The S3 cache always writes a new temporary file when a blob is created, even if the blob already exists. This can be improved to compute the digest without any file write in the case of a blob created from an existing File.
(Creating several times a blob with the same content happens for instance when placeholder images are used.)