I develop and present an improved method for modeling and removing the contamination from WFC3 Infrared Spectroscopic Parallel (WISP) Survey grism data. WISP takes slitless spectroscopy data, which means that spectra of di erent objects may overlap in the observations. The current cleaning algorithm assumes a multi-Gaussian intensity pro le for modeling the overlapping objects. The new algorithm I develop calculates the pro les from the direct image of the galaxy. I discuss current shortcomings of my method and where it can be improved. This algorithm will be made publicly available and eventually incorporated into the WISP pipeline.