arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2304.02602cs.CVcs.AIcs.GR
收藏

基于3D感知扩散模型(3D-Aware Diffusion Models)的生成式新视角合成

Generative Novel View Synthesis with 3D-Aware Diffusion Models

Stanford University(斯坦福大学);NVIDIA(英伟达)

Eric R. Chan, Koki Nagano, Matthew A. Chan, Alexander W. Bergman, Jeong Joon Park, Axel Levy, Miika Aittala, Shalini De Mello, Tero Karras, Gordon Wetzstein

AI总结:

提出一种基于扩散模型的方法,通过结合2D扩散骨干网络与3D特征体几何先验,从单张图像实现3D感知的生成式新视角合成及3D一致序列的自回归合成,并在合成与真实场景中取得先进效果。

摘要:

我们提出了一种基于扩散模型(diffusion-based model)的方法,用于从少至单张输入图像进行3D感知的生成式新视角合成。我们的模型从与输入一致的可能的渲染分布中进行采样,即使在存在歧义的情况下,也能够渲染出多样且合理的新视角。为了实现这一点,我们的方法利用了现有的2D扩散骨干网络(2D diffusion backbones),但关键在于以3D特征体(3D feature volume)的形式引入了几何先验。这一潜在特征场捕捉了可能场景表示的分布,并提升了我们方法生成视角一致的新渲染图的能力。除了生成新视角外,我们的方法还具备自回归合成3D一致序列的能力。我们在合成渲染和房间尺度的场景上展示了最先进的结果;我们还在具有挑战性的真实世界物体上展示了令人信服的结果。

英文摘要:

We present a diffusion-based model for 3D-aware generative novel view synthesis from as few as a single input image. Our model samples from the distribution of possible renderings consistent with the input and, even in the presence of ambiguity, is capable of rendering diverse and plausible novel views. To achieve this, our method makes use of existing 2D diffusion backbones but, crucially, incorporates geometry priors in the form of a 3D feature volume. This latent feature field captures the distribution over possible scene representations and improves our method's ability to generate view-consistent novel renderings. In addition to generating novel views, our method has the ability to autoregressively synthesize 3D-consistent sequences. We demonstrate state-of-the-art results on synthetic renderings and room-scale scenes; we also show compelling results for challenging, real-world objects.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