While at Penn, I conducted a senior thesis with Professors Masao Sako and Gary Bernstein on a new method of image subtraction to find variable objects such as Type Ia Supernovae for the Dark Energy Survey pipeline. We developed a new method based on the Alard and Lupton (1998) algorithm in which the seeing in the target and reference images is matched using a discrete pixel array kernel. Our new method had two unique properties: first, we solved for separate kernels to convolve with the target and reference images; second, we not only minimized chi squared but also maximized the signal to noise in the differenced image yielding a more user friendly and efficient program.