大规模场景实时渲染若干技术问题研究
|
摘要
大规模场景实时渲染的目的就是保证使用者在交互式的帧率下同大场景中的模型对象进行实时操作,并实时计算和动态绘制场景中的角色对象位姿,随着场景中视点的改变,画面的刷新速度要达到人眼觉察不到的闪烁,通常每秒20-30帧,不会产生跳帧现象或者内存被耗尽,出现死机等现象。目前,随着数字摄影测量和遥感技术的发展,获取海量数据已经成为可能,由这些海量数据构成虚拟现实系统大规模的场景,其中模型就有上万个甚至几百万个多边形,再加上高分辨率的卫星影像图作为纹理耗费了大量的存储空间,同时,还有可能需要环境光照,动画,声音效果等特效来增强现实感,如果再对地理数据的属性特征和位置特征进行实时再现,即使再高级的硬件配置也会显得力不从心。研究大规模场景在虚拟现实系统中实时渲染已经成为热门课题,它在数字城市的建设、3D游戏、虚拟战场、三维地理信息系统发展等众多领域都有着广阔的应用前景。
本文在研究虚拟场景渲染的理论和基于软件的加速技术基础上,以Multigen-Creator为建模工具,以Vega为场景的驱动平台,在没有昂贵的SGI图形工作站等硬件支持下,运用Polymesh算法、对Polymesh改进的Irregular Mesh算法、Delaunay算法以及对其改进的CAT算法对实验区域进行地形细节层次的生成,通过对生成细节层次模型的多边形数目统计、误差分析、计算速度和适用范围得出四种算法的优缺点;运用大规模数据库管理技术,对超出图形硬件生成能力的大规模场景模型进行实时管理和调度,既降低了场景中多边形的数目,提高了渲染效率,又保证了实时渲染的帧数均匀性。设计和实现的基于不同视点(地表和空中)虚拟仿真系统,经过测试能够在一般配置的微机上流畅运行,满足实时性的要求,对大规模场景的实时渲染做了有益尝试。
The aim of real-time rendering of large-scale scenes is to ensure that the user can have real-time operation on the object model of large-scale scenes at interactive frame rate, at the same time, calculate player's position and orientation in real time, render the player of scenes dynamically, with point of view's change, screen's refresh rate should make the human eyes not detect the flicker, usually 20-30 frames per second, frame-skipping and system halted phenomenon does not happen, memory is not exhausted. With the development of digital photogrammetry and remote sensing technology, massive data can be got possibly, these massive data constitute large-scale virtual reality scenes, of which there are tens of thousands of polygons or even millions of polygons, together with high-resolution satellite image as a texture to spend a lot of storage space, meanwhile, may need to have environmental illumination, animation, sound effects and other effects to enhance a sense's reality, if property characteristics and location characteristics of the geographical data reproduce in real time, even if the advanced hardware configuration also appeared to be inadequate. Research on real-time rendering of large-scale virtual reality scenes has become a hot topic, which has a wide prospect of application at the digital city building, 3D games, virtual battlefield, three-dimensional geographical information systems development, and many other fields.
Based on studying the theory of virtual scenes rendering and software-based acceleration technology, this paper uses Multigen-Creator for modeling tools and Vega for the scene-driven platform, in the absence of expensive SGI graphics workstations and other hardware support, the employment of Polymesh algorithm, Irregular Mesh algorithm to improve Polymesh algorithm, Delaunay algorithm and its improved algorithm CAT, to generate terrain levels of detail for the experimental region, through the LOD generation, statistics the polygon numbers of LOD models, analyse the LOD error, calculate the generation speed and summarize the application scope of the four algorithms to draw the advantages and disadvantages; The employment of large area database management technology to manage and attemper the models of large-scale scenes beyond the ability of graphics hardware in real time, which not only reduces the polygon numbers of scenes, improve rendering efficiency and make sure the frames uniformity of real-time rendering. Design and empolder the virtual simulation system based on the different point of view (surface and air), the virtual simulation system can run smoothly and meet the requirements of real-time in PC though testing, which makes a useful attempt to real-time rendering of large-scale of scenes.
