I am using ARFrame’s capturedImage to record an AR session. The idea is to record at 60 fps, at which the current session is running. The issue I have seen is that reading the ARFrame’s captureImage data from concurrent queue is slowing down the ARSCNView output and also the AVAssetWriter can only be used in serial queue. Without distributing either of these work loads I am not able to record at 60 fps; current results are at 25-28 fps. Is there a way to distribute this workload to improve performance?