Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Temporally Smooth Mesh Extraction for Procedural Scenes with Long-Range Camera Trajectories using Spacetime Octrees

Zeyu Ma, Adam Finkelstein, Jia Deng

TL;DR

The paper tackles the challenge of extracting temporally coherent 3D meshes from unbounded procedural scenes along long-range camera trajectories by introducing BinocMesher, a spacetime binary-octree that enables 4D mesh extraction and slicing. It combines a coarse-to-fine 4D meshing pipeline with dual contouring in $\mathbb{R}^4$ and a mesh-slicing step to generate temporally smooth 3D meshes across time, controlled by a transition parameter $\delta_t$. Key contributions include the spacetime-octree data structure, efficient 4D mesh extraction and slicing, and an implementation that achieves superior visual consistency at comparable cost to baselines like Spherical Mesher and OcMesher. The approach enables scalable, temporally coherent mesh representations for animation, synthetic data generation, and rendering of large procedurally generated scenes.

Abstract

The procedural occupancy function is a flexible and compact representation for creating 3D scenes. For rasterization and other tasks, it is often necessary to extract a mesh that represents the shape. Unbounded scenes with long-range camera trajectories, such as flying through a forest, pose a unique challenge for mesh extraction. A single static mesh representing all the geometric detail necessary for the full camera path can be prohibitively large. Therefore, independent meshes can be extracted for different camera views, but this approach may lead to popping artifacts during transitions. We propose a temporally coherent method for extracting meshes suitable for long-range camera trajectories in unbounded scenes represented by an occupancy function. The key idea is to perform 4D mesh extraction using a new spacetime tree structure called a binary-octree. Experiments show that, compared to existing baseline methods, our method offers superior visual consistency at a comparable cost. The code and the supplementary video for this paper are available at https://github.com/princeton-vl/BinocMesher.

Temporally Smooth Mesh Extraction for Procedural Scenes with Long-Range Camera Trajectories using Spacetime Octrees

TL;DR

The paper tackles the challenge of extracting temporally coherent 3D meshes from unbounded procedural scenes along long-range camera trajectories by introducing BinocMesher, a spacetime binary-octree that enables 4D mesh extraction and slicing. It combines a coarse-to-fine 4D meshing pipeline with dual contouring in and a mesh-slicing step to generate temporally smooth 3D meshes across time, controlled by a transition parameter . Key contributions include the spacetime-octree data structure, efficient 4D mesh extraction and slicing, and an implementation that achieves superior visual consistency at comparable cost to baselines like Spherical Mesher and OcMesher. The approach enables scalable, temporally coherent mesh representations for animation, synthetic data generation, and rendering of large procedurally generated scenes.

Abstract

The procedural occupancy function is a flexible and compact representation for creating 3D scenes. For rasterization and other tasks, it is often necessary to extract a mesh that represents the shape. Unbounded scenes with long-range camera trajectories, such as flying through a forest, pose a unique challenge for mesh extraction. A single static mesh representing all the geometric detail necessary for the full camera path can be prohibitively large. Therefore, independent meshes can be extracted for different camera views, but this approach may lead to popping artifacts during transitions. We propose a temporally coherent method for extracting meshes suitable for long-range camera trajectories in unbounded scenes represented by an occupancy function. The key idea is to perform 4D mesh extraction using a new spacetime tree structure called a binary-octree. Experiments show that, compared to existing baseline methods, our method offers superior visual consistency at a comparable cost. The code and the supplementary video for this paper are available at https://github.com/princeton-vl/BinocMesher.

Paper Structure

This paper contains 17 sections, 5 equations, 18 figures, 2 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries: Inputs, Octree and OcMesher
  4. Method: Binary-Octree and BinocMesher
  5. Binary-Octree
  6. Coarse-to-fine Algorithm
  7. 4D Mesh Extraction
  8. Mesh Slicing
  9. Experiments
  10. Visual Consistency
  11. Computational Cost
  12. Transition Control Parameter Settings
  13. Implementation Details
  14. Efficient Implementation of Tree Construction
  15. Node Groups and Dual Polyhedron Search
  16. ...and 2 more sections

Figures (18)

  • Figure 1: Several options for a camera-dependent tree structures. The $x$ and $y$ spatial axes are labeled, while $z$ is not shown. The binary-octree (c) supports the two spatial splits shown in (b) at different times.
  • Figure 2: Overview of BinocMesher
  • Figure 3: OcMesher builds the octree from coarse to fine. Two of the spatial axes $x$ and $y$ are shown. (a) Cubes closer to the camera have larger projected diameter, and are therefore refined. (b) Nodes intersecting the surface are found. (c) Tree is refined around these intersections.
  • Figure 4: Possible binary-octrees using different split decisions. A block of 3D spacetime is shown with two spatial axes $x$ and $y$.
  • Figure 5: Proposed temporal coherence criterion.
  • ...and 13 more figures