使用色彩扩散的分层景深渲染技术

139 浏览量更新于2024-08-27 收藏 479KB PDF 举报

"Layered depth-of-field rendering using color spreading" 是一篇关于在计算机图形学中实现高质高效深度景深效果的技术论文。该方法利用颜色传播来分层渲染场景，从而达到减少内存使用并提高渲染性能的目的。深度景深（Depth of Field, DoF）是摄影和电影中常用的一种技术，它通过模糊图像中远离焦点区域的物体来增加视觉深度感，使观众的注意力集中于特定的焦点区域。在计算机图形学中，DoF常用于增强虚拟现实应用和逼真图像渲染的真实感。本文提出了一种基于颜色传播的分层渲染框架来实现DoF效果。这一框架将场景分割成多个层次，并对每一层应用扩散滤波器，直接生成最终的模糊效果。这种方法的一个关键优点是减少了对内存的需求，因为它避免了存储大量的中间缓冲区。同时，通过优化后的算法，它还能提升渲染速度，这对于实时渲染和高性能计算来说尤其重要。在GPU（图形处理单元）上实现这种效果可以充分利用其并行计算能力，加快处理速度。后处理（post-processing）技术在这里发挥了关键作用，它允许在所有图层渲染完成后，对结果进行综合处理，进一步增强DoF的效果。文章中可能包括了以下内容： 1. **分层渲染**：将场景按距离或重要性分解为多个层次，每个层次分别处理。 2. **颜色传播**：使用扩散滤波器来模拟光学系统中的弥散效应，将模糊效果应用到不同层次。 3. **内存优化**：通过直接在每个层上应用滤波器，减少了临时存储的需要，节省了内存。 4. **GPU加速**：利用GPU的并行计算能力，加速DoF的计算过程。 5. **后处理**：在所有层渲染完毕后，将它们合并，形成最终的有深度景深效果的图像。 6. **应用场景**：讨论了这种方法在虚拟现实、游戏和电影制作等领域的应用潜力。论文还可能包含了实验结果，对比了传统方法和新方法在质量和性能上的差异，以及如何调整参数以适应不同的场景需求。此外，作者可能还探讨了如何处理边界情况和交互式应用中的实时性问题。总结起来，"Layered depth-of-field rendering using color spreading" 提出了一种创新的计算机图形学技术，旨在提供高质量、高性能的深度景深渲染，对于提升虚拟现实和图像渲染的沉浸感具有重要意义。通过有效的内存管理和利用GPU的计算能力，这种方法在保证效果的同时，降低了系统资源的消耗，为实时和高性能应用提供了可能。

Permission to make digital or hard copies of part or all of this work for personal or

classroom use is granted without fee provided that copies are not made or distributed

for commercial advantage and that copies bear this notice and the full citation on the

first page. Copyrights for components of this work owned by others than ACM must be

honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on

servers, or to redistribute to lists, requires prior specific permission and/or a fee.

Request permissions from permissions@acm.org.

VRCAI 2013, November 17 – 19, 2013, Hong Kong.

Layered Depth-of-Field Rendering Using Color Spreading

Kai Yu

∗ §

Shang Wu

Bin Sheng

§ ‡

Lizhuang Ma

† §

Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China

‡

State Key Lab. of Computer Science, Institute of Software, Chinese Academy of Sciences, Beijing, China

Figure 1: Depth of ﬁeld effects. Image on the left is the original scene, the middle one is DoF effect focusing on foreground, and the right

one is focusing on background.

Abstract

Depth of ﬁeld (DoF) is a depth range outside of which objects are

blurred. DoF has been widely used in photography and ﬁlms by

artists and is expected to be applied to realistic image rendering

and virtual reality applications in the area of computer graphics.

