Direct3D11 and SharpDX - How to pass a model instance's world matrix as an input to a vertex shader

Question

Using Direct3D11, I'm trying to pass a matrix into my vertex shader from the instance buffer that is associated with a given model's vertices and I can't seem to construct my InputLayout without throwing an exception.

The shader looks like this:

cbuffer ConstantBuffer : register(b0)
{
    matrix World;
    matrix View;
    matrix Projection;
}

struct VIn
{
    float4 position: POSITION;
    matrix instance: INSTANCE;
    float4 color: COLOR;
};

struct VOut
{
    float4 position : SV_POSITION;
    float4 color : COLOR;
};

VOut VShader(VIn input)
{
    VOut output;
    output.position = mul(input.position, input.instance);
    output.position = mul(output.position, View);
    output.position = mul(output.position, Projection);
    output.color = input.color;
    return output;
}

The input layout looks like this:

var elements = new[]
{
    new InputElement("POSITION", 0, Format.R32G32B32_Float, 0, 0, InputClassification.PerVertexData, 0),
    new InputElement("INSTANCE", 0, Format.R32G32B32A32_Float, 0, 0, InputClassification.PerInstanceData, 1),
    new InputElement("COLOR", 0, Format.R32G32B32A32_Float, 12, 0)
};
InputLayout = new InputLayout(device, signature, elements);

The buffer initialization looks like this:

public ModelDeviceData(Model model, Device device)
{
    Model = model;
    var vertices = Helpers.CreateBuffer(device, BindFlags.VertexBuffer, model.Vertices);
    var instances = Helpers.CreateBuffer(device, BindFlags.VertexBuffer, Model.Instances.Select(m => m.WorldMatrix).ToArray());
    VerticesBufferBinding = new VertexBufferBinding(vertices, Utilities.SizeOf<ColoredVertex>(), 0);
    InstancesBufferBinding = new VertexBufferBinding(instances, Utilities.SizeOf<Matrix>(), 0);
    IndicesBuffer = Helpers.CreateBuffer(device, BindFlags.IndexBuffer, model.Triangles);
}

The buffer creation helper method looks like this:

public static Buffer CreateBuffer<T>(Device device, BindFlags bindFlags, params T[] items)
    where T : struct
{
    var len = Utilities.SizeOf(items);
    var stream = new DataStream(len, true, true);
    foreach (var item in items)
        stream.Write(item);
    stream.Position = 0;
    var buffer = new Buffer(device, stream, len, ResourceUsage.Default,
        bindFlags, CpuAccessFlags.None, ResourceOptionFlags.None, 0);
    return buffer;
}

The line that instantiates the InputLayout object throws this exception:

*HRESULT: [0x80070057], Module: [General], ApiCode: [E_INVALIDARG/Invalid Arguments], Message: The parameter is incorrect.*

Note that the data for each model instance is simply an instance of SharpDX.Matrix.

EDIT

Based on Tordin's answer, it sems like I have to modify my code like so:

var elements = new[]
{
new InputElement("POSITION", 0, Format.R32G32B32_Float, 0, 0, InputClassification.PerVertexData, 0),
new InputElement("COLOR", 0, Format.R32G32B32A32_Float, 12, 0),
new InputElement("INSTANCE0", 0, Format.R32G32B32A32_Float, 0, 1, InputClassification.PerInstanceData, 1),
new InputElement("INSTANCE1", 1, Format.R32G32B32A32_Float, 16, 1, InputClassification.PerInstanceData, 1),
new InputElement("INSTANCE2", 2, Format.R32G32B32A32_Float, 32, 1, InputClassification.PerInstanceData, 1),
new InputElement("INSTANCE3", 3, Format.R32G32B32A32_Float, 48, 1, InputClassification.PerInstanceData, 1),
};

and in the shader:

struct VIn
{
    float3 position: POSITION;
    float4 color: COLOR;
    matrix instance: INSTANCE;
};

VOut VShader(VIn input)
{
    VOut output;
    output.position = mul(input.position, input.instance);
    output.position = mul(output.position, View);
    output.position = mul(output.position, Projection);
    output.color = input.color;
    return output;
}

However I still get an exception.

Answer 1

in your input layout, you are only specifying one part of the matrix. you have to specif the other three. You are saying that you have a VECTOR, not a matrix. to build a matrix, you must have 4 elements for a matrix, and saying that they belong into a specific register.

Reading this link, says on the second argument "a 4x4 matrix would have four components each with the semantic name"
MSDN LINK INPUTLAYOUTS

Answer 2

This may be a bit late but if anyone else would come across this topic, this answer is supposed to be a more extensive answer to the question as I believe some things are not properly pointed out.

First off, as clarified by a comment above the input layout has to use a predefined sematic name, in this case we could use:

{ "TEXCOORD", 1, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, 0, D3D11_INPUT_PER_INSTANCE_DATA, 1 },
{ "TEXCOORD", 2, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_INSTANCE_DATA, 1 },
{ "TEXCOORD", 3, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_INSTANCE_DATA, 1 },
{ "TEXCOORD", 4, DXGI_FORMAT_R32G32B32A32_FLOAT, 1, D3D11_APPEND_ALIGNED_ELEMENT, D3D11_INPUT_PER_INSTANCE_DATA, 1 },

Although this will only work if your instance vertex buffer is bound at slot 1. Also note that we start with the sematic name "TEXCOORD1" as most applications already have "TEXCOORD" bound to the actual texture-coordinates for the object.

Now in the vertex shader we will have to provide an appropriate input struct which should match the input layout we have defined, for example:

struct VS_IN
{
    float3 Pos : POSITION;
    float2 UV : TEXCOORD;
    float3 Normal : NORMAL;
    float3 Tangent : TANGENT;
    float4 InstancePos0 : TEXCOORD1;
    float4 InstancePos1 : TEXCOORD2;
    float4 InstancePos2 : TEXCOORD3;
    float4 InstancePos3 : TEXCOORD4;
};

Then we simply have to construct a 4x4 matrix from the instanced data we have provided:

float4x4 world = float4x4(input.InstancePos0, input.InstancePos1, input.InstancePos2, input.InstancePos3);