lochoi

i have to images that i mix on one same output with jit.alphablend.

they appear like separate fragments on a black screen. with random object i 

i need to know when the fragments meet. if i receive a bang at that moment i 

tell the fragments to stop moving and they stay connected at the position 

the project uses 6 to 9 fragments that move around randomly and when a 

fragment meets another they both stop until all fragments have met another 

i was thinking....each fragment having an alpha layer , is there an object 

that would recognise there are 6 or X alpha layers and when which meets 

hello,
i have to images that i mix on one same output with jit.alphablend.
they appear like separate fragments on a black screen. with random object i 
make the shapes move around.
i need to know when the fragments meet. if i receive a bang at that moment i 
tell the fragments to stop moving and they stay connected at the position 
when they first met.
the project uses 6 to 9 fragments that move around randomly and when a 
fragment meets another they both stop until all fragments have met another 
too.
i was thinking....each fragment having an alpha layer , is there an object 
that would recognise there are 6 or X alpha layers and when which meets 
which...?
thanx for any help
kranky

recongise-alpha-shape-and-position

hi, another boring bus trip today gave me an idea:

in order to detect collision between two shapes, you could count the number 

of white pixels in separate alpha channels (using jit.3m, f.e.), and add 

them together(+). Then count the number of white pixels in alpha channel 

after alpha blending, and compare. If the sum from separate channels is 

bigger than sum after alpha blending, shapes probably collided.

in order to detect collision between two shapes, you could count the number 
of white pixels in separate alpha channels (using jit.3m, f.e.), and add 
them together(+). Then count the number of white pixels in alpha channel 
after alpha blending, and compare. If the sum from separate channels is 
bigger than sum after alpha blending, shapes probably collided.
Tell me if you need a demo patch.

thanx for thinking about our little trouble on yr bus ride....

i'm gonna start working on that object, but since i'm running out of time, i 

could use some extra help and have a look at your demo patch.

thanx again....i hope this works, we could create random shapes with that 

function.i'll send you the patch and some of the images we'll use once 

hello,
thanx for thinking about our little trouble on yr bus ride....
i'm gonna start working on that object, but since i'm running out of time, i 
could use some extra help and have a look at your demo patch.
thanx again....i hope this works, we could create random shapes with that 
function.i'll send you the patch and some of the images we'll use once 
finished!!


thanx for your help...this is a great patch you sent me...

only problem is i have a hard time modifying it so it would work for more 

inputs....at the moment my patch is using 7 sources mixed .like 7 colored 

balls moving up and down, and zooming in and out randomly....the idea is 

when two meet they stop and so on until all have stoped.so out of the 7 i 

need to know which one meets which so i can individually stop the ones that 

when all have stoped i repeat the operation whith new fragments of forms to 

for yr example a fragment is a color ball and a shape is all 7 balls 

for my fragments i use psd images with alpha layer....

thanx nesa for yr help, hopefully soon i will understand how to add more 

howdy,
thanx for your help...this is a great patch you sent me...
only problem is i have a hard time modifying it so it would work for more 
inputs....at the moment my patch is using 7 sources mixed .like 7 colored 
balls moving up and down, and zooming in and out randomly....the idea is 
when two meet they stop and so on until all have stoped.so out of the 7 i 
need to know which one meets which so i can individually stop the ones that 
have collided...!!!?
when all have stoped i repeat the operation whith new fragments of forms to 
create another random form.
for yr example a fragment is a color ball and a shape is all 7 balls 
attached toguether.
for my fragments i use psd images with alpha layer....
thanx nesa for yr help, hopefully soon i will understand how to add more 
inputs to yr patch
ciao
lochoi


i guess the easiest way to do it would be to check two shapes at a time:

1st step: no checking, just put first shape, A. store it's alpha in 

2nd step: add second shape, B, check if collides, store in temporary 

matrix(from previous step) A.alpha+B.alpha

3rd step: add third shape, C, check if collides with A+B, store matrix A+B+C

4th step: add fourth shape, D, check if collides with A+B+C, store matrix 

and so on, until all shapes are checked. then go to step 1 for next frame.

this way you're not bound to fixed number of shapes.

imagine this algorithm: no erasing after last step. when there is no more 

space for current shape, make it smaller. this would produce some strange 

fractal-like effect(smaller and smaller circles, until all space is filled, 

1st step: no checking, just put first shape, A. store it's alpha in 
temporary matrix.
2nd step: add second shape, B, check if collides, store in temporary 
matrix(from previous step) A.alpha+B.alpha
3rd step: add third shape, C, check if collides with A+B, store matrix A+B+C
4th step: add fourth shape, D, check if collides with A+B+C, store matrix 
A+B+C+D
and so on, until all shapes are checked. then go to step 1 for next frame.

imagine this algorithm: no erasing after last step. when there is no more 
space for current shape, make it smaller. this would produce some strange 
fractal-like effect(smaller and smaller circles, until all space is filled, 
wich is - never)

