NDVI-based Vegetation Rendering CGIM ‘07 Stefan Roettger, University of Erlangen

Slides:



Advertisements
Similar presentations
Stefan Roettger, VIS Group, University of Stuttgart Ingo Frick, Massive Development, Mannheim The Terrain Rendering Pipeline
Advertisements

Graphics Pipeline.
Computer Graphic Creator: Mohsen Asghari Session 2 Fall 2014.
WSCG 2007 Hardware Independent Clipmapping A. Seoane, J. Taibo, L. Hernández, R. López, A. Jaspe VideaLAB – University of A Coruña (Spain)
Fast Volumetric Display of Natural Gaseous Phenonema Stefan Roettger Thomas Ertl University of Erlangen.
Real-Time Rendering TEXTURING Lecture 02 Marina Gavrilova.
Progressive Transmission of Appearance Preserving Octree-Textures Camille Perin Web3D 2008 August 9, 2008 Julien LacosteBruno Jobard LIUPPA University.
4.4. L EVEL OF D ETAIL Exploration of 3D mesh level-of-detail approaches.
Smooth view-dependent LOD control and its application to terrain rendering Hugues Hoppe Microsoft Research IEEE Visualization 1998.
EFFICIENT RENDERING LARGE TERRAINS USING MULTIRESOLUTION MODELLING AND IMAGE PROCESSING TECHNIQUES Ömer Nebil YAVEROĞLU Department of Computer Engineering.
Rasterization and Ray Tracing in Real-Time Applications (Games) Andrew Graff.
Smooth View-Dependent Level-of- Detail Control and its Application to Terrain Rendering Hugues Hoppe Microsoft Research.
1 Terrain Rendering and Level of Detail Week 7 Advanced Programming for 3D Applications CE
Adaptive Streaming and Rendering of Large Terrains: a Generic Solution WSCG 2009 Raphaël Lerbour Jean-Eudes Marvie Pascal Gautron THOMSON R&D, Rennes,
University of Texas at Austin CS 378 – Game Technology Don Fussell CS 378: Computer Game Technology Beyond Meshes Spring 2012.
Hidden Surface Removal
1 Computer Graphics Week13 –Shading Models. Shading Models Flat Shading Model: In this technique, each surface is assumed to have one normal vector (usually.
10/21/03CS679 - Fall Copyright Univ. of Wisconsin Last Time Terrain Dynamic LOD.
Computer Graphics: Programming, Problem Solving, and Visual Communication Steve Cunningham California State University Stanislaus and Grinnell College.
Real-time Rendering of Dynamic Vegetation Alexander Kusternig Vienna University Of Technology.
Basics of Rendering Pipeline Based Rendering –Objects in the scene are rendered in a sequence of steps that form the Rendering Pipeline. Ray-Tracing –A.
COMP 175: Computer Graphics March 24, 2015
3D Objects Subject:T0934 / Multimedia Programming Foundation Session:12 Tahun:2009 Versi:1/0.
02/26/02 (c) 2002 University of Wisconsin, CS 559 Last Time Canonical view pipeline Orthographic projection –There was an error in the matrix for taking.
Dynamic Meshing Using Adaptively Sampled Distance Fields
Week 2 - Wednesday CS361.
Computer Graphics World, View and Projection Matrices CO2409 Computer Graphics Week 8.
Rendering Forest Scenes in Real-Time EGSR 2004 Philippe Decaudin Fabrice Neyret GRAVIR/IMAG-INRIA, Grenoble, France.
Week 11 - Thursday.  What did we talk about last time?  Image processing  Blurring  Edge detection  Color correction  Tone mapping  Lens flare.
Buffers Textures and more Rendering Paul Taylor & Barry La Trobe University 2009.
VIS Group, University of Stuttgart Tutorial T4: Programmable Graphics Hardware for Interactive Visualization Adaptive Terrain Slicing (Stefan Roettger)
Real-Time Rendering SPEEDING UP RENDERING Lecture 04 Marina Gavrilova.
Computer Graphics and Multimedia Systems, University of Siegen, Germany 1 GPU-Based Responsive Grass Jens Orthmann, Christof Rezk-Salama, Andreas Kolb.
