I am working on a game and would like to add motion blur to videos that I export from the engine. Right now, I play back gameplay at 1/16th the original speed, giving me 16x more frames (e.g. instead of 60 frames per second, it's 960 frames per second) and averaging the frames. I am curious though, is this the proper weight or should I give frame samples closer to the middle more alpha? Also, should the frames sampled be sampled uniformly?
You might want to rethink motion blur and only do it for one frame, not blending between several. This will help you avoid a lot of edge cases where your frame rate is stuttering and you have to decide on which frames to blend and how much.
A common way to do this is using motion vectors, where you render your scene again, with extra data for each mesh, and render screen space velocity into a separate buffer. This buffer can then be used as a blur direction control. to compose the final blurred image.
You can also build depth of field information at the same time using this technique.