Drughi 34 Posted September 27, 2021 I have a little challange for you guys: How to interpolate a transformation matrix of an instance point in a non linear fashion? Here is a file for you to give it a try. interpolate_transforms.hiplc 1 Share this post Link to post Share on other sites
Drughi 34 Posted September 28, 2021 (edited) Just in case someone want's to do something similar. I've ended up just using the spline function. Instead of a matrix I've choose quaternions and the position vector wich gave me quite acurate results. Edited September 28, 2021 by Drughi Share this post Link to post Share on other sites
jkunz07 207 Posted September 28, 2021 (edited) There's no reliable way to slerp matrices as far as I know, typically people will convert the rotation to quaternion then slerp. You can then store the result as a matrix by converting from quaternion back to matrix. This example shows a lerp of the translates, scales and slerp of rotations with the result converted back to matrix. Edited September 28, 2021 by jkunz07 3 Share this post Link to post Share on other sites
mestela 752 Posted September 29, 2021 Sorta dodging your question really, but I found you can cheat this by getting 3 samples ($F, $F-0.5, $F-1) and feeding them to a sequence blend shape with cubic interpolation enabled. interpolate_transforms_via_blendshape.hiplc 4 1 Share this post Link to post Share on other sites
Drughi 34 Posted January 21 Thanks for your ideas. This is what I came up with in the end. I did only a rough interpolation and then used resample and subdivide to get adaptive and smooth samples. interpolate_transforms.hiplc Share this post Link to post Share on other sites