Neural Deformable Voxel Grid for Fast Optimization of Dynamic View Synthesis

نویسندگان

چکیده

Recently, Neural Radiance Fields (NeRF) is revolutionizing the task of novel view synthesis (NVS) for its superior performance. In this paper, we propose to synthesize dynamic scenes. Extending methods static scenes not straightforward as both scene geometry and appearance change over time, especially under monocular setup. Also, existing NeRF generally require a lengthy per-scene training procedure, where multi-layer perceptrons (MLP) are fitted model motions radiance. built on top recent advances in voxel-grid optimization, fast deformable radiance field method handle Our consists two modules. The first module adopts deformation grid store 3D features, light-weight MLP decoding that maps point observation space canonical using interpolated features. second contains density color scene. occlusion explicitly modeled further improve rendering quality. Experimental results show our achieves comparable performance D-NeRF only 20 minutes training, which more than 70 $$\times $$ faster D-NeRF, clearly demonstrating efficiency proposed method.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

diagnostic and developmental potentials of dynamic assessment for writing skill

این پایان نامه بدنبال بررسی کاربرد ارزیابی مستمر در یک محیط یادگیری زبان دوم از طریق طرح چهار سوال تحقیق زیر بود: (1) درک توانایی های فراگیران زمانیکه که از طریق برآورد عملکرد مستقل آنها امکان پذیر نباشد اما در طول جلسات ارزیابی مستمر مشخص شوند; (2) امکان تقویت توانایی های فراگیران از طریق ارزیابی مستمر; (3) سودمندی ارزیابی مستمر در هدایت آموزش فردی به سمتی که به منطقه ی تقریبی رشد افراد حساس ا...

15 صفحه اول

A Multiple View Layered Representation for Dynamic Novel View Synthesis

We propose a multiple view layered representation for tracking and segmentation of multiple objects in a scene. Existing layered approaches are dominated by the single view case and generally exploit only motion cues. We extend this to integrate static, dynamic and structural cues over a pair of views. The goal is to update coherent correspondence information sequentially, producing a multi-obj...

متن کامل

Fast dynamic simulation of rigid and deformable objects

This paper present a compleat dynamic modeling system: Robot. This system allows us to create a physical model of an object and to simulate its motion, its deformations, and its interaction with the environment and with other objects. This system takes into account three types of interactions: the collision, the static and dynamic friction, and the viscosity of the environment. An hierarchical ...

متن کامل

Neural dynamic optimization for control systems.III. Applications

For pt.II. see ibid., p. 490-501. The paper presents neural dynamic optimization (NDO) as a method of optimal feedback control for nonlinear multi-input-multi-output (MIMO) systems. The main feature of NDO is that it enables neural networks to approximate the optimal feedback solution whose existence dynamic programming (DP) justifies, thereby reducing the complexities of computation and storag...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Lecture Notes in Computer Science

سال: 2023

ISSN: ['1611-3349', '0302-9743']

DOI: https://doi.org/10.1007/978-3-031-26319-4_27