KFC-W: 从未经摆拍的互联网照片生成3D一致的视频
KFC-W: Generating 3D-Consistent Videos from Unposed Internet Photos
AI总结：
本文提出KFC-W模型，通过自监督学习从未经摆拍的互联网照片生成3D一致的视频，无需3D标注，优于现有方法，在几何和外观一致性上表现更佳，适用于3D高斯点扩散等应用。
摘要：
我们解决了从未经摆拍的互联网照片生成视频的问题。少量输入图像作为关键帧，我们的模型在它们之间插值以模拟相机路径。给定随机图像，模型捕捉底层几何、识别场景身份并关联帧的相机位置和方向的能力反映了对3D结构和场景布局的基本理解。然而，现有视频模型如Luma Dream Machine无法完成此任务。我们设计了一种自监督方法，利用视频的一致性和多视角互联网照片的多样性，训练一个可扩展的、3D感知的视频模型，无需任何3D标注如相机参数。我们验证了我们的方法在几何和外观一致性上优于所有基线。我们还展示了我们的模型适用于启用相机控制的应用，如3D高斯点扩散。我们的结果表明，仅使用视频和多视角互联网照片等2D数据即可扩展场景级3D学习。
英文摘要：
We address the problem of generating videos from unposed internet photos. A handful of input images serve as keyframes, and our model interpolates between them to simulate a path moving between the cameras. Given random images, a model's ability to capture underlying geometry, recognize scene identity, and relate frames in terms of camera position and orientation reflects a fundamental understanding of 3D structure and scene layout. However, existing video models such as Luma Dream Machine fail at this task. We design a self-supervised method that takes advantage of the consistency of videos and variability of multiview internet photos to train a scalable, 3D-aware video model without any 3D annotations such as camera parameters. We validate that our method outperforms all baselines in terms of geometric and appearance consistency. We also show our model benefits applications that enable camera control, such as 3D Gaussian Splatting. Our results suggest that we can scale up scene-level 3D learning using only 2D data such as videos and multiview internet photos.
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。
