Contents

cp_frame_binocular_frustum_matrix_for_drawable_target

Returns the transform which can be used for binocular frustum culling. A matrix to convert between the device coordinate space to normalized device coordinate space. This should be acquired after query frame via Cp_layer_renderer_query_next_frame and before calling into Endsubmission(). Renderer should not utilize this transform for actual rendering output.

Declaration

simd_float4x4 cp_frame_binocular_frustum_matrix_for_drawable_target(cp_frame_t frame, cp_drawable_target drawable_target, cp_axis_direction_convention convention, simd_float4 increase_tangents, simd_float2 depth_range);

Parameters

  • frame:

    Frame: The frame you finished preparing.

  • drawable_target:

    Whether this is intended for built_in or recorderdrawable

  • increase_tangents:

    The requested increased tangents for computing the projection transform. Values should be positive.

  • depth_range:

    The requested depth range for computing the projection transform. The range should be bigger than the one used for rendering

Return Value

A transform which combines perspective projection matrix that transforms coordinates from view coordinate space to clip space and a view matrix which brings device coordinate space to binocular camera space

Discussion

            /
           /
          / Monocular eye
         /------

Binocular / eye | \ ╱ \ ╱ \ ╱ \ Monocular eye ╲ ╲ ╲

The convention is defined in normalized device coordinates, which impact the rendering direction of rendering but have no impact on the convention used for model/view space. To convert from view space to the axis direction convention used by the projection matrix, a rotation/mirror matrix can be applied pre-projection matrix.

The axes are defined as: • X axes determines the horizontal direction of pixels – right convention means the left-most pixel is the left of the view • Y axes determines the vertical direction of pixels – up convention means the top-most pixel is the top of the view • Z axes determines the handedness/winding order The default convention is cp_axis_direction_convention_right_up_back for CompositorServices and Metal renderers which has a counter-clockwise winding order.

It computes corresponding binocular view relative to views, for defined transforms and projection matrix.

To avoid false positives during culling, transform should have broader tangents and depth range compared to one will be used for rendering. As the the time performing culling becomes more distant from presentation time, renderer should request for wider tangents/depth.

drawableTargetViewCount(target:) for given target should be more than 1 view.