Elysian Shadows

So here I am, once again, attempting to program something that somewhat resembles real-life physics. It's coming along fairly well, but I have a small issue that I need some help fixing.

First off a few implementation details (2D physics demo in SDL & OpenGL):

Creating an Object: The object class has a vector of vertices and, depending on how many you've fed into it, draws it accordingly (whether it's a point, line, triangle, quad, or polygon).

It also has a vector of applied forces which once applied are removed from the queue. The AddForce function call takes a magnitude and a direction (angle in degrees: 0/360 is up, 90 is right, 180 is down, 270 is left). The problem occurs when you provide an angle that is not only on a single axis (ie. 240 is down and left).

When updated, the object iterates through each force in the queue and takes it apart into its x and y components, then adds them to totalForcex and totalForcey respectively. What should happen after this is where I begin to struggle. Air resistance should always be opposite of the direction of motion (velocity) of an object. Right now I'm using the two if statements above (I removed second because it's same as the first but for the y-axis) to apply air resistance. This works fine when motion is on only one axis. However, when you are traveling in two directions (ie. down and left at 240 degrees) rather than traveling at an angle of 60 degrees right of the y-axis, it will travel at 45 degrees until the air resistance meets the vertical velocity, at which point it continues to travel directly left until the air resistance meets horizontal velocity. This is because both velocities are being updates at exactly the same rate even though they're different amounts.

My question to you is how would I go about finding and making the object travel at the correct slope (2 pixels over then 1 pixel down) for whatever the resultant angle of all of the acting forces is?

Sorry for the lengthy post, but without details it'd be rather difficult to guess what I'm trying to achieve.

Air resistance is not something I've handled but... don't you need to take the angle of travel into account by using sin or cos? What happens you apply air * sin(angle) to the y-vel and air * cos(angle) to the x-vel?

well if u r using a counter to measure the time elapse since the last "update " of ur objects velocities. then u just need to update the velocity along side the acceleration.
having the same air resistence on x and y axis means u r effectively reducing the velocity by same amount and it will update the velocity linearly. I guess to achieve the effect of air resistence (i.e variable velocity update at each update) you can try:-
1. to make ur air resistence dependent upon velocity.( and i guess it will be the best way)
2. or update the velocity according to the net acceleration, or update the net acceleration each frame.

i hope it helps. (P.S. i m not a pro,and its just my thoughts....)

Solved (for now) See Next Post

MarauderIIC wrote:Air resistance is not something I've handled but... don't you need to take the angle of travel into account by using sin or cos? What happens you apply air * sin(angle) to the y-vel and air * cos(angle) to the x-vel?

That's what I'm doing.

dandymcgee wrote: When updated, the object iterates through each force in the queue and takes it apart into its x and y components, then adds them to totalForcex and totalForcey respectively.

Here's the code for that:

Code: Select all

for( fAppIter = fApp.begin(); fAppIter != fApp.end(); fAppIter++ )
{
    magnitude = (*fAppIter)->Magnitude();
    angle = (*fAppIter)->Angle();
    radians = (angle * 3.14159265)/180;

    if( angle == 0 || angle == 360 ){
        totalForcey += magnitude;
    }else if( angle == 90 ){
        totalForcex += magnitude;
    }else if( angle == 180 ){
        totalForcey -= magnitude;
    }else if( angle == 270 ){
        totalForcex -= magnitude;
    }else if( angle > 0 && angle < 90 ){
        totalForcex += ( sin(radians) * magnitude );
        totalForcey += ( cos(radians) * magnitude );
    }else if( angle > 90 && angle < 180 ){
        totalForcex += ( sin(radians) * magnitude );
        totalForcey += ( cos(radians) * magnitude );
    }else if( angle > 180 && angle < 270 ){
        totalForcex += ( sin(radians) * magnitude );
        totalForcey += ( cos(radians) * magnitude );
    }else if( angle > 270 && angle < 360 ){
        totalForcex += ( sin(radians) * magnitude );
        totalForcey += ( cos(radians) * magnitude );
    }
}

The problem is them both being updated the same amount and one hits zero before the other.

hellknight wrote:well if u r using a counter to measure the time elapse since the last "update " of ur objects velocities. then u just need to update the velocity along side the acceleration.
having the same air resistence on x and y axis means u r effectively reducing the velocity by same amount and it will update the velocity linearly. I guess to achieve the effect of air resistence (i.e variable velocity update at each update) you can try:-
1. to make ur air resistence dependent upon velocity.( and i guess it will be the best way)
2. or update the velocity according to the net acceleration, or update the net acceleration each frame.

i hope it helps. (P.S. i m not a pro,and its just my thoughts....)

Could you give an example of how to update air resistance based on veclocity in my example?

EDIT: Just to let you guys know, this doesn't have to be accurate and true to real air resistance. I just need something that will be constantly opposing motion slightly so that all of my objects don't just accelerate to infinity.

Changed: To: Although at first glance I assumed these would do the same thing, the first one does not update the velocities with the air resistance each frame. I'll post back if I need further assistance, but for now I'm going to mark this as solved.

this is how i created my "friction" force in my engine about a week ago.....and it was working fine for me. (P.S. i rarely commant my own code,cause i m all alone. but i just did so everything is a bit clear to understand. :P the upper x& y r the co-ord of my object.)

as far as the resistence based upon velocity. here's how u can do it.

supoose resistence varies as (velocity)^2;
ff=kv^2;
a=k' v^2;//deviding by mass,we get another constant.
assume this is the only force acting.
then integrating both sides with prover limits with velocity changing from v1->v2 in time t.
we get:-

(1/v1-1/v2)=kt;
=>1/v2=1/v1-kt;//adjust k for ur desired resistence.

this is caluculated by assuming that only the resitence force is acting, since only resitence force varies as velocity other acceleration can be assumed to be constant (wrt velocity).

you can also perform it in 2 pass. pass 1 the static acceleration caluculation, part 2 the variable acceleration.

(P.S. like i always say i m not pro,this may not be the best or the optimal way. But i always think u dont need to optimise untill u really need to. :D. i hope it helps......if u see any typos,dont mind them i m too lazy to correct them. :P )

hellknight wrote: this is how i created my "friction" force in my engine about a week ago.....and it was working fine for me
[...]

Thanks for the example and effort explaining it, I appreciate it. I did get it working the way I want it for the time being, so all is good.

well actually i should thank you, you gave me an idea of adding force system to the physics engine......CHEERS!!! :D

hellknight wrote:well actually i should thank you, you gave me an idea of adding force system to the physics engine......CHEERS!!! :D

At first I didn't have a clue how I was going to remove them, but now I just give it a duration when I add it, and it gets deleted after that time interval (zero means forever, I use this for permanent forces such as gravity).

As far as vertices, you can still only add more.

Dandy is simulating friction through a damping force (resistance to changes is position, or velocity).

Hellknight is simulating friction as though it were a deceleration.

I guess that your approaches are nearly equivalent. You are both directly modifying the second derivative. One is multiplicative and one is additive, though.