We are using ISIS, the optimal image subtraction method developed by C. Alard & R.H. Lupton, to search for SNe in our images. After aligning the images of the same galaxy obtained at different epochs carefully, the image with the better seeing is convolved to match the others with poorer seeing. The kernel used for the convolution is determined by ISIS with a least squares fit using bright sources in the field. The method is able to correct the effect of the different seeing and conditions between the frames. The first picture shows a subtraction between two images of ESO255 without using ISIS:
The residuals around the nuclei are still quite considerable. The second image shows the same subtraction, but after the convolution of the best image with ISIS:

Although the result is certainly much more satisfactory, some residuals are still present in the central part of the nucleus. This effect reduces our sensitivity in the central arcsec of the galaxies. For this reason, we are suggested to search for nuclear SNe with NICMOS on HST, to exploit its stable PSF, sensitivity and angular resolution to detect from space nuclear obscured SNe which might have been missed by ground-based surveys.