I see; the fundamental issue is additive animation systems don't respond well to having animations removed and re-started.
Conversation
That sounds right. The notion of "re targeting" makes more sense to me than "removing and adding"
1
3
if you want to retarget, you don't need to remove the old animation. Add a new one from the old animation's destination.
3
2
React State Stream solves this API-wise by letting new animation be map() of the old one
2
that means every animation has to support the entire possibility space of old animations, though.
1
e.g., usually animation curves are 1-d. What happens if you add a new animation going a different direction?
1
Whoever adds new animations, decides how to merge them: sudden transition, additive, something else. That's the point
1
Check first and second example in live demo: rawgit.com/chenglou/react First disruptive, second additive, Same API
1
Correction: second is not additive, it's simpler. Third one on this page is additive: rawgit.com/gaearon/react-
2
Implementation is just using some frames from old stream and combining them with new stream
1
1
Replying to
This model seems reasonable but examples still suffer from weird non-physical timing e.g. interrupted animations (cont…)