• Adobe After Effects Expressions

  Ease In (Out) based on velocity between 2 keyframes…

  Posted by Mike Olson on February 9, 2009 at 6:35 pm

  I need help with an expression that would allow an object to slow to a stop AFTER a keyframe.

  Let me explain: If I change the position of a layer (or object) using 2 keyframes, I want the layer (or object) to float (keep moving in the same direction) to a stop based on the velocity it was moving between the 2 keyframes.

  Does anyone know if this can be done?

  Thanks in advance.

  Mike Olson replied 17 years, 7 months ago 3 Members · 4 Replies
• 4 Replies

• 

  Dan Ebberts

  February 9, 2009 at 10:37 pm

  Will there only ever be two keyframes? If not, what happens after the second one?

  Dan
• 

  Mike Olson

  February 9, 2009 at 10:49 pm

  Thanks for the reply Dan,

  Right now I only want to use two keyframes to bring my layer/object into position. Once the timeline hits the second keyframe, I want the layer/object to continue drifting to a stop.

  Using a third keyframe and Easing IN to the third keyframe just doesn’t “look right”. Yes, I could play with the animation (F) curves. But, I have a lot of things to move this way and having to mess with the F-curves for each object (they will all move different amounts and speeds) will be a real pain.

  Mike
• 

  Filip Vandueren

  February 10, 2009 at 6:14 am

  Hey Mike,

  Personally I like using the smooth- function a lot. It does a sort of ease ‘around’ your keyframes.
  something like smooth(width=0.2,samples=25);

  pros: – The timing is different and more natural than a normal ease.
  – You know the exact values at which the smooth will start and end (the keyframed values) and it will get there “width/2 seconds” later.

  cons: – it does it for every keyframe, not just the last one.
  – it ‘cuts corners’ on intermediate values: because of the rounding, you will not go exactly through each point.

  Back to what you were asking for:
  this expression should do the trick.

  
decayTime=0.2;

  if (numKeys && decayTime>0) {
  last_key_time=key(numKeys).time;
  if (time>last_key_time) {
  t=Math.min(time-last_key_time , decayTime);

  overshoot= (t - t*t/(decayTime*2))*velocityAtTime(last_key_time-0.001);

  value + overshoot;
} else {
value;
}
} else {
value;
}


  The objects will ‘overshoot’ their last keyframe. They will continue slowing down for ‘decayTime’ seconds untill they stop.
  If you look at the f-curves with Speed Graph and the expression f-curves enabled, you’ll see what’s happening.

  warning: by design you have no idea where the layer will settle, it will be beyond your keyframe.

  It’s based on the velocity (speed + direction) the layer was moving at a millisecond before the last keyframe. If you use the velocity exactly at the last keyframe, you get strange behaviours when the point has a very small bezier-handle.

  This will also work on 1d properties like rotation or 3d position
• 

  Mike Olson

  February 10, 2009 at 7:49 pm

  This is great Filip! Thank you so much. You are a living deity.

  However, now that I see it in action, I believe having my object move at half the speed(velocity) it was traveling between the two keyframes is closer to what I’m thinking of (very fast between the 2 frames, then overshooting the last keyframe at half or a third of the speed it was traveling, coming to a slow stop). I can get close to this idea by dividing the “overshoot” by 2:

  overshoot= (t – t*t/(decayTime*2))*velocityAtTime(last_key_time-0.001)/2;

  Unless you have a better way of achieving this?

  Thanks again!

Reply to this Discussion! Login or Sign Up