No! As you render each frame of the animation, you adjust the bucketsize to try and improve the render speed. Being an animation, each frame would be similar to the previous frame, so the optimum bucket size would be roughly the same. Over time, the picture would be very different, so when you get to the optimum bucket size you still adjust by 1 to allow for this. So you might get something like this.... Frame : BucketSize: Comment 1: 1 : First render, take a time 2: 16 : Compare this time to frame 1's render time, it's faster, so add another 16 to the bucket size. 3: 32 : Render was slower than Frame 2, so we deduct 8 from the bucket size. 4: 24 : Render was faster than Frame 3, so we try another reduction, this time by 4 5: 20 : This render was slower than frame 4, so we increase by 2 6: 22 : This render was faster than frame 5, so we increase by 1 7: 21: This render was slower than 6, so we decrease by 1 8: 20: This render was faster than 7, so we increase by 1 So by using the binary chop, we've discovered that the best bucket for this animation in the early frames is 20. However we keep fluctuating by 1 to test if the bucket size can be improved as the animation progresses. If so, the bucket size will gravitate towards the new bucket size.