readme.txt

For PD close loop control. Used in tracing trajectory or simulation.

neardd
cbloom
Dec 2007

Esc = exit
+/- = change # of NN
space = new random seed points
# keys = select mode #


Modes are :

0=	"points",
1=	"voronoi",
2=	"tris",
3=	"bary",
4=	"knn average",
5=	"knn fit",
6=	"all average",
7=	"all fit"


Demonstrates kNN and kNN-local regression


In the pref file "myconfig.txt" :
(autoreloads if you edit the text file)

	float lsqr_damping; // damping value for lsqr regression fit

	// toggles for different weight equations : these can be combined if you turn on several

	bool weight_linear;
	bool weight_inverse;
	bool weight_cubecube;
	bool weight_exp;
	bool weight_expsqr;
	bool weight_coslerp;

	// scales for the exponential weights :

	float expsqrScale;
	float expScale;


exponential weights don't really work with kNN , you have to use all neighbors


Note that for N sees, N neighbors is a bit different than the "all neighbors" mode, because
with k neighbors I'm actually only fitting on (k-1) seeds.