For Camera Frustum tool, I published a custom rig first featuring a calibrated pyramid driven by live expression evaluations.
 

Key Achievements:
 

• Converted live-action film language into mathematical programming.

• User-friendly design to easily increment the distance (+Z) from camera by dragging a constrained and bi-directionally synchronized locator.
  Being dragged towards or away from the camera, the pyramid gets scaled along the Z-axis, while a linked expression automatically drives the inverse proportion of the X and Y scales to keep the pyramid fitted to the viewport resolution and film back ratio.

• By utilizing the pyramid mechanism described above, the tool provides 8-corner matrix frustum that can alternate a legacy bounding box.
  With two pyramids designated each as foreground and background, each matching pyramid matrix can draw a linear line trace between the minimum and maximum coordinates, resulting in a fully dimensioned 8-corner frustum fit to the pyramids. Stage Operators could query the in--camera assets specifically by checking if the center pivot position of an asset is more than min and less than max range of frustum corner matrices. Tested to be more accurate than the bounding box matrices of pyramid, as they are intended to be uniform cubes at all times - it would falsely conclude assets that are not in-camera as within range, whereas 8-corner is synchronized with the camera field of view.

• By automation, the tool inspects either selected assets, or the entire Workspace into three Set membership;
  1. Assets placed between the camera lens and the foreground pyramid
  2. Assets placed within the 8-corner frustum
  3. Assets placed beyond the background pyramid.
  
  All members are in-camera. The tool is designed for Stage Operators to strategically execute optimization based on the result.

​