本文在研究虚拟场景渲染的理论和基于软件的加速技术基础上,以Multigen-Creator为建模工具,以Vega为场景的驱动平台,在没有昂贵的SGI图形工作站等硬件支持下,运用Polymesh算法、对Polymesh改进的Irregular Mesh算法、Delaunay算法以及对其改进的CAT算法对实验区域进行地形细节层次的生成,通过对生成细节层次模型的多边形数目统计、误差分析、计算速度和适用范围得出四种算法的优缺点;运用大规模数据库管理技术,对超出图形硬件生成能力的大规模场景模型进行实时管理和调度,既降低了场景中多边形的数目,提高了渲染效率,又保证了实时渲染的帧数均匀性。设计和实现的基于不同视点(地表和空中)虚拟仿真系统,经过测试能够在一般配置的微机上流畅运行,满足实时性的要求,对大规模场景的实时渲染做了有益尝试。
The aim of real-time rendering of large-scale scenes is to ensure that the user can have real-time operation on the object model of large-scale scenes at interactive frame rate, at the same time, calculate player's position and orientation in real time, render the player of scenes dynamically, with point of view's change, screen's refresh rate should make the human eyes not detect the flicker, usually 20-30 frames per second, frame-skipping and system halted phenomenon does not happen, memory is not exhausted. With the development of digital photogrammetry and remote sensing technology, massive data can be got possibly, these massive data constitute large-scale virtual reality scenes, of which there are tens of thousands of polygons or even millions of polygons, together with high-resolution satellite image as a texture to spend a lot of storage space, meanwhile, may need to have environmental illumination, animation, sound effects and other effects to enhance a sense's reality, if property characteristics and location characteristics of the geographical data reproduce in real time, even if the advanced hardware configuration also appeared to be inadequate. Research on real-time rendering of large-scale virtual reality scenes has become a hot topic, which has a wide prospect of application at the digital city building, 3D games, virtual battlefield, three-dimensional geographical information systems development, and many other fields.
Based on studying the theory of virtual scenes rendering and software-based acceleration technology, this paper uses Multigen-Creator for modeling tools and Vega for the scene-driven platform, in the absence of expensive SGI graphics workstations and other hardware support, the employment of Polymesh algorithm, Irregular Mesh algorithm to improve Polymesh algorithm, Delaunay algorithm and its improved algorithm CAT, to generate terrain levels of detail for the experimental region, through the LOD generation, statistics the polygon numbers of LOD models, analyse the LOD error, calculate the generation speed and summarize the application scope of the four algorithms to draw the advantages and disadvantages; The employment of large area database management technology to manage and attemper the models of large-scale scenes beyond the ability of graphics hardware in real time, which not only reduces the polygon numbers of scenes, improve rendering efficiency and make sure the frames uniformity of real-time rendering. Design and empolder the virtual simulation system based on the different point of view (surface and air), the virtual simulation system can run smoothly and meet the requirements of real-time in PC though testing, which makes a useful attempt to real-time rendering of large-scale of scenes.
引文
[1]高俊.地理空间数据可视化[M].北京:测绘工程,2000:10-65
[2]吴立新,史文中.地理信息系统原理与算法[M].北京:科学出版社,2002:380-397
[3]张祖勋,张剑清.数字摄影测量学[M].武汉:武汉测绘科技大学出版社,1996:34-87
[4]李成名,王继周,马照亭.数字城市三维地理空间框架的原理与方法[M].北京:科学出版社,2008:148-175
[5]危拥军,王家耀,陈广学.三维地理信息系统基本框架研究[J].测绘科学与工程,2005,25(2):18-22
[6]Kalvin A D,Taylor R H.Superfaces:Polygonal Mesh Simplification With Bounded Error[J].IEEE Computer Graphics and Application.1996:16(3)65-76
[7]Hoppe,H.Progressive mesh[A].Computer Graphics(SIG-GRAPH'96 Proceedings)[C].New Orleans,Louisiana:[s.n.],1996.99-108
[8]M.Garland,P.Heckbert.Surface simplification using quadric error metrics[R].In:Proceedings of SIGGRAPH 97,Computer Graphics Proceedings,Annual Conference Series,Los Angeles,1997
[9]Schroeder,W J.Decimation of triangle meshes[J].Computer Graphics,1992,V26(2):65-70
[10]潘志庚.虚拟环境中多细节层次模型自动生成算法[J].软件学报,1996,7(9):525-532
[11]张明敏,周昆,潘志庚.基于超包络的三角形网格简化算法[J].软件学报,1999,10(10):584-588
[12]Matthias Eck,Derose Tony,et al.Multiresolution Analysis of Arbitrary Meshes[J].Proceeding of SIGGRAPH,1995:173-182
[13]Rossignac,J.,Borrel,P.Multi-resolution 3D approximation for rendering complex scenes[J].Geometric Modeling in Computer Graphics.1993:455.-465
[14]Turk,G.Re-tiling polygonal surface[J].Computer Graphics,1996,V26(2):55-64
[15]Cohen,J.,Marc Olano.Simplifying Polygonal Models Using Successive Mappings[R].In:Proceedings of the 8th conference on Visualization '97,Oct.1997 - Oct.1997,Phoenix,Arizona,United States
[16]F.Losasso,H.Hoppe,S.Schaefer,et al.Smooth geometry,images[R].In:Proceedings of the Eurographics/ACM SIGGRAPH symposium on Geometry processing,June 23-25,2003,Aachen,Germany.
