I'm still learning ASM and trying to convert Basic programs i wrote, to ASM.
Heres one of my hilariuos Trig Formulas.
setup all the progs vars to single or double precision , radius of 125 is for 640X480
radian = ATN(1) / 45
radius = 125
FOR deg1 = 0 TO 360
c1 = COS(deg1 * radian)
s1 = SIN(deg1 * radian)
x1 = radius * c1
y1 = radius * s1
FOR deg2 = 0 TO 360 STEP 1
c2 = COS(deg2 * radian)
s2 = SIN(deg2 * radian)
x2 = radius * c2 * c2 * COS(COS(deg2 * s1 * c2 * radian) * c1) * c2 * c2 / c1 * s1 * COS(deg2 * radian * c1)
y2 = radius * s2 * s2 * SIN(COS(deg2 * c1 * s2 * radian) * s1) * s2 * s2 / s1 * s1 * SIN(deg2 * radian * s1)
PSET (xctr + x1 + x2 , yctr + y1 + y2), 9
NEXT
NEXT
try a Bresenham algo :t
http://en.wikipedia.org/wiki/Midpoint_circle_algorithm (http://en.wikipedia.org/wiki/Midpoint_circle_algorithm)
at any rate....
the FPU has intrinsic trig functions which should be reasonably fast
It's not going to be easy. The attachment contains the compiled FreeBASIC app.
looks interesting :biggrin:
Is this shape intended? Using my macros, the conversion is very easy - see attachment.
qWord,
I had previously downloaded your SmplMath package, and forgot to look at it. Now that I have I'm impressed, good job :t
The shape that your code produces does look somewhat like a cartoon version of a Mexican bandit:
http://fc02.deviantart.net/fs6/i/2005/030/f/4/Mexican_bandit_mascot_by_pas_designs.jpg (http://fc02.deviantart.net/fs6/i/2005/030/f/4/Mexican_bandit_mascot_by_pas_designs.jpg)
(http://fc02.deviantart.net/fs6/i/2005/030/f/4/Mexican_bandit_mascot_by_pas_designs.jpg)
Quote from: MichaelW on August 10, 2012, 01:32:32 AM
qWord,
I had previously downloaded your SmplMath package, and forgot to look at it. Now that I have I'm impressed, good job :t
I can echo that :t
Quote from: MichaelW on August 10, 2012, 01:32:32 AMI had previously downloaded your SmplMath package, and forgot to look at it. Now that I have I'm impressed, good job :t
thx :biggrin:
Quote from: MichaelW on August 10, 2012, 01:32:32 AM
The shape that your code produces does look somewhat like a cartoon version of a Mexican bandit:
that explains the name :lol:
@QWord
Your formula is not the correct height, its flat on the vertical.
ldl deg1 = 180 , radian = 0.01745329252 ; = Pi/180°
.while deg1 <= 360
ldl deg1=180 should be ldl deg1=0
ldl radius = 50 ; should be ~y_resolution / (somewhere betwen 5 to 7)
invoke BANDIT,ps.hdc,radius,150,150 ; the (150,150) should be x_resolution / 2 , y_resolution/2
i thought it was 1 radian = 180 degrees/pi ~ 57.3 degrees
Hi Dave,
Yes, a radian is an arc length on the circle equal to
the radius of the circle. There are two pi radians in the
circumference of a circle. So, about 57.3 degrees.
Cheers,
Steve N.
(http://l.yimg.com/us.yimg.com/i/mesg/emoticons7/18.gif)
In my trig formulas instead of looping -PI to PI or 0 to PI*2
I loop 0 to 360 degrees and multiply the degree by 1 degree worth or radians (.01745).
1 degree worth of radians is the ArcTangent of 1 (45 degrees) divided by 45.
I usually just call the "ATN(1)/45" a "rad" or "radian" or "r1 or r2" . but its not actually a radian, its 1 degree of radians.
and then do a
cosine = cos(deg1*rad) instead of a cosine = cos(theta)
Most everybody else uses radians or what they call theta.
loop 0 to PI*2 step theta (.01745)
c = cos(theta)
s = sin(theta)
I just use degrees , because i'm an untrained mathematician and probably a moron as well.
that is the fundamental flaw in your method
rather than stepping by degrees or radians....
step by pixels
for all the "orthogonals", it does not matter whether you step by X or by Y
the orthogonals are 0, 45, 90, 135, 180, 225, 270, 315, and 360
but - in between these angles, step by X if the slope of the tangent is between -1 and 1
otherwise, step by Y
you will be much happier with your circle :t
to expand on that a little.....
make 2 loops - one that steps by X, one that steps by Y
for 315 to 45 degrees, step by X
for 45 to 135 degrees, step by Y
for 135 to 225 degrees, step by X
for 225 to 315 degrees, step by Y
Hi,
To elaborate on Dave's comment; you should not step
by a degree because with a large circle you will be plotting
dots rather than a continuous circle. And with a small circle,
you will be plotting the pixels on the circle multiple times.
If you increment by pixels you end up with a continuous
line/circle. And you don't overplot pixels. This is one of
the ideas behind Bresenham's circle algorithm.
HTH,
Steve N.
yes - it will make a circle that can be filled :P
the main advantage of using Bresenham is that integer calculations are performed between steps
the circle algo was derived from the Bresenham Line algo
i have adapted the same idea to other things with good results
i happened to catch this photo on MSN
it is supposed to be a perseid meteor
but, it looks to me like Bresenham may have had a hand in it :lol:
(http://col.stb01.s-msn.com/i/E7/BF959D817C59911A59A248982D3CF.jpg)