||Scale Invariant feature transform is widely used in image analysis for detecting salient features in an image. SIFT is widely used in handwritten document analysis tasks such as transcription. However the features that are captured by SIFT that makes it wide spread in state of the art methods is not explored. In a attempt to better understand the features captured by this method when applied to character recognition, dense SIFT is applied to each pixel in a character image. The result of this approach at varying scale are presented.