[17]H.Hoppe,E.Praun.Shape Compression Using Spherical Geometry Images[J].Advances in Multiresolution for Geometric Modeling,N.Dodgson,M.Floater,M.Sabin(eds.),Springe-Verlag,2005:27-46
[18]JeffBolz,Ian Farmer,Eitan Grinspun,et al.Sparse Matrix Solvers on the GPU:Conjugate Gradients and Multigrid[C].ACM Transactions on Graphcs(Proceedings of SIGGRAPH),pages 9t7-924,July 2003.
[19]J.Gain,R.Southern.Creation and Control of Real-time Continuous Level of Detail on Program- mable Graphics Hardware.Computer Graphics Forum,March 2003
[20]F.Losasso,H.Hoppe,S.Schaefer,J.Warren et al.Geometry Clipmaps:Terrain Rendering using Nested Regular Grids[J].Proceedings of SIGGRAPH 2004,23(3)[21]C.Dachsbacher,M.Stamminger.Rendering Procedural Terrain by Geometry Image Warping[R].In:Eurographics Symposium on Rendering,Norrk(O|¨)ping,Sweden,June 21-23,2004
[22]冷志光.复杂场景绘制关键技术研究[D].长沙:国防科学技术大学,2005
[23]朱凌峰.大规模虚拟场景实时仿真技术研究[D].北京:中国科学院遥感应用研究所,2003
[24]翼俊峰.细节复杂模型实时绘制加速技术研究[D].成都:中国科学院软件研究所,2005
[25]秦志光.三维复杂场景实时绘制技术[D].成都:电子科技大学,2007
[26]郭阳明.虚拟现实实时性研究[D].西安:西北工业大学,2005
[27]邬彩霞.一种大规模地形场景绘制的方法[D].大连:大连理工大学,2007
[28]淮永建.虚拟场景中实时图形绘制关键技术研究[D].西安:西北工业大学,2002
[29]MultiGen Vega User's Guide Version 3.7.1[Z].MultiGen-Paradigm Inc.2003.
[30]王乘,李利军,周均清等.Vega实时三维视景仿真技术[M].武汉:华中科技大学出版社,2005:82-235
[31]安吉尔(Angel,E.)著,吴文国译.交互式计算机图形学——基于OpenGl的自顶向下方法[M].北京:清华大学出版社,2007:324-330
[32]Slater,M.,Chrysanthou,Y.View Volume culling using a probabilistic cashing scheme[J].Proceedings of Framework for Immersive Virtual Environments,1996,V5(12):71-77
[33]郑文庭,鲍虎军,彭群生.基于时空连贯性和几何简化技术的复杂场景快速消隐绘制算法[J].计算机学报.2000,23(8):836-837
[34]Kumar,S.,Manocha,D.,Garrett,W.,et al.Hierarchical back-face culling[J].Computers and Graphics,1995,V23(5):681-692
[35]潘志庚,马小虎,石教英.多细节层次模型自动生成技术综述[J].中国图像图形学报,1998,3(9):754-759
[36]Clark,J.Hierarchical Geometric Models for Visible Surface Algorithms[J].Communications of ACM,1976,547-554.
[37]Stefan R(o|¨)ttger,Volfgang Heidrich,Philipp Slussallek,et al.Real-time Generation of Continuous Levels of Detail for Height Fields[C]Proceeding of the 6th International Conference in Central Europe on Computer Graphics and Visiualization,1998,315-322.
[38]冷志光,汤晓安,郝建新等.大规模地形动态快速绘制技术研究[J].系统仿真学报,2006,18(10):2832-2835
[39]翟巍.三维GIS中大规模场景数据获取、组织、及调度方法的研究与实现[D].大连:大连理工大学,2003
[40]Mark Duehaineauy,Murray Wolinsky.ROAMing Terrain:Real-time Optimally Adapting Meshes[J]IEEE Visualization'97 Proeeedings(0-8186-8262-0),1997.
