Camera Conventions

Radient cameras follow the glTF and OpenUSD camera convention:

• The camera looks along its local negative Z axis.
• Local positive Y is up.
• The entity world transform stores the camera transform in scene space.
• The view matrix is the inverse of the camera world transform.

This means an identity camera transform looks down scene-space -Z. To look at an object placed at the origin from in front of it, place the camera on positive Z:

RadientEntityDesc CameraDesc{};
CameraDesc.Transform.Position = {0.f, 0.f, 5.f};

Projection

Diligent projection helpers use left-handed camera space, where positive Z is forward. Radient keeps the scene camera convention as local -Z and applies the camera-space adapter inside the projection matrix.

As a result:

• mView is the true inverse of the camera world transform.
• mViewInv is the camera world transform.
• mProj includes the conversion from Radient camera space to the camera space expected by Diligent projection helpers.

Renderer and post-processing code should treat the camera world transform as a Radient scene transform. It should not add another Z flip outside the Radient camera projection path.

Clip Range And Lens Values

RadientCameraComponent::ClippingRange stores near and far clipping distances in scene units. Perspective projection is derived from:

• FocalLength
• HorizontalAperture
• VerticalAperture

Orthographic projection uses HorizontalAperture and VerticalAperture as the view size.