# 3D surface fitting to N random points
# using inverse distance weighted averages.
# FB - 201003162
from PIL import Image
import random
import math
# image size
imgx = 512
imgy = 512
image = Image.new("RGB", (imgx, imgy))
# random color palette coefficients
kr = random.randint(1, 7)
kg = random.randint(1, 7)
kb = random.randint(1, 7)
ir = 2**kr
ig = 2**kg
ib = 2**kb
jr = 2**(8-kr)
jg = 2**(8-kg)
jb = 2**(8-kb)
# select n random points
n=random.randint(5, 50)
arx=[]
ary=[]
arz=[]
for i in range(n):
arx.append(random.randint(0, imgx-1))
ary.append(random.randint(0, imgy-1))
arz.append(random.randint(0, 255))
for y in range(imgy):
for x in range(imgx):
flag=False
sumv=0.0
sumw=0.0
for i in range(n):
dx=x-arx[i]
dy=y-ary[i]
if(dx==0 and dy==0):
flag=True
z=arz[i]
break
else:
# wgh=1.0/math.pow(math.sqrt(dx*dx+dy*dy),1.0) # linear
wgh=1.0/math.pow(math.sqrt(dx*dx+dy*dy),2.0) # quadratic
# wgh=1.0/math.pow(math.sqrt(dx*dx+dy*dy),3.0) # cubic
sumw+=wgh
sumv+=(wgh*arz[i])
if flag==False:
z=int(sumv/sumw)
# z to RGB
r = z % ir * jr
g = z % ig * jg
b = z % ib * jb
image.putpixel((x, y), b * 65536 + g * 256 + r)
image.save("rndSurface.png", "PNG")
