FabricPlain0059_5_thumbhuge_128

Click on the images in the 'Parameter estimation' and 'Procedural noise' columns (the images with a blue border) for more details about the parameter estimation and procedural evaluation.

Exemplar Gaussian
exemplar
Parameter
estimation1

Procedural
noise2

Grayscale kappa=1/256 N=1024
G=146, B=6
S=.89%
T=60.37s
FPS=32.0339
Power
spectrum
estimate
Power
spectrum3

Exemplar Gaussian
exemplar
Parameter
estimation
1
Procedural
noise
2
Color kappa=1/256 N=1024
G=145, B=6
S=.88%
T=56.67s
FPS=33.4815

1 kappa: relative regularization parameter; G: number of Gaussians; B: number of bandwidts; S: sparseness; T: time
2 N: impulse budget; FPS: frames per second
3 The power spectrum in the 'Procedural noise' column is a power spectrum estimate obtained by averaging 100 periodograms. We use this to validate our procedural evaluation.