Hi Glen, On 2015/02/12 21:49, Glen Huang wrote:
You examples make me read the spec again, and if I’m not wrong, the timing functions specified in each keyframe and in the animation node are *accumulative* (keyframes apply the specified time function to the time fractions they receive from their animation nodes), whereas in css animations, timing function in each keyframe *overrides* the one specified for the animation.
That's correct.
If that’s true, I prefer the #2 proposal too. But just to make sure I understand it correctly, it uses “ease” for standalone animation nodes, but uses “linear” for animation nodes contained in a group, by using a special “auto” keyword as the default value for AnimationTimingProperties.easing (and keeps using “linear” as the default value for keyframe time functions). Is that correct?
That's mostly correct. For nodes in a group, they would still use "ease" but the group itself would use "linear".
I discussed this with Shane and Doug today and we were of the view that requiring authors to undo the default "ease" timing on an animation in order to perform per-keyframe easing is unfortunate and would prefer to leave the defaults as "linear". What do you think?
Best regards, Brian
