Move a layer according to its own location

[Dan Ebberts]

The way expressions work is that an expression can only access the pre-expression value of the property hosting the expression. That means an expression has no access to results it calculated on other frames, only the keyframed (or static) value.

However, when an expression accesses the value of any other property, it gets the post-expression value of that property.

Depending on what you’re trying to do, there are usually ways to work around these “limitations”. However, if what you’re trying to do involves one or more expressions that affect the results of each other (creating a feedback loop) you may be in for disappointment.

[Sam Beckett]

Thanks a lot for your reply Dan.

I don’t expect any problem because each layer at a time makes a move (of any defined value) then the next frame launches the process again.

But without considering this, let’s imagine the comp has only one layer. How would you move it according to its current location ?

Here is a random example :

During this frame :
if its “x” value is odd, then add one to its “y” value
else, then substract one to its “y” value
if its “y” value is a multiple of 3, then add 1 to its “x” value
else, then substract one to its “x” value
[x,y]

And next frame, the script restatrs its routine and the layer moves in a certain way.

[Stephen Dixon]

Since expressions aren’t able to store previous values, to achieve something like this you have to calculate the results for every frame up to the current one.

e.g.

calculatedPos = thisProperty.valueAtTime(0); //initial (blue) value
for (let f = 0; f < time; f += thisComp.frameDuration){
    if (calculatedPos[0] % 2 > 0){ //x is odd
        calculatedPos[1]++ //add one to y
    }
    if (calculatedPos[1] % 3 === 0){ // y is multiple of 3
        calculatedPos[0]++ //add 1 to x
    } else {
        calculatedPos[0]-- //subtract 1 from x
    }
}
caculatedPos

[Dan Ebberts]

Generally, you have to convert the calculation to be time-based. If you can’t come up with a formula based on the current time (or frame number) then you can use the brute-force method, where you start at time = 0 and loop frame-by-frame to calculate the current result. Your example would look something like this:

val = valueAtTime(0);
f = 1;
while (f <= timeToFrames(time)){
  val += val[0]%2 ? [0,1] : [0,-1];
  val += val[1]%3 == 0 ? [1,0] : [-1,0];
  f++;
}
val

[Sam Beckett]

Thank you Dan.

I tried to add two slider controls that get the value of position. Then I tried to real these values.

I have 3 layers but only one has expressions.

Slider Control X :

Initialized to 1 (blue value)

transform.position[0]

Slider Control Y :

Initialized to 1

transform.position[1]

Layer Position :

Initialized to [1000,500]

seedRandom(0, timeless = true) ;
X_Pos = effect(“X“)(“Control“) ;
Y_Pos = effect(“Y“)(“Control“) ;

But it behaves the same. The layer doesn’t move frame by frame.

_______________________________________________________

As you suggest, I will have to calculate all the positions of every frames of every layers till the current time.

Each frame, the expression will recalculate all these and place the current layer according to time.

?To achieve this I will need to store all these informations into a 3 dimentional array (not a 3 cells array like a 3D position). I know this exists in C/C++ but does it in javascript ?