Custom WGSL position node in clip space (WebGPU)

Questions
, , , ,
1

I’m trying to implement a custom vertex shader in WGSL. I can do the following:

const params = {
	position: Nodes.attribute('position', 'vec3'),
};

const myPositionNode = Nodes.wgslFn(`
	fn getPosition(position: vec3<f32>) -> vec4<f32> {
		// Do my custom vertex shader calculations
		return vec4(...);
	}
`);

myNodeMaterial.positionNode = myPositionNode(params);

This works, but it applies the usual modelview and projection transformations on my output - after I return it, outside of my control. When inspecting the generated WGSL, this is what the main function looks like:

@vertex
fn main( @location( 0 ) position : vec3<f32> ) -> NodeVaryingsStruct {

	// system
	var NodeVaryings: NodeVaryingsStruct;

	// vars
	
	var nodeVar0 : vec4<f32>;
	var nodeVarying0 : vec3<f32>;
	var nodeVarying2 : vec3<f32>;


	// flow
	// code

	nodeVarying0 = position;
	nodeVarying0 = getPosition( position ).xyz;
	nodeVarying2 = vec3<f32>( 0.0, 0.0, 0.0 );
	NodeVaryings.nodeVarying1 = ( NodeUniforms.nodeUniform0 * nodeVarying2 );
	nodeVar0 = ( ( NodeUniforms.projectionMatrix * NodeUniforms.modelViewMatrix ) * vec4<f32>( nodeVarying0, 1.0 ) );

	// result
	NodeVaryings.Vertex = nodeVar0;

	return NodeVaryings;

}

So my question is, is there a way to avoid those transformations that are applied after my getPosition() function is called? Or is there a way to do this without a positionNode? For example, if I am working in clip space for some custom objects.

Also I’m still on r157, so I apologize if this has already had time to change in r158-r159, but I would be happy to know if that is the case. I was not able to find anything on this in the examples.