Boid Watching

[bplus]

OK here is a step or two further down from complete restart:

' Boids Restart b+ 2022-03-10

' >>>>>>>>>>>>>>>>>>>>>  Escape Key will now call it Quits!

global H$, xmax, ymax, pi, deg, rad, nb, speed, done
H$ = "gr"
xmax = 1200 '<== actual drawing space desired
ymax = 700 '<=== actual drawing space desired
pi = acs(-1)
deg = 180 / pi ' radian 2 degree mult
rad = pi / 180 ' degree 2 radian mult
nb = 100 ' number of birds
speed = 5
done = 0

dim bx(nb), by(nb), ba(nb), oldx(nb), oldy(nb)
for i = 1 to nb
   bx(i) = rand(100, xmax - 100) ' start random screen x, y away from borders
   by(i) = rand(100, ymax - 100)
   ba(i) = 2 * pi * rnd(0)       ' random headings
next

headMode = 1    ' on / off
sway = pi/6     'just turn neighbor towards neighbor
hf = .05         'heading factor how strong an influence  0 to 1

centerMode = 1  ' on / off
cf = .001        'centering factor how strong a pull from 0 to 1  .01 is week .1 pretty strong!

nomainwin

WindowWidth = xmax + 8
WindowHeight = ymax + 32
UpperLeftX = 100
UpperLeftY = 30

open "Boids Restart" for graphics_nsb_nf as #gr  '<======================= title
#gr "setfocus"
#gr "trapclose quit"
#gr "when leftButtonUp lButtonUp"
#gr "when characterInput charIn"
#gr "down"


while done = 0
   scan
   for i = 1 to nb 'draw then update positions
       'erase old
       #gr "color white"
       #gr "backcolor white"
       #gr "place ";oldx(i);" ";oldy(i);"; circlefilled ";2
       ' draw current
       #gr "color black"
       #gr "backcolor black"
       #gr "place ";bx(i);" ";by(i);"; circlefilled ";2
       oldx(i) = bx(i) : oldy(i) = by(i)


       [test]
       dx = speed * cos(ba(i)) : dy = speed * sin(ba(i))
       testx = bx(i) + dx : testy = by(i) + dy

       ' fix ba() for comparing
       if ba(i) < 0 then ba(i) = ba(i) + 2 * pi
       if ba(i) >= 2 * pi then ba(i) = ba(i) - 2 * pi

       'inbounds?
       if testx <= 0 then  'they seem to graviatate to left edge
           bx(i) = xmax + testx - rand(0, 15) : by(i) = rand(100, ymax - 100): ba(i) = pi + rnd(0) * pi/2 * rMinus()
       else

           if testx > 0 and testx < xmax and testy > 0 and testy < ymax then ' OK
               bx(i) = testx : by(i) = testy
           else
               ba(i) = ba(i) + rnd(0) * 2 * pi - pi  : goto [test]
           end if
       end if

       for j = i + 1 to nb
           scan
           dist = distance(bx(i), by(i), bx(j), by(j))
           if dist < 50 then  ' birds are close enough to influence each other by visual

               'sway the neighbors headings towards each other
               if headMode then
                   if ba(i) > ba(j) then   ' this is congregating all the birds to the left edge and bottom
                       ba(i) = ba(i) - sway * hf
                       ba(j) = ba(j) + sway * hf
                   else
                       ba(i) = ba(i) + sway * hf
                       ba(j) = ba(j) - sway * hf
                   end if

                   if ba(i) < 0 then ba(i) = ba(i) + 2 * pi
                   if ba(i) >= 2 * pi then ba(i) = ba(i) - 2 * pi
                   if ba(j) < 0 then ba(j) = ba(j) + 2 * pi
                   if ba(j) >= 2 * pi then ba(j) = ba(j) - 2 * pi
               end if
           end if

           if dist > 20 and dist < 100 then

               'stickiness stay close to neighbors, close distance between
               if centerMode then
                   if bx(i) > bx(j) then
                       bx(i) = bx(i) - cf * (bx(i) - bx(j))
                       bx(j) = bx(j) + cf * (bx(i) - bx(j))
                   else
                       bx(i) = bx(i) + cf * (bx(j) - bx(i))
                       bx(j) = bx(j) - cf * (bx(j) - bx(i))
                   end if
                   if by(i) > by(j) then
                       by(i) = by(i) - cf * (by(i) - by(j))
                       by(j) = by(j) + cf * (by(i) - by(j))
                   else
                       by(i) = by(i) + cf * (by(j) - by(i))
                       by(j) = by(j) - cf * (by(j) - by(i))
                   end if
               end if
           end if
       next 'j
   next ' i
   ' call pause 200  EDIT: This is not needed for this code
wend
wait

sub lButtonUp H$, mx, my  'must have handle and mouse x,y

end sub

sub charIn H$, c$
   if asc(c$) = 27 then call quit H$

end sub

'Need line: #gr "trapclose quit"
sub quit H$
   close #H$ '<=== this needs Global H$ = "gr"
   end       'Thanks Facundo, close graphic wo error
end sub

sub pause mil   'tsh version has scan built-in
   t0 = time$("ms")
   while time$("ms") < t0 + mil : scan : wend
end sub