Image-based Rendering. © 2002 James K. Hahn2 Image-based Rendering Usually based on 2-D imagesUsually based on 2-D images Pre-calculationPre-calculation.
Image Synthesis Rabie A. Ramadan, PhD 1. 2 About my self Rabie A. Ramadan My website and publications
Geometric Modeling. Volumetric o Collection device obtains regular grid of measurement values Examples: CT, MRI, PET, Ultrasound o Values are interpreted/visualized.
Computer Graphics The Rendering Pipeline - Review CO2409 Computer Graphics Week 15.
Pattern Based Procedural Textures Sylvain Lefebvre Fabrice Neyret iMAGIS - GRAVIR / IMAG - INRIA
1 Real-time visualization of large detailed volumes on GPU Cyril Crassin, Fabrice Neyret, Sylvain Lefebvre INRIA Rhône-Alpes / Grenoble Universities Interactive.
Real-time Graphics for VR Chapter 23. What is it about? In this part of the course we will look at how to render images given the constrains of VR: –we.
Game Programming 06 The Rendering Engine
Computer Graphics 2 Lecture 7: Texture Mapping Benjamin Mora 1 University of Wales Swansea Pr. Min Chen Dr. Benjamin Mora.
CHAPTER 8 Color and Texture Mapping © 2008 Cengage Learning EMEA.
Games Development 1 Camera Projection / Picking CO3301 Week 8.
09/16/03CS679 - Fall Copyright Univ. of Wisconsin Last Time Environment mapping Light mapping Project Goals for Stage 1.
Art-based Rendering with Continuous Levels of Detail Lee Markosian, Barb Meier, Michael Kowalski, Loring Holden, J. D. Northrup, and John Hughes.
- Laboratoire d'InfoRmatique en Image et Systèmes d'information
Basic 3D Concepts. Overview 1.Coordinate systems 2.Transformations 3.Projection 4.Rasterization.
Advanced Computer Graphics Spring 2014 K. H. Ko School of Mechatronics Gwangju Institute of Science and Technology.
Review on Graphics Basics. Outline Polygon rendering pipeline Affine transformations Projective transformations Lighting and shading From vertices to.
Representation and modelling 3 – landscape specialisations 4.1 Introduction 4.2 Simple height field landscapes 4.3 Procedural modeling of landscapes- fractals.
Graphics Graphics Korea University cgvr.korea.ac.kr 1 7. Speed-up Techniques Presented by SooKyun Kim.
Graphics Interface 2009 The-Kiet Lu Kok-Lim Low Jianmin Zheng 1.
Applied Mathematics 1 Distributed visualization of terrain models How to get the whole world into a coffee mug... Rune Aasgaard.
1 Perception and VR MONT 104S, Fall 2008 Lecture 20 Computer Graphics and VR.
Painterly Rendering for Animation Introduction speaks of focus and detail –Small brush strokes focus and provide detail –Large strokes are abstract and.
1 CSCE 441: Computer Graphics Hidden Surface Removal Jinxiang Chai.
Rendering Large Models (in real time)
Graphics, Modeling, and Textures Computer Game Design and Development.
COMPUTER GRAPHICS CS 482 – FALL 2015 SEPTEMBER 10, 2015 TRIANGLE MESHES 3D MESHES MESH OPERATIONS.
Visualization and Interactive Systems Group, University of Stuttgart Page 1 Terrain Rendering Dirk Ringe (phenomic), Stefan Roettger (VIS)
APE'07 IV INTERNATIONAL CONFERENCE ON ADVANCES IN PRODUCTION ENGINEERING June 2007 Warsaw, Poland M. Nowakiewicz, J. Porter-Sobieraj Faculty of.
Graphics Pipeline Bringing it all together. Implementation The goal of computer graphics is to take the data out of computer memory and put it up on the.
1 Real-Time High-Quality View-dependent Texture Mapping using Per-Pixel Visibility Damien Porquet Jean-Michel Dischler Djamchid Ghazanfarpour MSI Laboratory,

