Nanite does well when the camera is moiving at a slow to medium speed. At higher speeds things start to fall apart. Nanite works on the highest error clusters first with respect to sufrace area. This works great as long as cluster surface area is pretty consistant. However at higher speeds the surface area of a cluster is rapidly changing. This is when Nanite doesn’t do a great job. The current solution is to decrease the Fallback Relative Error per mesh. Unfortunately this does not entirely solve the problem and it will still occur and can be seen if going fast enough. So perhapse a dynamic Fallback Relative Error based upon velocity could be the solution. Which can translate to an extra account for velocity in error calculation. In conclusion Nanite, error calculation could improve with an account for velocity due to the limitatons of surface area based error calculation in Tringale Cluster Culling at high speeds.