Function Atan2(y,x)
   'Atan2 is a function which determines the angle between points
   'x1, y1 and x2, y2. The angle returned is in radians
   'The angle returned is always in the range of
   '-pi to pi radians (-180 to 180 degrees)
   '==============================================================
   'NOTE the position of Y and X arguments
   'This keeps Atan2 function same as other language versions
   '==============================================================
  If x = 0 Then
      If y < 0 Then
           Atan2 = -1.5707963267948967
       Else
           Atan2 = 1.5707963267948967
       End If
  Else
      chk = atn(y/x)
      If x < 0 Then
          If y < 0 Then
               chk = chk - 3.1415926535897932
           Else
               chk = chk + 3.1415926535897932
           End If
      End If
       Atan2 = chk
  End If
  'thanks Andy Amaya
End Function

function rand(lo, hi)
   rand = int((hi - lo + 1) * rnd(0)) + lo
end function

function rMinus()  ' plus or minus 1 multiplier
   if rnd(0) < .5 then rMinus = -1 else rMinus = 1
end function

function distance(x1, y1, x2, y2)
   distance = ( (x1 - x2) ^ 2 + (y1 - y2) ^ 2) ^ .5
end function



Enough code to see flocking action of dots from Random Start:

[bplus]

To Observable Flocking Action

[Rod]

Cool, I can't pretend I understand everything you are doing just yet. I have been thinking about speed. Sadly adding sprites to my old code did not make much difference. But looking again at my old code I see that I compare all boids against all others, so n^n computations. Many unnecessary. I started skipping boids too far apart but that failed when they all flocked together.

Now B+ reduced the second loop by only checking remaining boids. for j= I+1 to boids. That's a good start and I have put that in this code. I also added the sway but probably not in a way B+ expected. However we have flocking at a reasonable speed. My flocking code is poor but I am still at the stage of ensuring speed.

What I plan next is to sort the boids into x order then I will be able to get to groups of boids quickly and with less looping. Only then can I start proper separation, alignment and cohesion.

But this hints at animation success.

pi = acs(-1)
sway = pi/6
b=50
dim b(b,4) 'x,y,dx,dy
   x=1
   y=2
   dx=3
   dy=4
   for n= 1 to b
       b(n,x)=int(rnd(0)*800)
       b(n,y)=int(rnd(0)*600)
       b(n,dx)=8-rnd(0)*16
       b(n,dy)=8-rnd(0)*16
   next

   'nomainwin
   WindowWidth =800
   WindowHeight =600
   UpperLeftX = int((DisplayWidth-WindowWidth)/2)
   UpperLeftY = int((DisplayHeight-WindowHeight)/2)
   open "Boids" for graphics_nsb as #1
   #1 "down ; fill black ; backcolor cyan ; place 5 14; circlefilled 4 ; backcolor white ; place 5 4 ; circlefilled 4 ; getbmp boid 0 0 10 20"
   #1 "down ; fill black ; color white ; trapclose [quit]"
   #1 "getbmp bac 0 0 800 600 ; background bac"
   for n= 1 to b
       #1 "addsprite b";n;" boid"
       #1 "spritexy b";n;" ";b(n,x);" ";b(n,y)
   next

   [draw]
   #1 "drawsprites"
   for n=1 to b

       'find all near boids to be influenced by
       for m=n+1 to b
           if n<>m and abs(b(n,x)-b(m,x))<25 then
               dist=sqr((b(m,x)-b(n,x))*(b(m,x)-b(n,x))+(b(m,y)-b(n,y))*(b(m,y)-b(n,y)))
               if dist <20 and dist>5 then 'influence
                   b(n,dx)=(b(n,dx)+b(m,dx))/2
                   b(n,dy)=(b(n,dy)+b(m,dy))/2
               end if
               if dist <5 then 'too close
                   b(n,dx)=b(n,dx)*-1
                   b(n,dy)=b(n,dy)*-1
               end if
           end if
       next

       'move boid
       b(n,dx)=b(n,dx)+sway
       b(n,dy)=b(n,dy)-sway
       b(n,x)=b(n,x)+b(n,dx)
       b(n,y)=b(n,y)+b(n,dy)
       if b(n,x)>=800 then b(n,dx)=b(n,dx)*-1
       if b(n,y)>=600 then b(n,dy)=b(n,dy)*-1
       if b(n,x)<0 then b(n,dx)=b(n,dx)*-1
       if b(n,y)<0 then b(n,dy)=b(n,dy)*-1
       #1 "spritexy b";n;"  ";b(n,x);" ";b(n,y)

   next
   scan
   goto [draw]


[quit]
   close #1
   end




[Rod]

@ B+ Something else has been bugging me, its your close routine.