- Introduction - Graphics Pipeline
CS475 3D Game Development Level Of Detail Nodes (LOD)
Graphics, Modeling, and Textures
Finite Element Surface-Based Stereo 3D Reconstruction
Run-Time LOD Run-time algorithms may use static or dynamic LOD models:
Presentation transcript:

NDVI-based Vegetation Rendering CGIM ‘07 Stefan Roettger, University of Erlangen

NDVI-based Vegetation Rendering Overview - Recap of Terrain Rendering with C-LOD - How to get vegetation information - NDVI mapping - Volumetric representation of vegetation - Rendering with texture splatting and billboards - Double-buffered tile set caching - Results

NDVI-based Vegetation Rendering Height Fields A common representation of terrain is the so called height field, a regular matrix which stores height values. By transforming the elevation into grey-values the height field is given by a grey-scale image (left). Color information is given by an ortho-image (right).

NDVI-based Vegetation Rendering Rendering with Triangle Fans - the height field is decomposed into a quad tree - per quad tree node (dots) the decision to subdivide depends on the screen space error of the actual triangulation - each leave of the quad tree is rendered with a triangle fan (arrows) - to get a conforming mesh some vertices are skipped (crosses)

NDVI-based Vegetation Rendering C-LOD - screen-space error depends on distance to the eye point and the “surface roughness” - user-definable upper bound on the screen-space error continuously delimits the resolution of the triangulation low medium high

NDVI-based Vegetation Rendering Textured Terrain - an orthographic projection of the terrain texture yields the following visualization of the height field - geo-morphing of the vertices is used to suppress the popping effect

NDVI-based Vegetation Rendering Vegetation Information Left: Landsat ortho-image (pseudo-color) Middle: NDVI Index (automatic) Right: Corine Landcover Classification (hand-processed)

NDVI-based Vegetation Rendering NDVI Definition - Landsat has channels 1-8 for visible and invisible wave lengths = BGR, but not exactly, wave lengths are shifted towards blue - 4 = NIR (Near Infra-Red) Channels I 1-3

NDVI-based Vegetation Rendering NDVI Mapping Vegetation height and type are assumed to correlate with the NDVI via a monotone mapping: For NDVI=0 no vegetation exists, above a specific threshold the vegetation changes from grass to bushes or trees.

NDVI-based Vegetation Rendering Volumetric Representation We interprete the NDVI as a second height field that encodes H Plant and compute the triangulation based on the weighted error of both height fields. Then we take each base triangle of the mesh and stack a prism onto it where the height corresponds to the mapped NDVI. Now we can randomly place vegetation inside the prism.

NDVI-based Vegetation Rendering Texture Splatting If plant type is grass, e.g. the height is below a specific threshold, but not zero, we use texture splatting to show ground detail. A stack of grass textures is put into a 3D texture which is indexed by the mapped prism height. Then the lookup is perturbed with Perlin Noise, faded by distance and finally blended with the ortho-image. For bushes and trees the following billboard textures are scaled to fit the height of the prism:

NDVI-based Vegetation Rendering Paging and Caching The scene is divided into a regular tile set. All visible tiles are paged in and out depending on the LOD of each tile to minimize the [texture] memory footprint. For each frame only a subset of all the tiles is triangulated and the resulting part of the entire mesh is appended to a double-buffered vertex cache. While the back buffer of the cache is updated from frame to frame, the front buffer can be reused for a couple of subsequently rendered frames. Accordingly, for a single frame, only a subset of the plant geometry is updated and stored in the cache which leads to a much reduced CPU load since the vertex buffer can be redrawn with little overhead.

NDVI-based Vegetation Rendering Limiting Geometric Complexity Geo-morphing of both the terrain and the prisms allows for smooth transitions of the prisms and therefore the generated plant distribution. Prisms that are far away fade out to zero height and thus do not contain vegetation. As a result, the appearance of distant details is determined only by the ortho-image and plants are only generated in the vicinity of the viewer. This limits the overall number of geometric plant primitives to an amount that is easily managable by today’s graphics hardware. The total number of visible trees in a forest easily exceeds 50 million while the number of rendered trees in the vertex cache is only between 300k and a few million for densely vegetated areas.

NDVI-based Vegetation Rendering Results #1 We use green triangles as a simplified model for the plants in order to show the match of the NDVI (left) with the displayed plants (right).

NDVI-based Vegetation Rendering Results #2 The demo scene with and without vegetation rendering. Frame rate is about fps on a NVIDIA GeForce FX Here comes a live demo…

NDVI-based Vegetation Rendering CGIM ‘07 Thank you!