NewsForumsCodeLogsGallerySpecsSearch

Messed up polar coordinates

BlitzMax Forums/BlitzMax Programming/Messed up polar coordinates

smilertoo(Posted 2005) [#1]
Im trying to use polar coordinates to make an object orbit around a central point, but the object keeps drifing in towards the centre...as far as i can see the code is fine and the object should remain static until i have altered T variable.
Can anyone see whats wrong? (all values are double)




For Local PingObject:TPing=EachIn PingObject_List
X=PingObject.getX()
Y=PingObject.getY()

X1=X-384
Y1=Y-384

R=Sqr((X1*X1)+(Y1*Y1))
T=ATan2(Y1,X1)

X1=R*Cos(T)
Y1=R*Sin(T)

X=X1+384
Y=Y1+384

PingObject.setXY(X, Y)
Next


TomToad(Posted 2005) [#2]
Possibly rounding errors. Cos and Sin return numbers at very high decimal places and the computer needs to truncate at a certain place .I believe doubles truncate at 32 places, not 100% sure, but it does truncate the number resulting in small errors that can accumulate over time.
What I do when using polar coordinates in a game is I always use polar on that object throughout the game and only convert to x,y for rendering. If you must use X,Y (a.k.a. cartesian coordiantes), then make R a constant, possibly a Distance field in TPing. then you replace
R=Sqr((X1*X1)+(Y1*Y1))

with
R=PingObject.getDistance()

That way the distance never changes until you change it.


smilertoo(Posted 2005) [#3]
The issue seems to be that it gets confused when the angle is 90 or -90


TomToad(Posted 2005) [#4]
I still think you have a rounding issue somewhere. I tried some experimenting and 90 and -90 degrees are converted correctly. I tried incorporating your routine above into a test program, and everything behaves as expected. Are you sure that x and y within the TPing type are double? Here's the test program I made.