In the deep learning era, long video generation of high-quality still remains\nchallenging due to the spatio-temporal complexity and continuity of videos.\nExisting prior works have attempted to model video distribution by representing\nvideos as 3D grids of RGB values, which impedes the scale of generated videos\nand neglects continuous dynamics. In this paper, we found that the recent\nemerging paradigm of implicit neural representations (INRs) that encodes a\ncontinuous signal into a parameterized neural network effectively mitigates the\nissue. By utilizing INRs of video, we propose dynamics-aware implicit\ngenerative adversarial network (DIGAN), a novel generative adversarial network\nfor video generation. Specifically, we introduce (a) an INR-based video\ngenerator that improves the motion dynamics by manipulating the space and time\ncoordinates differently and (b) a motion discriminator that efficiently\nidentifies the unnatural motions without observing the entire long frame\nsequences. We demonstrate the superiority of DIGAN under various datasets,\nalong with multiple intriguing properties, e.g., long video synthesis, video\nextrapolation, and non-autoregressive video generation. For example, DIGAN\nimproves the previous state-of-the-art FVD score on UCF-101 by 30.7% and can be\ntrained on 128 frame videos of 128x128 resolution, 80 frames longer than the 48\nframes of the previous state-of-the-art method.\n