Data Visualization. Fall 2017

Transcription

1 Data Visualization Fall 2017

2 Vector Fields Vector field v: D R n D is typically 2D planar surface or 2D surface embedded in 3D n = 2 fields tangent to 2D surface n = 3 volumetric fields When visualizing vector fields, we have to: map D to 2D screen (like with scalar fields) Then we have 1 pixel for 2 or 3 scalar values (only 1 for scalar fields) Fall 2017 Data Visualization 2

3 Vector Fields A wind tunnel model of a Cessna 182 showing a wingtip vortex Source: Fall 2017 Data Visualization 3

4 Vector Fields We can compute derived scalar quantities from vector fields and use already known methods for these: Divergence (R n R) degree of field s convergence or divergence v x v v z x y z div v = + + or equivalently div v = y lim Γ 0 1 Γ Γ ( v n ) Γ ds Fall 2017 Data Visualization 4

5 Vector Fields Divergence (R n R) Source: Eric Shaffer, Vector Field Visualization Fall 2017 Data Visualization 5

6 Vector Fields Fall 2017 Data Visualization 6 Curl (rotor, vorticity) (R n R n ) rot v = or equivalently rot v = rot v is axis of the rotation and its magnitude is magnitude of rotation y v x v x v z v z v y v x y z x y z,, Γ Γ Γ ds v 1 lim 0

7 Vector Fields Curl(rotor, vorticity) (R n R n ) 2D fluid flow simulated by NSE and visualized using vorticity Fall 2017 Source: A Simple Fluid Solver based on the FFT, J. Stam, J. of Graphics Tools 6(2), 2001, Data Visualization 7

8 Approximating Derivatives Analytical functions Ideally we are given a function in a closed form and we can (probably) compute previous derivatives analytically Sometimes the given function is too complicated We compute the derivatives numerically via differentiation Discretely sampled data The best approach is to fit a function to the data and compute analytical derivatives Numerical evaluation may be fater than analytical Fall 2017 Data Visualization 8

9 Approximating Derivatives Central-difference furmulas of Order O(h 2 ): Fall 2017 Data Visualization 9

10 Approximating Derivatives Central-difference furmulas of Order O(h 4 ): Fall 2017 Data Visualization 10

11 Numerical Integration of ODE Initial value problem: y unknown function dt integration time step y (t) = f(t, y(t)) Fast but not very accurate Higher methods available Fall 2017 Data Visualization 11

12 Numerical Integration of ODE First order Euler method: x(t + dt) =x(t) +v(x(t))dt x represents spatial position v vector field dt integration time step Fast but not very accurate Higher methods available Fall 2017 Data Visualization 12

13 Numerical Integration of ODE Second order Runge-Kutta method: x(t + dt) =x(t) + ½ (K 1 + K 2 ) where K 1 = v(x(t))dt K 2 = v(x(t) + K 1 )dt Fall 2017 Data Visualization 13

14 Numerical Integration of ODE Fourth order Runge-Kutta method (aka RK4): x(t + dt) =x(t) + 1/6 (K 1 + 2K 2 + 2K 3 + K 4 ) where K 1 = v(x(t))dt K 2 = v(x(t) + ½ K 1 )dt K 3 = v(x(t) + ½ K 2 )dt K 4 = v(x(t) + K 3 )dt Fall 2017 Data Visualization 14

15 Glyphs Icons or signs for visualizing vector fields Placed by (sub)sampling the dataset domain Attributes (scale, color, orientation) map vector data at sample points Simplest glyph: Line segment (hedgehog plots) Line from x to x + kv Optionally color map v Fall 2017 Data Visualization 15

16 Glyphs Simplest glyph: Line segment (hedgehog plots) Source: Eric Shaffer, Vector Field Visualization Fall 2017 Data Visualization 16

17 Glyphs Another variant: 3D cone and 3D arrow glyphs Shows orientation better than lines Use shading to separate overlapping glyphs Source: Eric Shaffer, Vector Field Visualization Fall 2017 Data Visualization 17

18 Glyphs Related Problems No interpolation in glyph space (unlike for scalar plots) A glyph take more space than a pixel (clutter) Complicated visual interpolation of arrows Scalar plots are dense, glyph plots are sparse Glyph positioning is important: Uniform grid Rotated grid Random samples (generally best solution) Fall 2017 Data Visualization 18

19 Characteristic Lines Important approaches of characteristic lines in a vector field: Stream line curve everywhere tangential to the instantaneous vector (velocity) field (time independent vector field) Local technique initiated from one or a few particles Trace of the direction of the flow at one single time step Curves can look different at each time for unsteady flows Represents the direction of the flow at a given fixed time Computationally expensive The scene becomes cluttered when the number of streamlines is increased Fall 2017 Data Visualization 19

20 Characteristic Lines Extensions of stream lines for time-varying data follows: Path lines trajectories of massless particles in the flow Trace of a single particle in continuous time Time dependent vector field Fall 2017 Data Visualization 20

21 Characteristic Lines Streak lines trace of dye that is released into the flow at a fixed position Curve that connects the position of all the particles which are initiated at the certain point but at different times Time lines propagation of a line of massless elements in time In steady flow a path line = streak line = stream line Fall 2017 Data Visualization 21

22 Further Reading Fall 2017 Data Visualization 22