To assist in creating paper-based 2D animations, an Android application was developed allowing the user to photograph a sequence of drawings with a mobile device and show the resulting video on the screen. This is known as the pencil test in the classical animation process.
The application enables artists to review the animation they are working on with fewer restrictions than in the traditional process. Particularly, the sheet of paper can be photographed by hand instead of being fixes flat under the camera or scanned. Further, it is possible to draw several frames per page, their order determined by numbers written next to them.
The prototype application described in this thesis accepts one sheet of paper with 12 frames to produce a grayscale output.Using the OpenCV library functions, the image is searched for quadrilateral contour of the paper to correct the perspective. Probabilistic Hough line transorm is employed to find individual frames. In case of interfering background, it is also applied to find the paper corners. The handwritten digits are recognized by an artificial neural network, which is trained by the MNIST dataset.
Test sequences in various environments have been created with the prototype, demonstrating a way of drawing in a natural way and without specialized workplace or equipment. This application might serve as a foundation for transferring other traditional animation tools to mobile hardware.