Reran this on a subset of the dataset from the still image version of the Lora
Attempt to recreate my Dark Cinematic Lora for videos.
