[Solved] Unexpected tangent(interpolation) behaviour between two keyframes

lottie
#1

Hello,
I was trying to study how the tangents are decided/calculated from the TCB values. So I came across this file and saw how the in-tangent and out-tangent are calculated. It uses bezier between two waypoints.
Now I created a star layer and three waypoints and tried to predict there motion using the in-tangents and out-tangents but got some unexpected results.
The file I am using is MovingStar.sifz (861 Bytes)

All the waypoints used have TCB interpolation methods(both in and out).
I will show the calculation of in-tangent at waypoint 2:

waypoints

I am assuming that all the tangents are calculated with respect to the waypoints. That is the in-tangent is calculated with respect to waypoint-2. Please correct me if I got this wrong.

Now in the above image, “T” represents what I expected to be the tangent should have been. But when you open the attached file in synfig and run it, you will get the tangent “O” approximately(I don’t know how to find the exact co-ordinates).
Please tell me where am I going wrong or did I miss something?

#2

screenshot_019
screenshot_019.png838x440 115 KB

#3

The equation that you have show is for out tangent from point P1.
The one I wrote is for the in tangent for point P2.
Sorry for the confusion, I guess pointed towards the wrong line in the mentioned link. The correct one is this.. I will edit this in the main topic also.
Thank you!

#4

It looks like the case you referenced ( this) doesn’t happens for P2.

The condition is:

/// iter          next          after_next
/// ANY/ANY ------TCB/ANY ----- ANY/ANY
///
if(next_get_before==INTERPOLATION_TCB && after_next!=animated.waypoint_list_.end() && !is_angle())

As I understand, there are no “after next” point for P2 in your example, so after_next!=animated.waypoint_list_.end() is False.

#5

Please note that there are three points P1, P2 and P3.
So my iter is P1, next is P2, and after_next is P3, hence the code should go inside this “if” condition.
So we are calculating the out-tangent from iter and in-tangent for next.
Please correct me if I am wrong.

#6

So I got it where I was making a mistake. At waypoint-2, the in-tangent should look like T only. But I got confused because lottie format uses O as the in-tangent for waypoint-2.
So I guess this issue is resolved.
Thank you

#7

Ah, sorry. As you probably noticed, I do not have enough knowledge for this part of code yet.

Glad to hear you’ve got that sorted out! :slight_smile: