Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
2
Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
3
Introduction 1/2 Domain : Real-time Rendering Context : Real-time rendering of complex meshes Problematic : Geometrical complexity (i.e. visual accuracy) versus computation time Our approach : hybrid- image based & geometry rendering methods The method : Combine a simplified mesh and reference images of the original mesh to speed-up rendering time
4
Introduction 2/2 Related work : View-dependent Texture Mapping [Debevec et al. 96,98]
►no blending of reference images ►no subdivisions to avoid blurring
VDTM
Appearance Preserving Simplification [Cohen et al. 98] ►no need to compute bump-map for each LOD
APS
Relief Mapping
[Oliveira et al. 00, Policarpo et al. 05]
►no parametrization ►relief not restricted to height field Relief Mapping
5
Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
6
Overview of the method The method consists in
Replacing a complex mesh laid out in a scene with a geometrically simplified version of it ... ... and map it with full relief stored in reference images
7
Overview of the method The method consists in
Replacing a complex mesh laid out in a scene with a geometrically simplified version of it ... ... and map it with full relief stored in reference images
Precomputation step
Given a complex object, compute its low polygon count version Grab viewpoints of the original mesh
Rendering step
Select best viewpoints
Map lost relief onto the simplified surface 8
Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
Texture resolution : 512x512 Data size without optimization : 2 Mb per viewpoint Optimization : 1 Mb/view using indexed colors and normals textures 11
Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
12
Real-time rendering 1/5 Two steps: 1. CPU : Selection of three best viewpoints For a given virtual eye position we select the three closest viewpoints among the reference viewpoints set => We consider that these three viewpoints covers all of the object surface
2. GPU : Determination of the best viewpoint We extract one pixel from each of the three selected viewpoints, and determine the best
13
Real-time rendering 2/5 Virtual viewpoint
Rendered triangle
C1
A1
A2 A
B1 C Reference viewpoint #1
Vertex shader
C2
B2
Reference viewpoint #2
B Simplified mesh
14
Real-time rendering 3/5 Per-pixel visibility determination Camera position
V2 O
V1 O1
O2
Simplified surface
P Real object surface P1
Pixel shader
P2
P'
15
Real-time rendering 4/5 Per-pixel visibility determination Camera position
Shadow is “deformed” on flat triangles because we reconstruct original relief
Simplified mesh = 2000 faces
23
Results Videos
24
Plan 1.Introduction 2.Overview of the method 3.Precomputation step 4.Real-time rendering 5.Results 6.Conclusion & future work
25
Conclusion & future work
This paper presented a technique for realtime rendering of complex 3D objects using view-dependent texture mapping approach We combine reference images of a mesh and geometrical simplification to speed-up rendering for a little lost of visual accuracy Mapping done in object space ➔
3D object to speed-up rendering at the cost of a little loss of visual accuracy. During a .... then, in order to compensate lost details, a bump texture represent-.
Mar 7, 2005 - They store the result of all possible ray intersection queries within a three-dimensional volume in a five-dimensional map indexed by three ...
... as virtual indexed and virtual tagged, which means that both the index and the tag are ..... [1] The RT Wiki: http://rt.wiki.kernel.org/- index.php/Main Page.
RealTime. SmartTag. ACTIVITE, RUMINATION & www.boumatic.fr. U n e tra ite confortable, rapide et com p lète. â¢. BouMatic. LOCALISATION DES VACHES ...
Keywords: Segmentation; Hierarchical; Colour; Cartography; Land cover. 1. Introduction ..... nan, 1991) are filters localized in space and frequency, which can be ...
however, our numerical study focuses on the white noise case,. i.e.,. , with the precision .... Examples of texture pairs with the same PSD, but different law for the.
1IPAL (Image & Pervasive Access Lab) - CNRS (French National Research Center), ... variations in diagnosis) but also for research applications (e.g., to understand the bio- ... [6] proposed fuzzy c-mean clustering algorithm along ... cal operations a
connected parts (patches) and of a piecewise 2D param- eterization (cf for example [16] ... The second improvement path is precisely pixel-wise .... otherwise. (6).
Aug 15, 2005 - naturally into the separation framework, producing separate layers as a by-product of the inpainting process. ... the geometric principle has been implemented through several different PDEs, aiming to get the most .... sparse solutions
The elasticity of a texture was also simulated by a deformation of the image ... assimilated to the amount of âpressureâ applied to the associated texture by .... The viscoelastic model presented previously can be easily implemented by means.
[4] to a parametric regression model using non linear basis functions. Kernel .... a kernel function, applied to δz and a basis vector denoted δâm z . Let W = (w1 ...
May 11, 2012 - and the Google cars. Google⢠= a set of services accessible from the web for free. Google Maps⢠and Google Earth⢠= Mapping and ...
Surveys for air pollutants as NOx, NO2, O3 and SO2 were conducted using passive ..... modeling: a new use of nanotechnology based solid state gas sensors.
an AND-OR-INVERT. ... the area calculations, or modify the delay cost and .... "Logic Decomposition Algorithms for the Time Optimization of Multi-Level Logic", p.
into the image based reference frame. Temporal matching of W can be seen as the estimation pk of the state system, for each new image of the video sequence.
Mar 22, 2000 - basic scheme is to build a grammar for the texture and then parse the ... If we express the frequency-domain coordinates in polar coordinates, ...
Feb 4, 2009 - Go to the Math function menu and select the *CDeconvolve operator. Set Source 1 .... â¢http://cp.literature.agilent.com/litweb/pdf/5989-3777EN.pdf.
and extended services provided by user-level managers. ... manager builds on the kernel facilities to provide additional process-level semantics, ..... Founded in 1980, QNX Software Systems is the industry leader in realtime, microkernel OS.