This paper presents a framework that produces DoF effect with

high quality and performance. It is a layered method rendering

the scene into layers where spreading ﬁlter is used to directly get

the output ,reduce memory usage and improve the result comparing

with other layered methods using gathering ﬁlter. Depth peeling is

used to solve partial occlusion problem and modiﬁed to discard the

insigniﬁcant occluded pixels so as to reduce the number of occluded

layers. This framework can also eliminate artifacts such as intensity

leakage and depth discontinuity, and is GPU-friendly.

CR Categories: I.3.3 [Computer Graphics]: Image Generation—

Display algorithms;

Keywords: depth of ﬁeld, layer, post-processing, GPU

1 Introduction

In most computer applications, renders use one camera to render

the whole scene, thus all objects in the scene are clear because of

pinhole camera model, where there is an assumption that all light-

rays travel through one point before hitting the image plane [Schedl

and Wimmer 2012]. But in real world, cameras, and also human

eyes, have ﬁnite-aperture lenses, which make the light-rays can

travel along more paths besides the one through the center of the

lens. It results in a phenomenon called focal blur, meaning that the

∗

email: wsrlyk@gmail.com

†

email: ma-lz@cs.sjtu.edu.cn

objects in the scene are clear only when they are in certain range

of depth, which is called depth of ﬁeld (DoF), and other objects

are blurred, as shown in Figure 1. Focal blur has been widely used

in photography and ﬁlms, and has become a key tool to control

people’s attentions and hide the objects artists do not want to show.

In order to render scenes with DoF effect in real-time applications

such as games and virtual reality (VR) systems, a fast method is

needed. The main idea of most real-time approaches is to blur

the rendered scene with different radiuses according to the circle

of confusion (CoC) [Potmesil and Chakravarty 1981], calculated

using depth map. These methods are usually called post-processing

methods. But many such methods suffer some artifacts, for instance

intensity leakage and partial occlusion. Intensity leakage happens

when simply blurring the background with pixels in focus, while

partial occlusion is caused by the fact that a lens with a ﬁnite

aperture allows more of the scene to be visible than would be seen

through an inﬁnitesimal pinhole [Barsky et al. 2005].

This paper proposes an approach to simulate the DoF effect in

3D scenes with GPU acceleration. In this approach, scenes

are rendered into layers, which are generated using spreading

ﬁlter [Kosloff et al. 2009]. To get the occluded pixels, depth

peeling [Everitt 2001] is used when rendering the scenes, thus

removing the artifacts of partial occlusion. Some methods are used

to overcome the problems caused by splitting one object into two

or more layers. At last all layers are blended from back to front.

The contribution of this paper is a layered DoF effect framework

that gives results with high quality and performance by spreading

the fragments to build scene layers, leading to lower memory cost

compared with those that store the layers for blurring. And a

modiﬁed depth peeling is proposed to reduce the number of depth

layers.

2 Previous work

DoF algorithms can be divided into two categories, which are

object-space algorithm and image-space algorithm [Barsky and

Kosloff 2008]. Object-space algorithms directly operate the 3D

scenes, and build DoF effects directly into the rendering pipeline.

The most classic object-space method is distributed ray trac-

ing [Cook et al. 1984], which tracing several rays per pixel to

下载后可阅读完整内容，剩余5页未读，立即下载

weixin_38654220

粉丝: 10
资源: 931

使用色彩扩散的分层景深渲染技术

N-Layered.Domain-Oriented.Architecture.Guide.with.NET.4.0

SWT Layered Graph-开源

N-Layered Domain-Oriented Architecture Guide with .NET 4.0

New Layered SOA-Based Architecture for Multi-robots Cooperative Online SLAM

Isotropic layered soil-structure interaction caused by stationary random excitations

numerical-modeling-of-wave-field 各向同性、VTI、TTI以及双相介质波场数值模拟 Fortra

Polarization beam splitting through an anisotropic metamaterial slab realized by a layered metal-dielectric structure

Multi-Layered-Web-Map-Application

层叠式非对称裁剪光正交频分复用技术(Layered ACO-OFDM)及其性能提升方法

android-layered-architecture-example

最新资源