'Need line: #gr "trapclose quit"
sub quit H$
   close #H$ '<=== this needs Global H$ = "gr"
   end       'Thanks Facundo, close graphic wo error
end sub


Not sure where this all came from. You don't need a global handle. Passing global variables as parameters is known to cause problems particularly if the variable name in the sub is the same as the global name. There are easy ways to force the error you were having. Not having an END statement in the Sub is one way, also having the trapclose attached to the GRAPHICBOX rather than the window is another.

But this works

sub quit h$
   close #h$
   end
end sub


The trap close event will call the sub and pass the handle #gr If you want to use Esc then just call the close sub with a normal variable "#gr"

[bplus]

Oh I see I made Global variable same as variable parameter in SUBs, nope didn't mean to do that.

[Rod]

Yes, but global wasn’t needed either. Next will be your pause code, but in the interim keep dazzling us with graphics, I will simply wait for a timer event!

[bplus]

Mar 11, 2022 15:05:39 GMT -5 Rod said:

Yes, but global wasn’t needed either. Next will be your pause code, but in the interim keep dazzling us with graphics, I will simply wait for a timer event!

The global comes from template code I have for JB which has been pieced together over time so I can write up some JB code fairly quickly.

What if the timer event goes off before you are done with a massive set of calculations?

Will it wait until they are done?

Pause is never called until calculations are done and it's there only to reduce flicker, usually when clearing screen which this version doesn't need! now that we mention it. Yah, dump it!

[Rod]

Yeh I knew it was template code. A block of code will always get done, the program will only break away from a scan or wait statement. If the code takes longer than the timer event the event queues. It will get actioned as soon as the code reaches the scan or wait.

That isn’t good. If the timer fires too quickly events stack up. That’s why programs can appear to become unresponsive, say to the close event. What is happening is the queued timer events are getting actioned ahead of the queued trap close event and the program seems to be ignoring you.

Also the while wend pause is not good because it ties the program into a dumb loop. You can imagine that other queuing events get ignored while you are pausing that way.

But it isn’t clear cut, sometimes loop scan go to loop is good sometimes timer x code wait is good if timed correctly. The benefit of the first is speed the benefit if the second is smooth paced animation.

The problem of flicker is harder to fix because it is not a timing issue. It is caused by windows randomly redrawing the screen when it has time. This may catch us out mid draw. So drawing quick and leaving it alone for a while works sometimes, over drawing mostly works.

The good news is that Lb5 will have a screen buffer that will fix this issue.

Hope you are learning a little more about Liberty from these comments.

[Walt Decker]

You can use double buffering in LB to eliminate flicker if you are willing to use API code.

[Rod]

Yes, I have often promoted the blitter but it falls on deaf ears:(

alycesrestaurant.com/lbpe/Fun%20with%20the%20blitter.html

[Walt Decker]

Seems like most LB users are deaf when one mentions API or 3rd party dlls. Why, I have no clue.

[Carl Gundel]

Mar 12, 2022 13:02:55 GMT -5 Walt Decker said:

Seems like most LB users are deaf when one mentions API or 3rd party dlls. Why, I have no clue.

I think it might be because most people think of BASIC is being a language for casual programming, and they gravitate to it because they don't won't to think about the nuts and bolts of things.  This discussion might deserve a thread of its own.

On the other hand, an easy language that also allows for less abstract tools such as CALLDLL (or in the old school, PEEK, POKE, SYS, etc.) can provide a ramp for the beginner to become more of an expert as curiosity or need demand.

[Rod]

Walt, you forget how much you know and how little others do. API ain't BASIC either so it is not a natural transition. For me it is a necessary evil. LB5 will focus the mind since it will support API less.

[Carl Gundel]

Mar 12, 2022 14:04:44 GMT -5 Rod said:

LB5 will focus the mind since it will support API less.

LB5 will certainly, at least initially be less capable with API control or widgets, but the widgets that come with it are more powerful.  My intent is that the general mechanisms of calling out to APIs and libraries will be improved.  Then later on (LB 5.1 or so) there will be once again native widgets (giving you more options), and the use of APIs will apply to them.

[Carl Gundel]

Mar 12, 2022 14:04:44 GMT -5 Rod said:

Walt, you forget how much you know and how little others do. API ain't BASIC either so it is not a natural transition. For me it is a necessary evil. 

This is how people tend to behave in all realms of human endeavor.  Walt is on the right hand side of the bell curve, the top percentile of Liberty BASIC programmers.  Almost everyone else is not going to dip their toes in the API and DLL space, even if they would like to.