Convex hull, quick algorithm - Marcus - 05-02-2024
I need to calculate and display the convex hull of 2D polygons in the editor for that 3d thing, so I implemented a quick (as in "divide and conquer") algorithm for it. I remember writing a quick algorithm for the convex hull of a set of random points in 3D at the university, which is a bit trickier.
If you ever need to calculate the convex hull for a polygon or a bunch of points, you can use this
Code: ' Convex hull test
' ----------------
set window "test", 640, 480
set redraw off
do
set color 0, 0, 0
cls
' Create 100 random points.
points = []
for i = 1 to 100
a = rnd()*2*PI
r = rnd(200)
points[sizeof(points)] = 320 + cos(a)*r
points[sizeof(points)] = 240 + sin(a)*r
next
' Draw points.
set color 255, 255, 255
pcount = sizeof(points)/2
for i = 0 to pcount - 1
draw pixel points[i*2], points[i*2 + 1]
next
' Display and wait one second.
redraw
wait 1000
' Calculate convex hull and display it.
hull = ConvexHull(points)
draw poly hull
set caret width(primary)/2, 4
center "Press space bar to continue ..."
redraw
while not keydown(KEY_SPACE, true) fwait 60
loop
' ConvexHull
' ----------
' Calculate the convex hull for a set of points in an array [x0, y0, x1, y1 ..], return as a polygon
' in the same format.
function ConvexHull(points)
pcount = sizeof(points)/2
xminIndex = 0; xmaxIndex = 0
for i = 1 to pcount - 1
x = points[i*2]
if x < points[xminIndex*2] xminIndex = i
if x > points[xmaxIndex*2] xmaxIndex = i
next
lpoints = []; rpoints = []
for i = 0 to pcount - 1
if i <> xminIndex and i <> xmaxIndex
cp = CrossProd(points[xminIndex*2], points[xminIndex*2 + 1],
points[xmaxIndex*2], points[xmaxIndex*2 + 1],
points[i*2], points[i*2 + 1])
if cp < 0
lpoints[sizeof(lpoints)] = points[i*2]
lpoints[sizeof(lpoints)] = points[i*2 + 1]
elseif cp > 0
rpoints[sizeof(rpoints)] = points[i*2]
rpoints[sizeof(rpoints)] = points[i*2 + 1]
endif
endif
next
hull = []
hull[sizeof(hull)] = points[xminIndex*2]
hull[sizeof(hull)] = points[xminIndex*2 + 1]
HullRec(hull, lpoints, points[xminIndex*2], points[xminIndex*2 + 1], points[xmaxIndex*2], points[xmaxIndex*2 + 1])
hull[sizeof(hull)] = points[xmaxIndex*2]
hull[sizeof(hull)] = points[xmaxIndex*2 + 1]
HullRec(hull, rpoints, points[xmaxIndex*2], points[xmaxIndex*2 + 1], points[xminIndex*2], points[xminIndex*2 + 1])
return hull
function HullRec(hull, points, x0, y0, x1, y1)
if not sizeof(points) return
pcount = sizeof(points)/2
maxd = 0; maxdIndex = -1
for i = 0 to pcount - 1
x = points[i*2]; y = points[i*2 + 1]
d = LinePointDist(x0, y0, x1, y1, x, y)
if d > maxd
maxd = d
maxdIndex = i
endif
next
npoints = []
for i = 0 to pcount - 1
if i <> maxdIndex
x = points[i*2]; y = points[i*2 + 1]
cp = CrossProd(x0, y0, points[maxdIndex*2], points[maxdIndex*2 + 1], x, y)
if cp < 0
npoints[sizeof(npoints)] = x
npoints[sizeof(npoints)] = y
endif
endif
next
if sizeof(npoints)
HullRec(hull, npoints, x0, y0, points[maxdIndex*2], points[maxdIndex*2 + 1])
endif
hull[sizeof(hull)] = points[maxdIndex*2]
hull[sizeof(hull)] = points[maxdIndex*2 + 1]
npoints = []
for i = 0 to pcount - 1
if i <> maxdIndex
x = points[i*2]; y = points[i*2 + 1]
cp = CrossProd(points[maxdIndex*2], points[maxdIndex*2 + 1], x1, y1, x, y)
if cp < 0
npoints[sizeof(npoints)] = x
npoints[sizeof(npoints)] = y
endif
endif
next
if sizeof(npoints)
HullRec(hull, npoints, points[maxdIndex*2], points[maxdIndex*2 + 1], x1, y1)
endif
endfunc
function CrossProd(ax, ay, bx, by, cx, cy)
return (bx - ax)*(cy - ay) - (by - ay)*(cx - ax)
endfunc
function LinePointDist(x0, y0, x1, y1, px, py)
dx = x1 - x0; dy = y1 - y0; d = dx*dx + dy*dy
if d <= 0 return 0
return |(x1 - x0)*(py - y0) - (px - x0)*(y1 - y0)|/sqr(d)
endfunc
endfunc
RE: Convex hull, quick algorithm - 1micha.elok - 05-03-2024
CONVEX HULL
click image to zoom in
Convex Hull is very useful.
It can be used in 3D graphics, which you are working now ... and it can be used also to simulate football in 2D game.
RE: Convex hull, quick algorithm - Marcus - 05-03-2024
(05-03-2024, 05:18 AM)1micha.elok Wrote: CONVEX HULL
click image to zoom in
Convex Hull is very useful.
It can be used in 3D graphics, which you are working now ... and it can be used also to simulate football in 2D game.
I didn't know it was used in soccer, haha! But then, my interest in sports happens to be pretty much equal to zero
I need it in the editor because portals/doorways between sectors/rooms must only exist on the convex hull of the sectors, else you're doomed to get rendering errors (like seeing other sectors through walls).
RE: Convex hull, quick algorithm - johnno56 - 05-03-2024
Doomed! Doom! Great idea... Time to play!
RE: Convex hull, quick algorithm - Marcus - 05-04-2024
(05-03-2024, 08:28 AM)johnno56 Wrote: Doomed! Doom! Great idea... Time to play!
Just curious, what version of doom do you prefer?
RE: Convex hull, quick algorithm - johnno56 - 05-04-2024
My favourite is classic... out of the box Doom and Doom2... Of course I have many, many mods and total conversions... I use GZDoom as my front-end...
RE: Convex hull, quick algorithm - Marcus - 05-04-2024
(05-04-2024, 12:25 PM)johnno56 Wrote: My favourite is classic... out of the box Doom and Doom2... Of course I have many, many mods and total conversions... I use GZDoom as my front-end...
Any way to play the doom 64 levels (originally on the N64 console) on that? They're very different from the original doom, creepier somehow, unsettling.
RE: Convex hull, quick algorithm - johnno56 - 05-04-2024
Doom 64 is one of the files I have... As I had never played the N64 version I cannot compare Doom 64 with N64... I also have "creepy" doom files... Not my "cup of tea" but good for a chuckle... lol
|