It’s mainly due to the encoding and metadata requirements when storing blobs using Walrus. The smallest size for an encoded file, inclusive of replication metadata, is typically 62 MB. This large increase is a result of the encoding process which uses erasure codes and other metadata handling mechanisms to ensure reliability and availability of the data across storage nodes.

You're correct in thinking that metadata plays a significant role here. The Walrus storage system encodes blobs using erasure codes which involves expanding the original data for better stability and access in distributed storage systems. This expansion factor can be about 4.5 to 5 times the original size. You can find more details about this encoding in the Walrus documentation: https://docs.walrus.site/design/encoding.html.