_USAGE_DYNAMIC; 
    ibd.ByteWidth = sizeof(D3DXMATRIX) * (UINT)m_instanceData->size(); 
    ibd.BindFlags = D3D10_BIND_VERTEX_BUFFER; 
    ibd.CPUAccessFlags = D3D10_CPU_ACCESS_WRITE; 
    ibd.MiscFlags = 0; 
    D3D10_SUBRESOURCE_DATA vinitData; 
    vinitData.pSysMem = &m_instanceData[0]; 
    m_D3DDevice->CreateBuffer(&ibd, &vinitData, &m_InstanceBuffer); 

    D3DXMATRIX* pMatrices = NULL; 
    m_InstanceBuffer->Map(D3D10_MAP_WRITE_DISCARD, NULL, (void**)&pMatrices); 

    memcpy(pMatrices, m_InstanceBuffer, m_instanceData->size() * sizeof(D3DXMATRIX)); 

    m_InstanceBuffer->Unmap(); 

}

第二编辑：想这是一个比特的预成熟。似乎在着色器中我的模型矩阵正在被“正确”填充，因为每一行都在正确的位置得到了一个值。但是，这个价值仍然不是我通过的价值。

这真诚地在一分钟内变得陌生。当我在Visual Studio中调试程序时，我会看到绿色的网格。在PIX中，它显示为黑色。我真的可以使用这个帮助，我不知道现在发生了什么。

来源

2012-02-02 Mr Dudeface

那么什么值在GPU上的实例缓冲区（使用PIX检查它）？另外，网格是否正确渲染？你说它被渲染成绿色，而PIX是黑色的？你可以张贴一些截图吗？ – 2012-02-03 12:48:20

更晚的时候，我再也猜不到，万一有人在这个问题上发愁。更改BuildInstanceBuffer到：

void Model::BuildInstanceBuffer() 
{ 
    D3D10_BUFFER_DESC ibd; 
    ibd.Usage = D3D10_USAGE_DYNAMIC; 
    ibd.ByteWidth = sizeof(D3DXMATRIX) * (UINT)m_instanceData->size(); 
    ibd.BindFlags = D3D10_BIND_VERTEX_BUFFER; 
    ibd.CPUAccessFlags = D3D10_CPU_ACCESS_WRITE; 
    ibd.MiscFlags = 0; 

    m_D3DDevice->CreateBuffer(&ibd, NULL, &m_InstanceBuffer); 
    D3DXMATRIX* pMatrices = NULL; 
    m_InstanceBuffer->Map(D3D10_MAP_WRITE_DISCARD, NULL, (void**)&pMatrices); 
    memcpy(pMatrices, m_instanceData->data(), m_instanceData->size() * sizeof(D3DXMATRIX)); 
    m_InstanceBuffer->Unmap(); 
}

基本上，从那里一块蛋糕。

来源

2012-03-01 17:11:47

DirectX10，实例渲染，顶点缓冲区输入

回答

相关问题