The project implements a collaborative neural rendering pipeline that takes hand-drawn anime character sheets as input and outputs vivid dancing videos. It uses deep learning techniques and PyTorch to enable the neural rendering of anime-style content from static 2D character illustrations. The system was developed by Megvii Research and accepted to the IJCAI 2023 Special Track on AI, the Arts, and Creativity.