|matrix||Transformation matrix to use.|
|material||Material to use.|
|shaderPass||Which pass of the shader to use (or -1 for all passes).|
|topology||Topology of the procedural geometry.|
|properties||Additional material properties to apply just before rendering. See MaterialPropertyBlock.|
|bufferWithArgs||Buffer with draw arguments.|
|argsOffset||Byte offset where in the buffer the draw arguments are.|
Add a "draw procedural geometry" command.
When the command buffer executes, this will do a draw call on the GPU, without any vertex or index buffers.
The amount of geometry to draw is read from a ComputeBuffer. Typical use case is generating arbitrary amount
of data from a ComputeShader and then rendering that, without requiring a readback to the CPU.
This is only useful on Shader Model 4.5 level hardware where shaders can read arbitrary data from ComputeBuffer buffers.
Buffer with arguments,
bufferWithArgs, has to have four integer numbers at given
vertex count per instance, instance count, start vertex location, start instance location.
This very much maps to Direct3D11 DrawInstancedIndirect and equivalent functions on other graphics APIs. On OpenGL versions before 4.2 and all OpenGL ES versions that support indirect draw, the last argument is reserved and therefore must be zero.
In the vertex shader, you'd typically use SV_VertexID and SV_InstanceID input variables to fetch data from some buffers.
See Also: DrawProcedural, MaterialPropertyBlock, Graphics.DrawProceduralIndirect, ComputeBuffer.CopyCount, SystemInfo.supportsComputeShaders.