GFlow: Recovering 4D World from Monocular Video

• URL: http://arxiv.org/abs/2405.18426v2
• Date: Tue, 31 Dec 2024 07:05:28 GMT
• Title: GFlow: Recovering 4D World from Monocular Video
• Authors: Shizun Wang, Xingyi Yang, Qiuhong Shen, Zhenxiang Jiang, Xinchao Wang,
• Abstract summary: We introduce GFlow, a framework that lifts a video to a 4D scene as a flow of 3D Gaussians through space and time. GFlow segment the video into still and moving parts, then alternates between optimizing camera poses and the dynamics of the 3D Gaussian points. GFlow estimates the camera poses for each frame, enabling novel view synthesis by changing camera pose.
• Score: 58.63051670458107
• License:
• Abstract: Recovering 4D world from monocular video is a crucial yet challenging task. Conventional methods usually rely on the assumptions of multi-view videos, known camera parameters, or static scenes. In this paper, we relax all these constraints and tackle a highly ambitious but practical task: With only one monocular video without camera parameters, we aim to recover the dynamic 3D world alongside the camera poses. To solve this, we introduce GFlow, a new framework that utilizes only 2D priors (depth and optical flow) to lift a video to a 4D scene, as a flow of 3D Gaussians through space and time. GFlow starts by segmenting the video into still and moving parts, then alternates between optimizing camera poses and the dynamics of the 3D Gaussian points. This method ensures consistency among adjacent points and smooth transitions between frames. Since dynamic scenes always continually introduce new visual content, we present prior-driven initialization and pixel-wise densification strategy for Gaussian points to integrate new content. By combining all those techniques, GFlow transcends the boundaries of 4D recovery from causal videos; it naturally enables tracking of points and segmentation of moving objects across frames. Additionally, GFlow estimates the camera poses for each frame, enabling novel view synthesis by changing camera pose. This capability facilitates extensive scene-level or object-level editing, highlighting GFlow's versatility and effectiveness. Visit our project page at: https://littlepure2333.github.io/GFlow

Related papers

• GS-DiT: Advancing Video Generation with Pseudo 4D Gaussian Fields through Efficient Dense 3D Point Tracking [38.104532522698285]
  Training a video Diffusion Transformer (DiT) directly to control 4D content requires expensive multi-view videos. Inspired by Monocular Dynamic novel View Synthesis (MDVS), we bring pseudo 4D Gaussian fields to video generation. We finetune a pretrained DiT to generate videos following the guidance of the rendered video, dubbed as GS-DiT.
  arXiv  Detail & Related papers  (2025-01-05T23:55:33Z)
• Deblur4DGS: 4D Gaussian Splatting from Blurry Monocular Video [64.38566659338751]
  We propose the first 4D Gaussian Splatting framework to reconstruct a high-quality 4D model from blurry monocular video, named Deblur4DGS. We introduce exposure regularization to avoid trivial solutions, as well as multi-frame and multi-resolution consistency ones to alleviate artifacts. Beyond novel-view, Deblur4DGS can be applied to improve blurry video from multiple perspectives, including deblurring, frame synthesis, and video stabilization.
  arXiv  Detail & Related papers  (2024-12-09T12:02:11Z)
• RoMo: Robust Motion Segmentation Improves Structure from Motion [46.77236343300953]
  We propose a novel approach to video-based motion segmentation to identify the components of a scene that are moving w.r.t. a fixed world frame. Our simple but effective iterative method, RoMo, combines optical flow and epipolar cues with a pre-trained video segmentation model. More importantly, the combination of an off-the-shelf SfM pipeline with our segmentation masks establishes a new state-of-the-art on camera calibration for scenes with dynamic content, outperforming existing methods by a substantial margin.
  arXiv  Detail & Related papers  (2024-11-27T01:09:56Z)
• Cavia: Camera-controllable Multi-view Video Diffusion with View-Integrated Attention [62.2447324481159]
  Cavia is a novel framework for camera-controllable, multi-view video generation. Our framework extends the spatial and temporal attention modules, improving both viewpoint and temporal consistency. Cavia is the first of its kind that allows the user to specify distinct camera motion while obtaining object motion.
  arXiv  Detail & Related papers  (2024-10-14T17:46:32Z)
• DreamScene4D: Dynamic Multi-Object Scene Generation from Monocular Videos [21.93514516437402]
  We present DreamScene4D, the first approach to generate 3D dynamic scenes of multiple objects from monocular videos via novel view synthesis. Our key insight is a "decompose-recompose" approach that factorizes the video scene into the background and object tracks. We show extensive results on challenging DAVIS, Kubric, and self-captured videos with quantitative comparisons and a user preference study.
  arXiv  Detail & Related papers  (2024-05-03T17:55:34Z)
• COLMAP-Free 3D Gaussian Splatting [88.420322646756]
  We propose a novel method to perform novel view synthesis without any SfM preprocessing. We process the input frames in a sequential manner and progressively grow the 3D Gaussians set by taking one input frame at a time. Our method significantly improves over previous approaches in view synthesis and camera pose estimation under large motion changes.
  arXiv  Detail & Related papers  (2023-12-12T18:39:52Z)
• Make-It-4D: Synthesizing a Consistent Long-Term Dynamic Scene Video from a Single Image [59.18564636990079]
  We study the problem of synthesizing a long-term dynamic video from only a single image. Existing methods either hallucinate inconsistent perpetual views or struggle with long camera trajectories. We present Make-It-4D, a novel method that can generate a consistent long-term dynamic video from a single image.
  arXiv  Detail & Related papers  (2023-08-20T12:53:50Z)
• FlowCam: Training Generalizable 3D Radiance Fields without Camera Poses via Pixel-Aligned Scene Flow [26.528667940013598]
  Reconstruction of 3D neural fields from posed images has emerged as a promising method for self-supervised representation learning. Key challenge preventing the deployment of these 3D scene learners on large-scale video data is their dependence on precise camera poses from structure-from-motion. We propose a method that jointly reconstructs camera poses and 3D neural scene representations online and in a single forward pass.
  arXiv  Detail & Related papers  (2023-05-31T20:58:46Z)
• Consistent Depth of Moving Objects in Video [52.72092264848864]
  We present a method to estimate depth of a dynamic scene, containing arbitrary moving objects, from an ordinary video captured with a moving camera. We formulate this objective in a new test-time training framework where a depth-prediction CNN is trained in tandem with an auxiliary scene-flow prediction over the entire input video. We demonstrate accurate and temporally coherent results on a variety of challenging videos containing diverse moving objects (pets, people, cars) as well as camera motion.
  arXiv  Detail & Related papers  (2021-08-02T20:53:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.