Numerous images of atarget human subject or ascene are required for training. Second, a learned model has limited generalization capability.

We study the problem of video-to-video synthesis, whose goal is to learn a mapping function from an input source video (e.g., a sequence of semantic segmentation masks) to an output photorealistic video that precisely depicts the content of the source video. While its image counterpart, the image-to-image translation problem, is a popular topic, the video-to-video synthesis problem is less explored in the literature. Without modeling temporal dynamics, directly applying existing image synthesis approaches to an input video often results in temporally incoherent videos of low visual quality. In this paper, we propose a video-to-video synthesis approach under the generative adversarial learning framework. Through carefully-designed generators and discriminators, coupled with a spatio-temporal adversarial objective, we achieve high-resolution, photorealistic, temporally coherent video results on a diverse set of input formats including segmentation masks, sketches, and poses. Experiments on multiple benchmarks show the advantage of our method compared to strong baselines. In particular, our model is capable of synthesizing 2K resolution videos of street scenes up to 30 seconds long, which significantly advances the state-of-the-art of video synthesis. Finally, we apply our method to future video prediction, outperforming several competing systems. Code, models, and more results are available at our website: https://github.com/NVIDIA/vid2vid. (Please use Adobe Reader to see the embedded videos in the paper.)

This paper focuses on video-2-video synthesis, i.e. given a real video the goal is to learn a model that outputs a new photorealistic and temporally consistent video with (ideally) the same data distribution, preserving the content and style of the source video. Existing image-2-image methods produce photorealistic images, but they do not account for the temporal dimension, resulting in high-frequency artifacts across time. This work builds on existing image-2-image works and mainly extends them into the temporal dimension to ensure temporal coherence. By employing conditional GANs the method provides high-level control over the output, e.g. Although the theoretical background and components are employed from past work, there is significant amount of effort in putting them together and adding the temporal extension.

2D animation is a common factor in game development, used for characters, effects and background art. It involves work that takes both skill and time, but parts of which are repetitive and tedious. Automated animation approaches exist, but are designed without animators in mind. The focus is heavily on real-life video, which follows strict laws of how objects move, and does not account for the stylistic movement often present in 2D animation. We propose a problem formulation that more closely adheres to the standard workflow of animation. We also demonstrate a model, SketchBetween, which learns to map between keyframes and sketched in-betweens to rendered sprite animations. We demonstrate that our problem formulation provides the required information for the task and that our model outperforms an existing method.

We study the problem of video-to-video synthesis, whose goal is to learn a mapping function from an input source video (e.g., a sequence of semantic segmentation masks) to an output photorealistic video that precisely depicts the content of the source video. While its image counterpart, the image-to-image translation problem, is a popular topic, the video-to-video synthesis problem is less explored in the literature. Without modeling temporal dynamics, directly applying existing image synthesis approaches to an input video often results in temporally incoherent videos of low visual quality. In this paper, we propose a video-to-video synthesis approach under the generative adversarial learning framework. Through carefully-designed generators and discriminators, coupled with a spatio-temporal adversarial objective, we achieve high-resolution, photorealistic, temporally coherent video results on a diverse set of input formats including segmentation masks, sketches, and poses.