For large images, when correcting with higher orders (typically N>4), the math produces very large numbers, that will exceed the capacity of the floating point values used in the code. There is a scaling constant, called Scalar that must be set to values > 1 as the order gets larger. If you get this error message, you must increase the scalar value. Increase it by factors of 10 until the correction runs. Running the correction with an unnecessarily high scalar value will introduce round-off errors.