DirectX 10/11 – Basic Shader Reflection – Automatic Input Layout Creation

This is gonna be a very brief post on a pretty simple but powerful tool for any hobby programmers tinkering with Direct3D. One of the most annoying things, at least for me, was always having to manually create input layouts for each different vertex shader I used in my hobby projects. I have now restarted work on my hobby game engine and have upgraded the renderer to DX11 and in doing so ripped out the DirectX effects framework. I’m actually planning on writing a post on moving from DX10 to DX11 but thats for another time. Right now lets get back to input layouts, if you can remember the input layout basically describes the memory layout of the input to the vertex shader program ( i.e. the input HLSL struct). In all my previous tutorials, we had to describe this input layout by hand and then create it using the compiled vertex shader for validation. We used to do this as follows:

const D3D10_INPUT_ELEMENT_DESC vertexInputLayout[] =

const int vertexInputLayoutNumElements = sizeof(vertexInputLayout)/sizeof(vertexInputLayout[0]);

D3D10_PASS_DESC PassDesc;

pTechnique->GetPassByIndex( 0 )->GetDesc( &PassDesc );
if ( FAILED( pD3DDevice->CreateInputLayout( vertexInputLayout,
											&pVertexLayout ) ) )
	return fatalError("Could not create Input Layout!");

The above sample is still using the effects framework but that’s irrelevant. There is another problem with this approach with regards to flexibility, if you wish to swap shaders on the fly, they either all need to use the same vertex shader input layout or you need to somehow create all your input layouts in advance and then link them with the shaders you wish to load. Well, considering the fact that we use the compiled shader to actually validate the input layout then that means that the compiled shader has all the necessary information available for us to create the input layout. Basically we’re just going to reverse engineer the validation step to create an input layout based on the shader.

Now if we don’t use the effects framework then we have to manually compile our shaders. I’m not going to go into much detail regarding this as the info is readily available in the SDK tutorials. Once you have compiled your shader you have the compiled shader blob which we can then use to create the final vertex shader program as shown below:

if( FAILED( D3DX11CompileFromFile( pFilename, NULL, NULL, pFunctionName, "vs_4_0", shaderFlags, NULL, NULL, &pCompiledShaderBlob, &pErrorBlob, NULL ) ) )
	if ( pErrorBlob != NULL )
		PRINT_ERROR_STRING( pErrorBlob->GetBufferPointer() );
	return false;

HRESULT hr = pD3DDevice->CreateVertexShader( pCompiledShaderBlob->GetBufferPointer(), pCompiledShaderBlob->GetBufferSize(), NULL, &shader.pVertexShader);

We can very easily create our input layout using the following function:

HRESULT CreateInputLayoutDescFromVertexShaderSignature( ID3DBlob* pShaderBlob, ID3D11Device* pD3DDevice, ID3D11InputLayout** pInputLayout )
	// Reflect shader info
	ID3D11ShaderReflection* pVertexShaderReflection = NULL;
	if ( FAILED( D3DReflect( pShaderBlob->GetBufferPointer(), pShaderBlob->GetBufferSize(), IID_ID3D11ShaderReflection, (void**) &pVertexShaderReflection ) ) )
		return S_FALSE;

	// Get shader info
	D3D11_SHADER_DESC shaderDesc;
	pVertexShaderReflection->GetDesc( &shaderDesc );

	// Read input layout description from shader info
	std::vector<D3D11_INPUT_ELEMENT_DESC> inputLayoutDesc;
	for ( uint32 i=0; i< shaderDesc.InputParameters; i++ )
		pVertexShaderReflection->GetInputParameterDesc(i, &paramDesc );

		// fill out input element desc
		D3D11_INPUT_ELEMENT_DESC elementDesc;
		elementDesc.SemanticName = paramDesc.SemanticName;
		elementDesc.SemanticIndex = paramDesc.SemanticIndex;
		elementDesc.InputSlot = 0;
		elementDesc.AlignedByteOffset = D3D11_APPEND_ALIGNED_ELEMENT;
		elementDesc.InputSlotClass = D3D11_INPUT_PER_VERTEX_DATA;
		elementDesc.InstanceDataStepRate = 0;	

		// determine DXGI format
		if ( paramDesc.Mask == 1 )
			if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_UINT32 ) elementDesc.Format = DXGI_FORMAT_R32_UINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_SINT32 ) elementDesc.Format = DXGI_FORMAT_R32_SINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_FLOAT32 ) elementDesc.Format = DXGI_FORMAT_R32_FLOAT;
		else if ( paramDesc.Mask <= 3 )
			if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_UINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32_UINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_SINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32_SINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_FLOAT32 ) elementDesc.Format = DXGI_FORMAT_R32G32_FLOAT;
		else if ( paramDesc.Mask <= 7 )
			if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_UINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32_UINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_SINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32_SINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_FLOAT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32_FLOAT;
		else if ( paramDesc.Mask <= 15 )
			if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_UINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32A32_UINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_SINT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32A32_SINT;
			else if ( paramDesc.ComponentType == D3D_REGISTER_COMPONENT_FLOAT32 ) elementDesc.Format = DXGI_FORMAT_R32G32B32A32_FLOAT;

		//save element desc

	// Try to create Input Layout
	HRESULT hr = pD3DDevice->CreateInputLayout( &inputLayoutDesc[0], inputLayoutDesc.size(), pShaderBlob->GetBufferPointer(), pShaderBlob->GetBufferSize(), pInputLayout );

	//Free allocation shader reflection memory
	return hr;

What this function does is peek inside the compiled shader using a shader reflection. A shader reflection is simply an interface for reading all the shader details at runtime. We first reflect the shader information using the D3DReflect function and then get the description of the shader using the reflection interface. From this description we can see how many input elements the vertex shader takes and then for each one we can get it descriptions. Using this data we can fill out our inputlayoutdesc structure. The above function is very simple and is not robust to be able to handle any shader thrown at it. I quickly slapped it together just to get things up and running with the intention of extending it as needed in the future. I just figured its something, that at least to my knowledge, isn’t readily known or mentioned and I figured just pointing it out would be useful for any hobbyist programmers  :P

Oh btw, before I forget to actually use the shader reflection functions you need to include the D3DCompiler.h header and link against the D3Dcompiler.lib static library.

4 thoughts on “DirectX 10/11 – Basic Shader Reflection – Automatic Input Layout Creation

  1. This is helping me a lot, Thanks very much! But there is something I know that you don’t: If you set the
    D3D11_INPUT_ELEMENT_DESC::AlignedByteOffset field to D3D11_APPEND_ALIGNED_ELEMENT,
    then the offset is automatically computed and assigned for you, which saves having to keep track of the offset yourself.

  2. D3DReflect function (Windows)
    HRESULT D3DReflect(
    in LPCVOID pSrcData,
    in SIZE_T SrcDataSize,
    in REFIID pInterface,
    out void **ppReflector
    Note You can use this API to develop your Windows Store apps, but you can’t use it in apps that you submit to the Windows Store.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s