@DanielAsarnow in your case, is it possible to split the dataset into subsets by frame count, and process them separately?