Does it matter if two images are different sizes when using sp.signal.correlate

Question

I am just wondering if having padding around one image effects the correlation functions scipy.signal.correlate() result or does it ignore that. basically I am trying to see