QRhiVertexInputAttribute Class

Detailed Description

Describes a single vertex input element.

The members specify the binding number, location, format, and offset for a single vertex input element.

Note: For HLSL it is assumed that the vertex shader uses TEXCOORD<location> as the semantic for each input. Hence no separate semantic name and index.

As an example, assume a vertex shader with the following inputs:

  layout(location = 0) in vec4 position;
  layout(location = 1) in vec2 texcoord;

Now let's assume that we have 3 component vertex positions (x, y, z) and 2 component texture coordinates (u, v) are provided in a non-interleaved format in a buffer (or separate buffers even). Once two bindings are defined, the attributes could be specified as:

  QRhiVertexInputLayout inputLayout;
  inputLayout.setBindings({
      { 3 * sizeof(float) },
      { 2 * sizeof(float) }
  });
  inputLayout.setAttributes({
      { 0, 0, QRhiVertexInputAttribute::Float3, 0 },
      { 1, 1, QRhiVertexInputAttribute::Float2, 0 }
  });

Once a graphics pipeline with this vertex input layout is bound, the vertex inputs could be set up like the following for drawing a cube with 36 vertices, assuming we have a single buffer with first the positions and then the texture coordinates:

  const QRhiCommandBuffer::VertexInput vbufBindings[] = {
      { cubeBuf, 0 },
      { cubeBuf, 36 * 3 * sizeof(float) }
  };
  cb->setVertexInput(0, 2, vbufBindings);

When working with interleaved data, there will typically be just one binding, with multiple attributes referring to that same buffer binding point:

  QRhiVertexInputLayout inputLayout;
  inputLayout.setBindings({
      { 5 * sizeof(float) }
  });
  inputLayout.setAttributes({
      { 0, 0, QRhiVertexInputAttribute::Float3, 0 },
      { 0, 1, QRhiVertexInputAttribute::Float2, 3 * sizeof(float) }
  });

and then:

  const QRhiCommandBuffer::VertexInput vbufBinding(interleavedCubeBuf, 0);
  cb->setVertexInput(0, 1, &vbufBinding);