Abstract: In one or more embodiments, a computing device is configured to modify an original video by applying a machine learning model. The computing device obtains multiple training data sets, with each particular training data set including an original video and a corresponding modified video. One or more frames from the original video are cropped to generate corresponding frames in the corresponding modified video. The computing device trains a machine learning model, using the training data sets, to generate modified videos from original videos such that one or more frames in the original videos are modified to generate corresponding frames in respective modified videos. Once the machine learning model is trained, the computing device obtains a target original video and applies the trained machine learning model to the target original video to generate a target modified video.