Does sourceRectAtTime() report layer bounds before or after masks and effects?

When building expressions that rely on sourceRectAtTime() to drive layout or collision logic, it is essential to know whether the function returns the layer's original source dimensions, the dimensions after masks have been applied, or the dimensions after both masks and effects have altered the visible area. After Effects processes each layer in a fixed order—masks first, then effects, then transform properties—but the documentation does not specify at which point sourceRectAtTime() samples the geometry.

This uncertainty affects rigs that attempt to mask‑crop footage before applying blurs, distortions, or third‑party plugins, because the expression may read stale or premature bounds. In a typical composition with a masked solid and a Gaussian blur, does sourceRectAtTime() report the mask‑constrained size, the blurred‑extended size, or the unmodified source size?