Image Space Based Visualization of Unsteady Flow on
Image Space Based Visualization of Unsteady Flow on Surfaces Robert S. Laramee 1, Bruno Jobard 2 Helwig Hauser 1 1 VRVis Research Center, Austria, www. VRVis. at, {Laramee, Hauser}@VRVis. at 2 University of Pau, France, www. univ-pau. fr, bjobard@univ-pau. fr 1 Laramee@VRVis. at
Overview § goals § previous work § image space based vis of unsteady flow on surfaces § results § conclusions, future work 2 Laramee@VRVis. at
Our Goals & Motivation § application: vis of CFD simulation data § dense rep. of unsteady flow on surfaces § visualize flow on complex, adaptive resolution surfaces § user-interaction § fast § vis flow on dynamic meshes § no parameterization 3 Laramee@VRVis. at
Previous Work § Lagrangian. Eulerian Advection by Jobard et al. § path integration (Lagrangian) § update of color pixels (Eulerian) § Image Based Flow Visualization by Van Wijk § advection of images 4 Laramee@VRVis. at
Method Overview Vector Field Projection Compute Advection Mesh Image Advection Noise Blending Edge Blending Static Case Dynamic Case Edge Detection Image Overlay Application 5 Laramee@VRVis. at
Vector Field Encoding § Assign colors to the mesh vertices as a function of velocity § Colored image used as the simplified (view dependent) 3 D vector field Velocity Image § 3 D vectors are projected to image space transforming the computation from 3 D to 2 D § No more computation time spent on occluded polyons 6 Laramee@VRVis. at
Advection Mesh Computation and Boundary Treatment Euler approximation of a pathline (like IBFV) pk+1 = pk + vp(pk; t) dt Backward integration (like LEA) Pk-1 = pk - vp(pk-1; t) dt Advect Noise 7 Laramee@VRVis. at
Edge Detection and Blending Discontinuity Condition |zk-1 - zk | > e |pk-1 - pk| 8 Laramee@VRVis. at
Edge Detection and Blending § can also be used to prevent background color(s) from “leaking in” § edge detection enabled 9 Laramee@VRVis. at
Noise Injection and Blending § Why noise injection and blending? –for full coveratge § both spatial and temporal characteristics: linearly interpolated sequence of random values § Temporal characteristics: a black and white pulsing function 10 Laramee@VRVis. at
Image Overlay Application § final stage § computed once for each dynamic case § applied last § User controlled opacity 11 Laramee@VRVis. at
Putting Pieces Together Vector Field Projection Compute Advection Mesh Image Advection Noise Blending Edge Blending Static Case Dynamic Case Edge Detection Image Overlay Application 12 Laramee@VRVis. at
Texture Clipping § Texture clipped along edges to reduce artifacts § Example is exaggerated for exposition 13 Laramee@VRVis. at
Results, Large, Complex Data Sets 14 Laramee@VRVis. at
Results, Large, Complex Data Sets 15 Laramee@VRVis. at
Results, Zooming 16 Laramee@VRVis. at
Results, Medical Simulation Data 17 Laramee@VRVis. at
Results, Performance Times # mesh polygons 10 K 48 K 221 K § § § % image space 75 74 84 Not the same as in paper! Nvidia 980 XGL Quadro 2. 8 GHz dual processor Advection mesh res. FPS static FPS dynamic 32^2 64 64 40 35 128^2 64 18 256^2 512^2 32^2 64^2 32 15 -16 64 64 8 2. 3 13 13 128^2 64 10 -11 256^2 512^2 32^2 64^2 32 15 64 64 6 2 5 4 128^2 63 -64 4 256^2 512^2 32 13 2. 9 1. 5 18 Laramee@VRVis. at
Results, time-dependent mesh geometry and topology 19 Laramee@VRVis. at
Conclusions and Future Work We presented an algorithm with the following properties: § dense representation of unsteady flow on surfaces § vis flow on large surfaces, independent of surface’s complexity and resolution § supports user-interaction § fast § vis flow on dynamic meshes § does not rely on parameterization Future work § extension to unsteady, 3 D flow § application of more specialized HW, e. g. programmable per-pixel operations 20 Laramee@VRVis. at
Acknowledgements Thanks to: § § § Helmut Doleisch Tom Laramee Michael Mayer Jürgen Schneider Jarke van Wijk Austrian governmental research program, Kplus (www. kplus. at) § AVL (www. avl. com) Many result animations available at: http: //www. VRVis. at/ar 3/pr 2/vis 03/ 21 Laramee@VRVis. at
- Slides: 21