I think you should check the PMPD library; much more efficient than doing it

I just posted an example patch that uses the PMPD library (Physical

modelling) in max/Jitter to simulate balls on a pool table (2d environment -

collision / friction / gravity / attraction...)

http://mathieu.chamagne.free.fr/max/mc.PMPD-jitter-example-1

I think you should check the PMPD library; much more efficient than doing it
in jitter...
I just posted an example patch that uses the PMPD library (Physical
modelling) in max/Jitter to simulate balls on a pool table (2d environment -
collision / friction / gravity / attraction...)

 http://mathieu.chamagne.free.fr/max/mc.PMPD-jitter-example-1 .pat

Mathieu Chamagne
http://mathieu.chamagne.free.fr
www.maxobjects.com

expanded a bit on the idea given by nesa, this image based collision

detection implements the most basic collision *response* possible, but it

gives the nice effect and can work with movies also.

i prototyped it on javascript but it runs nicly.

-the bounce (response) is fake, can be much better at price of knowing

-at certain Angles the object can get inside the colliding object, to solve

-cant do 1% of what pmpd can do, but as far as i know pmpd only works with

geometry, this is good for organic forms (a.k.a video)

look at nesa's example, you have to edit the lines a bit, some are warped

included, max, javascript and a simple collision map.

//name this javascript "collision.js" and put in same dir.

var circle = {_x:0,_y:0}; //circle is an object

var lcd = {width:200,high:200};//size of output

var speedx=(Math.random()+1)*3; //speed is random

function bang()//simple,nice(&fake) bounding bounce

//if this not checked can happen a "stuck" circle

speedx *= -1; //if reach bound reverse direction

}else if(circle._y>= lcd.high-10 || circle._y

function colide(){speedx*=-1;speedy*=1}//this changes the direction on

function list()//size of lcd, you can send list

> I think you should check the PMPD library; much more efficient than doing

> I just posted an example patch that uses the PMPD library (Physical

> modelling) in max/Jitter to simulate balls on a pool table (2d environment

> collision / friction / gravity / attraction...)

http://mathieu.chamagne.free.fr/max/mc.PMPD-jitter-example-1.pat

expanded a bit on the idea given by nesa, this image based collision
detection implements the most basic collision *response* possible, but it
gives the nice effect and can work with movies also.

problems:
-every object has to have separate map.
-the bounce (response)  is fake, can be much better at price of knowing
physics
-at certain Angles the object can get inside the colliding object, to solve
-cant do 1% of what pmpd can do, but as far as i know pmpd only works with
geometry, this is good for organic forms (a.k.a video)

look at nesa's example, you have to edit the lines a bit, some are warped
included, max, javascript and a simple collision map.
please help make this better.

//++++++++++++++++++++++++++++++++++
//name this javascript "collision.js" and put in same dir.
//++++++++++++++++++++++++++++++++++
//javascript begin
autowatch=1;
outlets=2
post("nn__compiled__nn");
var circle = {_x:0,_y:0}; //circle is an object
var lcd = {width:200,high:200};//size of output
var speedx=(Math.random()+1)*3; //speed is random
var speedy=(Math.random()+1)*3;

function bang()//simple,nice(&fake) bounding bounce
{
    if(circle._x>= lcd.width-10  || circle._x
        //if this not checked can happen a "stuck" circle
        if(circle._x> lcd.width-10){
            circle._x= lcd.width-10 ;}
            else if(circle._x
            circle._x=0;
        }
    speedx *= -1; //if reach bound reverse direction

    }else if(circle._y>= lcd.high-10  || circle._y
        if(circle._y> lcd.high-10 ){
            circle._y= lcd.high-10 ;}
            else if(circle._y
            circle._y=0;}
    speedy *= -1;
    }
    circle._x += speedx;
    circle._y += speedy;
    outlet(0,circle._x,circle._y);
}
function colide(){speedx*=-1;speedy*=1}//this changes the direction on
jit.3m message

function list()//size of lcd, you can send list
{
    var a = arrayfromargs(arguments);
    lcd.width=a[0]; lcd.high=a[1];
}
function newRndSpeed()//speed at random
{
    speedx=(Math.random()+1)*3;
    speedy=(Math.random()+1)*3;
}

2006/4/12, Mathieu Chamagne :
> I think you should check the PMPD library; much more efficient than doing
it
> in jitter...
> I just posted an example patch that uses the PMPD library (Physical
> modelling) in max/Jitter to simulate balls on a pool table (2d environment
-
> collision / friction / gravity / attraction...)
>
> http://mathieu.chamagne.free.fr/max/mc.PMPD-jitter-example-1.pat
>
> best
>
> Mathieu Chamagne
> http://mathieu.chamagne.free.fr
> www.maxobjects.com
>
>
>
>
>

recongise alpha, shape and position