08-19-2022, 01:05 AM
(08-18-2022, 04:54 PM)tyFlow Wrote: Hi cgdvs,
My quaternion slerp function used to interpolate rotation values first normalizes the rotation values. This is needed in 99% of cases but the downside is that the rotation won't go past 180 degrees. As soon as you go to 181 degrees or higher, the rotation will be interpolated from the other direction, to ensure the rotation takes the smallest path possible. This prevents a lot of unnecessary flipping in the majority of cases, but it does prevent very large single rotations like you're attempting. For that kind of rotation, I'd recommend setting some animation curves on the rotation value directly and synchronizing the rotation with the keyframes rollout, instead of using interpolation.
OK, I used to use keyframes. Now, this "curve interpolation" can avoid the confusion caused by too many key frames on the max time axis without using K key frames on the time axis. If there are many operators in tyflow that need K key frames, it will be confused over time. Of course, the curve editor can solve the problem, but it is not concise enough. This is one of the reasons why I prefer to use "curve interpolation" to adjust! Imagine that if some important key parameters in tyflow can achieve the effect of key frames through this "curve interpolation", first of all, we will not be confused by the key frames on the time axis, and we can more accurately adjust the start and end of time. I think the "curve interpolation" designed by you is a very good method to achieve the effect of key frames only by parameters.
Of course, this is not critical. If you have time in the future, you may consider putting "curve interpolation" on more key parameters, so that tyflow can get rid of the trouble of key frames. thank you