`Camera.LayerMask` has no effect on a camera that renders into its own `FrameBuffer`

[SimonMeysman]

Evergine 2026.5.26.921 · .NET 10 · Windows 11 · DX11 head (and WebGPU head where noted).
Desktop application: a 3D viewer for building models.

What happens

A Camera3D with its own FrameBuffer renders the entire scene into that target regardless of its layer
mask. We want it to draw exactly one render layer, so we can build a screen-space mask of selected objects.

What we expect

With LayerMaskDefaultValue = false and a single true entry, the camera draws only that layer — the way the
main camera does.

Evidence

A RenderDoc capture of one frame shows 652 colour draws and 782 depth draws into a target that should
receive one draw — model, ground grid and UI alike. Dumping the colour target confirms it: a complete render of
the scene.

At the moment of those draws the camera's state, logged from our own code, is exactly what it should be:

LayerMaskDefaultValue = False
LayerMask entries     = 9        (8 explicit false, 1 true for the layer we want)
requested layer key   = present, value = True

What we already ruled out

• Not the key scheme. We use camera.LayerMask[layer.Id.GetHashCode()], and the main camera honours the
  same dictionary with the same keys — its exclusions work.
• Not ordering. Tried setting the entry before and after entityManager.Add(...).
• Not creation time. Tried building the camera during scene setup and lazily later; no difference.
• Not a missed change notification. Tried re-asserting LayerMaskDefaultValue after attach, and writing the
  entry false then true to force a change.
• Not reliance on the default. Tried naming every layer explicitly with false rather than depending on
  LayerMaskDefaultValue.
• Camera.LayerBoolMask is null — but it is null on the main camera too, so that does not appear to be the
  consulted state.

Questions

1. Is LayerMask supposed to apply to a camera with a FrameBuffer? If not, is that documented anywhere?
2. If this is by design, what is the supported way to render a subset of the scene into a texture? A
   per-camera replacement/override material, a custom render step, or a second Scene rendered to a
   FrameBuffer — whichever is intended. A pointer to a sample would be ideal.