En el manual dice esto:
Rapid Motion Blur
The Rapid Motion scanline rendering algorithm is a way of rendering motion blur that is particularly well-suited to complex scenes with a lot of motion blur where it produces high-quality blurs in a fraction of the time that it would take to render raytraced motion blur. It works by subdividing object surfaces into increasingly smaller triangles. Once the surface is subdivided, the triangles are drawn on screen, according to the number of samples that you specify.
Y en cuanto a la opción que preguntas:
1.Adjust the Collect Rate value. This specifies the number of samples used per pixel to draw the subdivided triangles.
For motion-blurred objects, this means the number of time samples that are used per rendered pixel, since the object may or may not be in the pixel when the pixel is sampled at a given time.
La teoría parece clara. Ahora la práctica a saber, yo tampoco lo he probado. El motion blur lo pongo en composicion, con Twixtor.
Saludos,
Unaided