基于音视频的自动化低成本VR视频生成方法研究

Research on Automated Low-cost Virtual Reality Video Generation Method Based on Raw Audio and Video Materials

虚拟现实(VR)技术的发展和相应硬件设备的普及,使得VR视频内容具有非常大的发展潜力。但VR视频的制作存在两个方面的挑战:一是新内容的VR视频生成成本很高;二是过去的影音资料难以重新录制成VR视频。本文提出一种新颖、低成本的利用已有影音资料生成VR视频的方法,该方法结合自然语言处理技术(NLP)、3D建模、虚拟现实等技术,可以快速、低成本生成VR视频。实验表明,本文方法可以大幅度节省制作成本,过去的音视频也可以生成沉浸感强的VR视频。

With the development of virtual reality (VR) technology and the popularity of corresponding hardware devices, VR videos present a very bright developing prospect among the emerging technologies. However, there are two major challenges in the production of VR videos. First, the cost of producing new VR video is very high. Second, it is difficult to transform past audio or video data to VR formats using the normal VR generation approaches. Addressing these problems, this paper proposes an automatic and low-cost method to generate VR videos using the existing low-cost audio and video materials. The proposed method integrates the technologies of Natural Language Processing (NLP), 3D modeling, virtual reality to produce high quality VR videos in a fast, low cost and automatic way. Experimental results show that cost can be greatly saved by the proposed method. Furthermore, it is a novel way to provide VR videos for the old precious audios or videos.