In my quest for the smoothest camera of all time I’ve been wondering if selecting Radians instead of Degrees in using the Quaternion BP Library’s QinterpTo function would give me a smoother interpolation?

I’m guessing not, but if anyone knows… it would be great to be sure!

You are kinda asking if apples are equal to apples.

The answer is that while no 2 apples are the same, there likely is no practical difference.

2 pie radians = 1 360 angle.

360 degrees = 2 pie radian angle.

So the real question is:

How does the BP you use convert from one to the other?

Which is way is lees prone to floating point errors?

To determine that you have to open the source and check what the code does.

In a “general” FQuat implementation:

The original rotation (yaw, roll and pitch) is multiplied by PI and divided by 180 to get to radians.

After that it usually uses Double to do internal math - so float point precision is high (and it’s very important when dealing with rotations).

The double values are then all “assembled” so to speak in the required X/Y/Z/W outputs.

This means that you can get better performance if you pass Radians into the current rotation at the beginning of the process - by next to nothing because * PI / 180 takes next to nothing to process.

Thank you… very interesting I’m doing everything in quats at the moment… getting the start/target rotation, interpolating it and setting it in quats. So I guess whether the interpolation is in degrees or radians, the performance/smoothness is about the same either way. Nice one!