[41]周艳,朱庆,黄铎.三维城市模型中建筑物LOD模型研究[J].测绘科学,2006,31(5):74-77
[42]王乘,周均清,李利军.Creator可视化仿真建模技术[M].武汉:华中科技大学出版社,2005,251-255
[43]汤国安,刘学军,闾国年.数字高程模型及地学分析的原理与方法[M].北京:科学出版社,2005:2-15
[44]MultiGen Creator Creating Terrain for Simulations Version 3.0 for Windows[Z].MultiGen- Paradigm Inc.2004
[45]乔勇军.MultiGen视景仿真中大地形与航迹生成方法研究[J].航空计算技术,2005,35(2):54-56
[46]董博.基于MFC的Vega应用程序设计[J].仿真技术,2002,2(2):264-265
[47]MultiGen Vega Man Pages Version 3.7.1[Z].MultiGen-Paradigm Inc.2003.
[48]张继伟.基于MultiGen vegaa拟显示平台的开发与应用[D].青岛:中国海洋大学,2006
[49]张亮.基于VR技术的三维GIS实验系统的研究[D].成都:西南交通大学,2007
[50]MultiGen Creator Creating Models for Simulations Version 3.0 for Windows[Z].MultiGen- Paradign'~Inc.2004
[2]吴立新,史文中.地理信息系统原理与算法[M].北京:科学出版社,2002:380-397
[3]张祖勋,张剑清.数字摄影测量学[M].武汉:武汉测绘科技大学出版社,1996:34-87
[4]李成名,王继周,马照亭.数字城市三维地理空间框架的原理与方法[M].北京:科学出版社,2008:148-175
[5]危拥军,王家耀,陈广学.三维地理信息系统基本框架研究[J].测绘科学与工程,2005,25(2):18-22
[6]Kalvin A D,Taylor R H.Superfaces:Polygonal Mesh Simplification With Bounded Error[J].IEEE Computer Graphics and Application.1996:16(3)65-76
[7]Hoppe,H.Progressive mesh[A].Computer Graphics(SIG-GRAPH'96 Proceedings)[C].New Orleans,Louisiana:[s.n.],1996.99-108
[8]M.Garland,P.Heckbert.Surface simplification using quadric error metrics[R].In:Proceedings of SIGGRAPH 97,Computer Graphics Proceedings,Annual Conference Series,Los Angeles,1997
[9]Schroeder,W J.Decimation of triangle meshes[J].Computer Graphics,1992,V26(2):65-70
[10]潘志庚.虚拟环境中多细节层次模型自动生成算法[J].软件学报,1996,7(9):525-532
[11]张明敏,周昆,潘志庚.基于超包络的三角形网格简化算法[J].软件学报,1999,10(10):584-588
[12]Matthias Eck,Derose Tony,et al.Multiresolution Analysis of Arbitrary Meshes[J].Proceeding of SIGGRAPH,1995:173-182
[13]Rossignac,J.,Borrel,P.Multi-resolution 3D approximation for rendering complex scenes[J].Geometric Modeling in Computer Graphics.1993:455.-465
[14]Turk,G.Re-tiling polygonal surface[J].Computer Graphics,1996,V26(2):55-64
[15]Cohen,J.,Marc Olano.Simplifying Polygonal Models Using Successive Mappings[R].In:Proceedings of the 8th conference on Visualization '97,Oct.1997 - Oct.1997,Phoenix,Arizona,United States
[16]F.Losasso,H.Hoppe,S.Schaefer,et al.Smooth geometry,images[R].In:Proceedings of the Eurographics/ACM SIGGRAPH symposium on Geometry processing,June 23-25,2003,Aachen,Germany.
[17]H.Hoppe,E.Praun.Shape Compression Using Spherical Geometry Images[J].Advances in Multiresolution for Geometric Modeling,N.Dodgson,M.Floater,M.Sabin(eds.),Springe-Verlag,2005:27-46
[18]JeffBolz,Ian Farmer,Eitan Grinspun,et al.Sparse Matrix Solvers on the GPU:Conjugate Gradients and Multigrid[C].ACM Transactions on Graphcs(Proceedings of SIGGRAPH),pages 9t7-924,July 2003.
