This problem is caused by the bottleneck I inadvertently created when I forced the lens_inverse function to return a ray.
The debug.lua (previously cube.lua) is using plate_to_ray(plate,u,v). So the plate coordinate is mapped to a ray vector, then the ray vector is mapped back to the closest plate. This was bad because we're not actually seeing the whole plate, but it is also great because we can see how much of the plate is being used if we apply the rubix command. This helped me debug the tetrahedron problem so I could see how much of the plate was being used. I will keep it as is.
This problem is caused by the bottleneck I inadvertently created when I forced the lens_inverse function to return a ray.
The debug.lua (previously cube.lua) is using plate_to_ray(plate,u,v). So the plate coordinate is mapped to a ray vector, then the ray vector is mapped back to the closest plate. This was bad because we're not actually seeing the whole plate, but it is also great because we can see how much of the plate is being used if we apply the rubix command. This helped me debug the tetrahedron problem so I could see how much of the plate was being used. I will keep it as is.