[19]J.Gain,R.Southern.Creation and Control of Real-time Continuous Level of Detail on Program- mable Graphics Hardware.Computer Graphics Forum,March 2003
[20]F.Losasso,H.Hoppe,S.Schaefer,J.Warren et al.Geometry Clipmaps:Terrain Rendering using Nested Regular Grids[J].Proceedings of SIGGRAPH 2004,23(3)[21]C.Dachsbacher,M.Stamminger.Rendering Procedural Terrain by Geometry Image Warping[R].In:Eurographics Symposium on Rendering,Norrk(O|¨)ping,Sweden,June 21-23,2004
[22]冷志光.复杂场景绘制关键技术研究[D].长沙:国防科学技术大学,2005
[23]朱凌峰.大规模虚拟场景实时仿真技术研究[D].北京:中国科学院遥感应用研究所,2003
[24]翼俊峰.细节复杂模型实时绘制加速技术研究[D].成都:中国科学院软件研究所,2005
[25]秦志光.三维复杂场景实时绘制技术[D].成都:电子科技大学,2007
[26]郭阳明.虚拟现实实时性研究[D].西安:西北工业大学,2005
[27]邬彩霞.一种大规模地形场景绘制的方法[D].大连:大连理工大学,2007
[28]淮永建.虚拟场景中实时图形绘制关键技术研究[D].西安:西北工业大学,2002
[29]MultiGen Vega User's Guide Version 3.7.1[Z].MultiGen-Paradigm Inc.2003.
[30]王乘,李利军,周均清等.Vega实时三维视景仿真技术[M].武汉:华中科技大学出版社,2005:82-235
[31]安吉尔(Angel,E.)著,吴文国译.交互式计算机图形学——基于OpenGl的自顶向下方法[M].北京:清华大学出版社,2007:324-330
[32]Slater,M.,Chrysanthou,Y.View Volume culling using a probabilistic cashing scheme[J].Proceedings of Framework for Immersive Virtual Environments,1996,V5(12):71-77
[33]郑文庭,鲍虎军,彭群生.基于时空连贯性和几何简化技术的复杂场景快速消隐绘制算法[J].计算机学报.2000,23(8):836-837
[34]Kumar,S.,Manocha,D.,Garrett,W.,et al.Hierarchical back-face culling[J].Computers and Graphics,1995,V23(5):681-692
[35]潘志庚,马小虎,石教英.多细节层次模型自动生成技术综述[J].中国图像图形学报,1998,3(9):754-759
[36]Clark,J.Hierarchical Geometric Models for Visible Surface Algorithms[J].Communications of ACM,1976,547-554.
[37]Stefan R(o|¨)ttger,Volfgang Heidrich,Philipp Slussallek,et al.Real-time Generation of Continuous Levels of Detail for Height Fields[C]Proceeding of the 6th International Conference in Central Europe on Computer Graphics and Visiualization,1998,315-322.
[38]冷志光,汤晓安,郝建新等.大规模地形动态快速绘制技术研究[J].系统仿真学报,2006,18(10):2832-2835
[39]翟巍.三维GIS中大规模场景数据获取、组织、及调度方法的研究与实现[D].大连:大连理工大学,2003
[40]Mark Duehaineauy,Murray Wolinsky.ROAMing Terrain:Real-time Optimally Adapting Meshes[J]IEEE Visualization'97 Proeeedings(0-8186-8262-0),1997.
[41]周艳,朱庆,黄铎.三维城市模型中建筑物LOD模型研究[J].测绘科学,2006,31(5):74-77
[42]王乘,周均清,李利军.Creator可视化仿真建模技术[M].武汉:华中科技大学出版社,2005,251-255
[43]汤国安,刘学军,闾国年.数字高程模型及地学分析的原理与方法[M].北京:科学出版社,2005:2-15
[44]MultiGen Creator Creating Terrain for Simulations Version 3.0 for Windows[Z].MultiGen- Paradigm Inc.2004
[45]乔勇军.MultiGen视景仿真中大地形与航迹生成方法研究[J].航空计算技术,2005,35(2):54-56
[46]董博.基于MFC的Vega应用程序设计[J].仿真技术,2002,2(2):264-265
[47]MultiGen Vega Man Pages Version 3.7.1[Z].MultiGen-Paradigm Inc.2003.
[48]张继伟.基于MultiGen vegaa拟显示平台的开发与应用[D].青岛:中国海洋大学,2006
[49]张亮.基于VR技术的三维GIS实验系统的研究[D].成都:西南交通大学,2007
[50]MultiGen Creator Creating Models for Simulations Version 3.0 for Windows[Z].MultiGen- Paradign'~Inc.2004